Low-Rank Adaptation (LoRA) is a widely used parameter-efficient fine-tuning (PEFT) method for large language models.Under a fixed rank budget, LoRA parameterizes each adapted weight through a single low-dimensional inputside pathway, which may couple heterogeneous behaviors through shared input directions and induce interference during optimization.We propose Static Orthogonal Subspace LoRA (SOS-LoRA), a drop-in extension that reparameterizes a rank-r tot update as a sum of K static (always-on, non-routed) low-rank experts.SOS-LoRA (i) decomposes the total rank across experts, (ii) applies a fixed multi-scale scaling scheme to encourage scale-separated optimization dynamics, and (iii) promotes diverse input-side directions via cross-expert orthogonal initialization and a lightweight regularizer.SOS-LoRA remains fully mergeable, adding no inference-time parameters or latency after merging.Experiments on reasoning and knowledge-intensive benchmarks (Llama 2/3), encoder-based NLU (GLUE), and math reasoning (GSM8K/MATH) show consistent gains over matched-budget LoRA baselines and recent variants.Code is available at https://github.com/llm172/sos-lora.

WebAgents have demonstrated strong capabilities in autonomously completing complex web tasks, yet their computational efficiency vulnerabilities have received limited attention.Adversaries can inject malicious prompts into webpages, causing WebAgents to generate unnecessarily long reasoning processes and incur excessive computational cost, termed Computational Cost Attacks (CCA).In this paper, to systematically study this vulnerability under realistic black-box settings, we propose CostBomb, a generation-then-selection attack framework that leverages large language models to generate diverse adversarial prompts and a reinforcement learning-enhanced selector to identify the most effective perturbations.Extensive experiments on multiple real-world web benchmarks reveal that existing WebAgents are highly vulnerable to CCA, suffering substantial increases in computational cost without compromising successful task completion.Our findings highlight an overlooked dimension of WebAgent robustness and underscore the urgent need for efficiency-aware defenses.

Multimodal recommendation, which integrates user-item interactions with rich item multimodal features to infer user preferences, has emerged as one of the state-of-the-art recommendation paradigms. However, most existing multimodal recommendation methods focus solely on the multimodal fusion process of items, overlooking the impact of modality-specific noise characteristics and inter-modal distribution shifts on the fusion process. As a result, noise from specific modalities may be inadvertently amplified during the fusion of multimodal representations, thereby degrading the quality of fused representations and ultimately undermining recommendation accuracy. To address this issue, we propose a novel framework, DAMPS, which Denoises and cAlibrates multimodal representations in the spectral domain from the perspectives of both aMplitude and Phase Spectrum. Specifically, we first apply the Fast Fourier Transform to decompose multimodal representations into amplitude spectrum and phase spectrum components. We then propose the adaptive phase calibrator, which resolves the phase distribution shift between modalities through the optimal minimum variance unbiased estimate of phase offset and adaptive phase rotation. We further develop the amplitude variance-ratio filter, which suppresses modality-specific amplitude noise while preserving prominent modality-specific representations. Finally, we introduce the inter-modal coherence filter, which quantifies and enhances the intrinsic shared information between modalities through amplitude-squared coherence, thereby eliminating irrelevant modality noise. Technically, DAMPS is model-agnostic and highly flexible, enabling seamless integration with various multimodal recommendation backbones. Extensive evaluations on five public multimodal recommendation datasets demonstrate that DAMPS consistently improves recommendation performance and achieves state-of-the-art results.

Recentworks have shown the potential of diffusion models in computer vision and natural language processing. Apart from the classical supervised learning fields, diffusion models have also shown strong competitiveness in reinforcement learning (RL) by formulating decision-making as sequential generation. However, incorporating temporal information of sequential data and utilizing it to guide diffusion models to perform better generation is still an open challenge. In this paper, we take one step forward to investigate controllable generation with temporal conditions that are refined from temporal information. We observe the importance of temporal conditions in sequential generation in sufficient scenarios and provide a comprehensive discussion and comparison of different temporal conditions. Based on the observations, we propose an effective temporally-conditional diffusion model coined Temporally-Composable Diffuser (TCD), which extracts temporal information from interaction sequences and explicitly guides generation with temporal conditions. Specifically, we separate the sequences into three parts according to time expansion and identify historical, immediate, and prospective conditions accordingly. Each condition preserves non-overlapping temporal information of sequences, enabling more controllable generation when we jointly use them to guide the diffuser. Finally, we conduct extensive experiments and analysis to reveal the favorable applicability of TCD in offline RL tasks, where our method reaches or matches the best performance compared with prior SOTA baselines.

Tabular data have been playing a mostly important role in diverse real-world fields, such as healthcare, engineering, finance, etc. The recent success of deep learning has fostered many deep networks (e.g., Transformer, ResNet) based tabular learning methods. Generally, existing deep tabular machine learning methods are along with the two paradigms, i.e., in-learning and pre-learning. In-learning methods need to train networks from scratch or impose extra constraints to regulate the representations which nonetheless train multiple tasks simultaneously and make learning more difficult, while pre-learning methods design several pretext tasks for pre-training and then conduct task-specific fine-tuning, which however need much extra training effort with prior knowledge. In this paper, we introduce a novel deep Tabular Representation Corrector, TRC, to enhance any trained deep tabular model's representations without altering its parameters in a model-agnostic manner. Specifically, targeting the representation shift and representation redundancy that hinder prediction, we propose two tasks, i.e., (i) Tabular Representation Re-estimation, that involves training a shift estimator to calculate the inherent shift of tabular representations to subsequently mitigate it, thereby re-estimating the representations and (ii) Tabular Space Mapping, that transforms the above re-estimated representations into a light-embedding vector space via a coordinate estimator while preserves crucial predictive information to minimize redundancy. The two tasks jointly enhance the representations of deep tabular models without touching on the original models thus enjoying high efficiency. Finally, we conduct extensive experiments on state-of-the-art deep tabular machine learning models coupled with TRC on various tabular benchmarks which have shown consistent superiority.

In real-world decision-making tasks, it is critical for reinforcement learning (RL) methods to be both stable and robust. Maximum entropy RL methods typically generate a robust policy with entropy augmented reward. While incorporating entropy into the reward offers the benefit of exploration, it presents limited universal applicability and persistent convergence difficulties, such as suboptimal policy stabilization and unstable $Q$ value update. From optimization, we define these two issues as tremulous policy and spiky Q-function, investigating their underlying causes and relationships. Analysis with this, the maximum entropy principle leads to a spiky $Q$ -function update, which ultimately results in a tremulous policy. We thus introduce a beta-symmetric Kullback-Leibler (KL) divergence objective to mitigate such issues under the maximum entropy framework. With this objective function, the tremulous nature of the policy could be controlled with a large beta value. The spiky $Q$ -function could be avoided by annealing the entropy in the target $Q$ value, as the beta-symmetric KL divergence is an upper bound of the original reverse KL divergence. Theoretically, we prove that minimizing our new objective function results in a new policy that presents an improvement in the $Q$ value. Guaranteed by these results, we ultimately derive the optimal policy by iteratively updating the $Q$ value and policy, and we call this method max-entropy stable optimization (MeSO). Experimental results on the Mujoco and Roboschool platforms demonstrate that our algorithm maintains stability while offering better flexibility and overall performance.

Offline reinforcement learning (RL) provides a promising solution to learning an agent fully relying on a data-driven paradigm. However, constrained by the limited quality of the offline dataset, its performance is often suboptimal. Therefore, it is desired to further finetune the agent via extra online interactions before deployment. Unfortunately, offline-to-online RL can be challenging due to two main challenges: constrained exploratory behavior and state-action distribution shift. In view of this, we propose a simple unified uncertainty-guided (SUNG) framework, which naturally unifies the solution to both challenges with the tool of uncertainty. Specifically, SUNG quantifies uncertainty via a variational autoencoder (VAE)-based state-action visitation density estimator. To facilitate efficient exploration, SUNG presents a practical optimistic exploration strategy to select informative actions with both high value and high uncertainty. Moreover, SUNG develops an adaptive exploitation method by applying conservative offline RL objectives to high-uncertainty samples and standard online RL objectives to low-uncertainty samples to smoothly bridge offline and online stages. SUNG achieves state-of-the-art online finetuning performance when combined with different offline RL methods, across various environments and datasets in the D4RL benchmark. Codes are made publicly available in https://github.com/guosyjlu/ACBR.

Explainable fake news detection aims to assess the veracity of news claims while providing human-friendly explanations. Existing methods incorporating investigative journalism are often inefficient and struggle with breaking news. Recent advances in large language models (LLMs) enable leveraging externally retrieved reports as evidence for detection and explanation generation, but unverified reports may introduce inaccuracies. Moreover, effective explainable fake news detection should provide a comprehensible explanation for all aspects of a claim to assist the public in verifying its accuracy. To address these challenges, we propose a graph-enhanced defense framework (G-Defense) that provides fine-grained explanations based solely on unverified reports. Specifically, we construct a claim-centered graph by decomposing the news claim into several sub-claims and modeling their dependency relationships. For each sub-claim, we use the retrieval-augmented generation (RAG) technique to retrieve salient evidence and generate competing explanations. We then introduce a defense-like inference module based on the graph to assess the overall veracity. Finally, we prompt an LLM to generate an intuitive explanation graph. Experimental results demonstrate that G-Defense achieves state-of-the-art performance in both veracity detection and the quality of its explanations.

The remarkable achievements of Large Language Models (LLMs) have captivated the attention of both academia and industry, transcending their initial role in dialogue generation. To expand the usage scenarios of LLM, some works enhance the effectiveness and capabilities of the model by introducing more external information, which is called the agent paradigm. Based on this idea, we propose a new method that integrates the agent paradigm into out-of-distribution (OOD) detection task, aiming to improve its robustness and adaptability. Our proposed method, Concept Matching with Agent (CMA), employs neutral prompts as agents to augment the CLIP-based OOD detection process. These agents function as dynamic observers and communication hubs, interacting with both In-distribution (ID) labels and data inputs to form vector triangle relationships. This triangular framework offers a more nuanced approach than the traditional binary relationship, allowing for better separation and identification of ID and OOD inputs. Our extensive experimental results showcase the superior performance of CMA over both zero-shot and training-required methods in a diverse array of real-world scenarios.

Sequential Recommender Systems (SRS) has stood out as a highly promising technique in numerous domains due to its impressive capability of capturing complex user preferences. Current SRS have employed transformer-based models to give the next-item prediction. Nevertheless, its quadratic computational complexity has often resulted in notable inefficiencies, posing a significant obstacle to real-time recommendation processes. Recently, Mamba has demonstrated its exceptional effectiveness in time series prediction, delivering substantial improvements in both efficiency and effectiveness. However, directly applying Mamba to SRS poses certain challenges. Its unidirectional structure may impede the ability to capture contextual information in user-item interactions, while its instability in state estimation may hinder the ability to capture short-term patterns in interaction sequences. To address these issues, we propose a novel framework called Selective Gated Mamba for Sequential Recommendation (SIGMA). By introducing the Partially Flipped Mamba (PF-Mamba), we construct a special bi-directional structure to address the context modeling challenge. Then, to consolidate PF-Mamba's performance, we employed an input-dependent Dense Selective Gate (DS Gate) to allocate the weights of the two directions and further filter the sequential information. Moreover, for short sequence modeling, we devise a Feature Extract GRU (FE-GRU) to capture the short-term dependencies. Experimental results demonstrate that SIGMA significantly outperforms existing baselines across five real-world datasets. Our implementation code is available in Supplementary Material to ease reproducibility.

Black-box prompt tuning has become a prevalent parameter-efficient paradigm that leverages the capabilities of large language models (LLMs) for customized applications in specific downstream tasks. In practical scenarios, downstream tasks frequently involve data distributions that are heavily imbalanced. Such imbalances tend to impair performance, causing severe performance collapse in minority classes. Conducting effective imbalanced black-box prompt tuning to mitigate the adverse effects of imbalanced data distribution on prompt performance remains a significant challenge. In this paper, we propose black-box prompt tuning with first and zeroth order gradient (BPT-FZG) for handling the imbalanced data. Specifically, BPT-FZG introduces AUC maximization as the objective for prompt tuning and equivalently formulates it as a nonconvex-concave saddle point problem to avoid the construction of sample pairs from opposite classes. Indeed, BPT-FZG optimizes the latent representation of the continuous prompt in the low-dimensional subspace with AUC loss and leverages the first and zeroth order gradients alternately to update the parameters. Furthermore, we establish the theoretical convergence guarantee for BPT-FZG under common assumptions, showing that our method can find a stationary point of the objective function. Our experiments on RoBERTa-large, GPT2-XL, and Llama3 show that BPT-FZG achieves improvement on various imbalanced datasets, emphasizing the effectiveness of our methods.

Zichuan Fu, Xian Wu, Yejing Wang, Wanyu Wang, Shanshan Ye, Hongzhi Yin, Yi Chang, Yefeng Zheng, Xiangyu Zhao. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Semi-supervised learning (SSL) has garnered significant attention due to its ability to leverage limited labeled data and a large amount of unlabeled data to improve model generalization performance. Recent approaches achieve impressive successes by combining ideas from both consistency regularization and pseudo-labeling. However, these methods tend to underperform in the more realistic situations with relatively scarce labeled data. We argue that this issue arises because existing methods rely solely on the model’s confidence, making them challenging to accurately assess the model’s state and identify unlabeled examples contributing to the training phase when supervision information is limited, especially during the early stages of model training. In this paper, we propose a novel SSL model called CGMatch, which, for the first time, incorporates a new metric known as Count-Gap (CG). We demonstrate that CG is effective in discovering unlabeled examples beneficial for model training. Along with confidence, a commonly used metric in SSL, we propose a fine-grained dynamic selection (FDS) strategy. This strategy dynamically divides the unlabeled dataset into three subsets with different characteristics: easy-to-learn set, ambiguous set, and hard-to-learn set. By selective filtering subsets, and applying corresponding regularization with selected subsets, we mitigate the negative impact of incorrect pseudo-labels on model optimization and generalization. Extensive experimental results on several common SSL benchmarks indicate the effectiveness of CGMatch especially when the labeled data are particularly limited. Source code is available at https://github.com/BoCheng-96/CGMatch.

Visual Prompt Learning (VPL) has emerged as a powerful strategy for harnessing the capabilities of large-scale pretrained models (PTMs) to tackle specific downstream tasks. However, the opaque nature of PTMs in many real-world applications has led to a growing interest in gradient-free approaches within VPL. A significant challenge with existing black-box VPL methods lies in the high dimensionality of visual prompts, which necessitates considerable API queries for tuning, thereby impacting efficiency. To address this issue, we propose a novel query-efficient framework for blackbox visual prompting, designed to generate input-dependent visual prompts efficiently for large-scale black-box PTMs. Our framework is built upon the insight of reparameterizing prompts using neural networks, improving the typical pretraining-fine-tuning paradigm through the subspace learning strategy to maximize efficiency and adaptability from both the perspective of initial weights and parameter dimensionality. This tuning intrinsically optimizes low-dimensional representations within the well-learned subspace, enabling the efficient adaptation of the network to downstream tasks. Our approach significantly reduces the necessity for substantial API queries to PTMs, presenting an efficient method for leveraging large-scale black-box PTMs in visual prompting tasks. Most experimental results across various benchmarks demonstrate the effectiveness of our method, showcasing substantial reductions in the number of required API queries to PTMs while maintaining or even enhancing performance on downstream tasks.

Fine-tuning vision-language models (VLMs) with large amounts of unlabeled data has recently garnered significant interest. However, a key challenge remains the lack of high-quality pseudo-labeled data. Current pseudo-labeling strategies often struggle with mismatches between semantic and visual information, leading to sub-optimal performance of unsupervised prompt learning (UPL) methods. In this paper, we introduce a simple yet effective approach called Augmenting Discriminative Richness via Diffusions (AiR), toward learning a richer discriminating way to represent the class comprehensively and thus facilitate classification. Specifically, our approach includes a pseudo-label generation module that leverages high-fidelity synthetic samples to create an auxiliary classifier, which captures richer visual variation, bridging text-image-pair classification to a more robust image-image-pair classification. Additionally, we exploit the diversity of diffusion-based synthetic samples to enhance prompt learning, providing greater information for semantic-visual alignment. Extensive experiments on five public benchmarks, including RESISC45 and Flowers102, and across three learning paradigms-UL, SSL, and TRZSL-demonstrate that AiR achieves substantial and consistent performance improvements over state-of-the-art unsupervised prompt learning methods. Code is available.

Federated Learning (FL) has emerged as a promising framework for distributed machine learning, enabling collaborative model training without sharing local data, thereby preserving privacy and enhancing security. However, data heterogeneity resulting from differences across user behav-iors, preferences, and device characteristics poses a significant challenge for federated learning. Most previous works overlook the adjustment of aggregation weights, relying solely on dataset size for weight assignment, which often leads to unstable convergence and reduced model performance. Recently, several studies have sought to refine aggregation strategies by incorporating dataset characteristics and model alignment. However, adaptively adjusting aggregation weights while ensuring data security—without requiring additional proxy data—remains a significant challenge. In this work, we propose Federated learning with Adaptive Weight Aggregation (FedAWA), a novel method that adaptively adjusts aggregation weights based on client vectors during the learning process. The client vector captures the direction of model updates, reflecting local data variations, and is used to optimize the aggregation weight without requiring additional datasets or violating privacy. By assigning higher aggregation weights to local models whose updates align closely with the global optimization direction, FedAWA enhances the stability and generalization of the global model. Extensive experiments under diverse scenarios demonstrate the superiority of our method, providing a promising solution to the challenges of data heterogeneity in federated learning.

Large language models (LLMs) require continual knowledge updates to keep pace with the evolving world.While various model editing methods have been proposed, most face critical challenges in the context of lifelong learning due to two fundamental limitations: (1) Edit Overshooting -parameter updates intended for a specific fact spill over to unrelated regions, causing interference with previously retained knowledge; and (2) Knowledge Entanglement -polysemantic neurons' overlapping encoding of multiple concepts makes it difficult to isolate and edit a single fact.In this paper, we propose MicroEdit, a neuron-level editing method that performs minimal and controlled interventions within LLMs.By leveraging a sparse autoencoder (SAE), MicroEdit disentangles knowledge representations and activates only a minimal set of necessary neurons for precise parameter updates.This targeted design enables fine-grained control over the editing scope, effectively mitigating interference and preserving unrelated knowledge.Extensive experiments show that MicroEdit outperforms prior methods and robustly handles lifelong knowledge editing across QA and Hallucination settings on LLaMA 1 and Mistral 2 .

Multimodal Retrieval-Augmented Generation (MRAG) systems enhance LMMs by integrating external multimodal databases, but introduce unexplored privacy vulnerabilities.While text-based RAG privacy risks have been studied, multimodal data presents unique challenges.We provide the first systematic analysis of MRAG privacy vulnerabilities across vision-language and speech-language modalities.Using a novel compositional structured prompt attack in a black-box setting, we demonstrate how attackers can extract private information by manipulating queries.Our experiments reveal that LMMs can both directly generate outputs resembling retrieved content and produce descriptions that indirectly expose sensitive information, highlighting the urgent need for robust privacy-preserving MRAG techniques.The code is available at

Modeling high-resolution spatiotemporal representations, including both global dynamic contexts (e.g., holistic human motion tendencies) and local motion details (e.g., high-frequency changes of keypoints), is essential for video-based human pose estimation (VHPE). Current state-of-the-art methods typically unify spatiotemporal learning within a single type of modeling structure (convolution or attention-based blocks), which inherently have difficulties in balancing global and local dynamic modeling and may bias the network to one of them, leading to suboptimal performance. Moreover, existing VHPE models suffer from quadratic complexity when capturing global dependencies, limiting their applicability especially for high-resolution sequences. Recently, the state space models (known as Mamba) have demonstrated significant potential in modeling long-range contexts with linear complexity; however, they are restricted to 1D sequential data. In this paper, we present a novel framework that extends Mamba from two aspects to separately learn global and local high-resolution spatiotemporal representations for VHPE. Specifically, we first propose a Global Spatiotemporal Mamba, which performs 6D selective space-time scan and spatial- and temporal-modulated scan merging to efficiently extract global representations from high-resolution sequences. We further introduce a windowed space-time scan-based Local Refinement Mamba to enhance the high-frequency details of localized keypoint motions. Extensive experiments on four benchmark datasets demonstrate that the proposed model outperforms state-of-the-art VHPE approaches while achieving better computational trade-offs.

The performance of offline reinforcement learning (RL) is sensitive to the proportion of high-return trajectories in the offline dataset. However, in many simulation environments and real-world scenarios, there are large ratios of low-return trajectories rather than high-return trajectories, which makes learning an efficient policy challenging. In this paper, we propose a method called Contrastive Diffuser (ContraDiff) to make full use of low-return trajectories and improve the performance of offline RL algorithms. Specifically, ContraDiff groups the states of trajectories in the offline dataset into high-return states and low-return states and treats them as positive and negative samples correspondingly. Then, it designs a contrastive mechanism to pull the planned trajectory of an agent toward high-return states and push them away from low-return states. Through the contrast mechanism, trajectories with low returns can serve as negative examples for policy learning, guiding the agent to avoid areas associated with low returns and achieve better performance. Through the contrast mechanism, trajectories with low returns provide a ``counteracting force'' guides the agent to avoid areas associated with low returns and achieve better performance. Experiments on 27 sub-optimal datasets demonstrate the effectiveness of our proposed method. Our code is publicly available at https://github.com/Looomo/contradiff.

In Federated Learning (FL), weighted aggregation of local models is conducted to generate a new global model, and the aggregation weights are typically normalized to 1. A recent study identifies the global weight shrinking effect in FL, indicating an enhancement in the global model’s generalization when the sum of weights (i.e., the shrinking factor) is smaller than 1, where how to learn the shrinking factor becomes crucial. However, principled approaches to this solution have not been carefully studied from the adequate consideration of privacy concerns and layer-wise distinctions. To this end, we propose a novel model aggregation strategy, Federated Learning with Adaptive Layer-wise Weight Shrinking (FedLWS), which adaptively designs the shrinking factor in a layer-wise manner and avoids optimizing the shrinking factors on a proxy dataset. We initially explored the factors affecting the shrinking factor during the training process. Then we calculate the layer-wise shrinking factors by considering the distinctions among each layer of the global model. FedLWS can be easily incorporated with various existing methods due to its flexibility. Extensive experiments under diverse scenarios demonstrate the superiority of our method over several state-of-the-art approaches, providing a promising tool for enhancing the global model in FL.

Anomaly detection, indicating to identify the anomalies that significantly deviate from the majority normal instances of data, has been an important role in machine learning and related applications. Despite the significant success achieved in anomaly detection on image and text data, the accurate Tabular Anomaly Detection (TAD) has still been hindered due to the lack of clear prior semantic information in the tabular data. Most state-of-the-art TAD studies are along the line of reconstruction, which first reconstruct training data and then use reconstruction errors to decide anomalies; however, reconstruction on training data can still hardly distinguish anomalies due to the data entanglement in their representations. To address this problem, in this paper, we propose a novel approach Decomposed Representation Learning (DRL), to re-map data into a tailor-designed constrained space, in order to capture the underlying shared patterns of normal samples and differ anomalous patterns for TAD. Specifically, we enforce the representation of each normal sample in the latent space to be decomposed into a weighted linear combination of randomly generated orthogonal basis vectors, where these basis vectors are both data-free and training-free. Furthermore, we enhance the discriminative capability between normal and anomalous patterns in the latent space by introducing a novel constraint that amplifies the discrepancy between these two categories, supported by theoretical analysis. Finally, extensive experiments on 40 tabular datasets and 16 competing tabular anomaly detection algorithms show that our method achieves state-of-the-art performance.

Large Scale Pre-Trained Language Models (PTMs) have demonstrated unprecedented capabilities across diverse natural language processing tasks. Adapting such models to downstream tasks is computationally intensive and time-consuming, particularly in black-box scenarios common in Language-Model-as-a-Service (LMaaS) environments, where model parameters and gradients are inaccessible. Recently, black-box prompt learning using zeroth-order gradients has emerged as a promising approach to address these challenges by optimizing learnable continuous prompts in embedding spaces, starting with \textit{randomly initialized discrete text prompts}. However, its reliance on randomly initialized discrete prompts limits adaptability to diverse downstream tasks or models. To address this limitation, this paper introduces ZO-PoG, a novel framework that optimizes prompts through a collaborative approach, combining Policy Gradient optimization for initial discrete text prompts and Zeroth-Order optimization for continuous prompts in embedding space. By optimizing collaboratively between discrete and continuous prompts, ZO-PoG maximizes adaptability to downstream tasks, achieving superior results without direct access to the model’s internal structures. Importantly, we establish the sub-linear convergence of ZO-PoG under mild assumptions. The experiments on different datasets demonstrate significant improvements in various tasks compared to the baselines.

Black-Box Discrete Prompt Learning (BDPL) is a prompt-tuning method that optimizes discrete prompts without accessing model parameters or gradients, making the prompt tuning on a cloud-based Large Language Model (LLM) feasible. Adapting Federated Learning (FL) to BDPL could further enhance prompt tuning performance by leveraging data from diverse sources. However, all previous research on federated black-box prompt tuning had neglected the substantial query cost associated with the cloud-based LLM service. To address this gap, we conducted a theoretical analysis of query efficiency within the context of federated black-box prompt tuning. Our findings revealed that degrading FedAvg to activate only one client per round, a strategy we called \textit{FedOne}, enabled optimal query efficiency in federated black-box prompt learning. Building on this insight, we proposed the FedOne framework, a federated black-box discrete prompt learning method designed to maximize query efficiency when interacting with cloud-based LLMs. We conducted numerical experiments on various aspects of our framework, demonstrating a significant improvement in query efficiency, which aligns with our theoretical results.

In this work, we explore the potential of large language models (LLMs) for generating functional test scripts, which necessitates understanding the dynamically evolving code structure of the target software. To achieve this, we propose a case-based reasoning (CBR) system utilizing a 4R cycle (i.e., retrieve, reuse, revise, and retain), which maintains and leverages a case bank of test intent descriptions and corresponding test scripts to facilitate LLMs for test script generation. To improve user experience further, we introduce Re4, an optimization method for the CBR system, comprising reranking-based retrieval finetuning and reinforced reuse finetuning. Specifically, we first identify positive examples with high semantic and script similarity, providing reliable pseudo-labels for finetuning the retriever model without costly labeling. Then, we apply supervised finetuning, followed by a reinforcement learning finetuning stage, to align LLMs with our production scenarios, ensuring the faithful reuse of retrieved cases. Extensive experimental results on two product development units from Huawei Datacom demonstrate the superiority of the proposed CBR+Re4. Notably, we also show that the proposed Re4 method can help alleviate the repetitive generation issues with LLMs.

Black-box attacks have emerged as a significant threat to deep neural networks. This challenge is particularly difficult in discrete sequential data compared to continuous data. Recently, the Blockwise Bayesian Attack (BBA) leveraging discrete Bayesian optimization with an adapted RBF kernel has gained prominence as a cutting-edge solution. However, it relies solely on alignment information (i.e., positional differences) within the RBF kernel, which may not fully capture the information (such as statistical, structural, and semantic information) inherent in discrete sequential data and potentially lacks the desired inductive bias necessary to approximate the target function accurately. To overcome this limitation, this paper proposes a novel bilevel Bayesian optimization approach to adaptively learn a hybrid space that better captures the similarity between discrete sequences. Specifically, we introduce a multi-kernel mechanism that incorporates multiple types of information, creating a more comprehensive similarity measure. Moreover, we develop a bilevel Bayesian optimization algorithm, where the outer-level objective determines the optimal weights of the multiple kernels, while the inner-level objective identifies the optimal adversarial sequence. Extensive experiments conducted on discrete sequential data demonstrate that our approach ensures secure multi-kernel selection and achieves a higher attack success rate with only a few additional queries, compared to BBA and other traditional optimization strategies.

Long-horizon planning in robotic manipulation tasks requires translating underspecified, symbolic goals into executable control programs satisfying spatial, temporal, and physical constraints. However, language model-based planners often struggle with long-horizon task decomposition, robust constraint satisfaction, and adaptive failure recovery. We introduce InstructFlow, a multi-agent framework that establishes a symbolic, feedback-driven flow of information for code generation in robotic manipulation tasks. InstructFlow employs a InstructFlow Planner to construct and traverse a hierarchical instruction graph that decomposes goals into semantically meaningful subtasks, while a Code Generator generates executable code snippets conditioned on this graph. Crucially, when execution failures occur, a Constraint Generator analyzes feedback and induces symbolic constraints, which are propagated back into the instruction graph to guide targeted code refinement without regenerating from scratch. This dynamic, graph-guided flow enables structured, interpretable, and failure-resilient planning, significantly improving task success rates and robustness across diverse manipulation benchmarks, especially in constraint-sensitive and long-horizon scenarios.

Recent advancements in Large Language Models (LLMs) have emphasized the critical role of fine-tuning (FT) techniques in adapting LLMs to specific tasks, especially when retraining from scratch is computationally infeasible. Fine-tuning enables LLMs to leverage task- or domain-specific data, producing models that more effectively meet the requirements of targeted applications. However, conventional FT approaches often suffer from catastrophic forgetting and suboptimal data efficiency, limiting their real-world applicability. To address these challenges, this paper proposes DEAL, a novel framework that integrates Low-Rank Adaptation (LoRA) with a continuous fine-tuning strategy. By incorporating knowledge retention and adaptive parameter update modules, the framework mitigates the limitations of existing FT methods while maintaining efficiency. Experiments on 15 diverse datasets show that DEAL consistently outperforms baseline methods, yielding substantial gains in task accuracy and resource efficiency. These findings demonstrate the potential of our approach to advance continual adaptation in LLMs by enhancing task performance while improving resource efficiency. The source code is publicly available at https://github.com/Applied-Machine-Learning-Lab/DEAL.

Continual offline reinforcement learning (CORL) has shown impressive ability in diffusion-based continual learning systems by modeling the joint distributions of trajectories. However, most research only focuses on limited continual task settings where the tasks have the same observation and action space, which deviates from the realistic demands of training agents in various environments. In view of this, we propose Vector-Quantized Continual Diffuser, named VQ-CD, to break the barrier of different spaces between various tasks. Specifically, our method contains two complementary sections, where the quantization spaces alignment provides a unified basis for the selective weights activation. In the quantized spaces alignment, we leverage vector quantization to align the different state and action spaces of various tasks, facilitating continual training in the same space. Then, we propose to leverage a unified diffusion model attached by the inverse dynamic model to master all tasks by selectively activating different weights according to the task-related sparse masks. Finally, we conduct extensive experiments on 15 continual learning (CL) tasks, including conventional CL task settings (identical state and action spaces) and general CL task settings (various state and action spaces). Compared with 17 baselines, our method reaches the SOTA performance.

Conditional decision generation with diffusion models has shown powerful competitiveness in reinforcement learning (RL). Recent studies reveal the relation between energy-function-guidance diffusion models and constrained RL problems. The main challenge lies in estimating the intermediate energy, which is intractable due to the log-expectation formulation during the generation process. To address this issue, we propose the Analytic Energy-guided Policy Optimization (AEPO). Specifically, we first provide a theoretical analysis and the closed-form solution of the intermediate guidance when the diffusion model obeys the conditional Gaussian transformation. Then, we analyze the posterior Gaussian distribution in the log-expectation formulation and obtain the target estimation of the log-expectation under mild assumptions. Finally, we train an intermediate energy neural network to approach the target estimation of log-expectation formulation. We apply our method in 30+ offline RL tasks to demonstrate the effectiveness of our method. Extensive experiments illustrate that our method surpasses numerous representative baselines in D4RL offline reinforcement learning benchmarks.

Vertical Federated Learning (VFL) enables participants to collaboratively train models on aligned samples while keeping their heterogeneous features private and distributed. Despite their utility, VFL models remain vulnerable to adversarial attacks during inference. Adversarial Training (AT), which generates adversarial examples at each training iteration, stands as the most effective defense for improving model robustness. However, applying AT in VFL settings (VFAL) faces significant computational efficiency challenges, as the distributed training framework necessitates iterative propagations across participants. To this end, we propose **_DecVFAL_** framework, which substantially accelerates **_VFAL_** training through a dual-level ***Dec***oupling mechanism applied during adversarial sample generation. Specifically, we first decouple the bottom modules of clients (directly responsible for adversarial updates) from the remaining networks, enabling efficient _lazy sequential propagations_ that reduce communication frequency through delayed gradients. We further introduce _decoupled parallel backpropagation_ to accelerate delayed gradient computation by eliminating idle waiting through parallel processing across modules. Additionally, we are the first to establish convergence analysis for VFAL, rigorously characterizing how our decoupling mechanism interacts with existing VFL dynamics, and prove that _DecVFAL_ achieves an $\mathcal{O}(1/\sqrt{K})$ convergence rate matching that of standard VFLs. Experimental results show that _DecVFAL_ ensures competitive robustness while significantly achieving about $3\sim10\times$ speed up.

Tabular data have been playing a vital role in diverse real-world fields, including healthcare, finance, etc. With the recent success of Large Language Models (LLMs), early explorations of extending LLMs to the domain of tabular data have been developed. Most of these LLM-based methods typically first serialize tabular data into natural language descriptions, and then tune LLMs or directly infer on these serialized data. However, these methods suffer from two key inherent issues: (i) data perspective: existing data serialization methods lack universal applicability for structured tabular data, and may pose privacy risks through direct textual exposure, and (ii) model perspective: LLM fine-tuning methods struggle with tabular data, and in-context learning scalability is bottle-necked by input length constraints (suitable for few-shot learning). This work explores a novel direction of integrating LLMs into tabular data through logical decision tree rules as intermediaries, proposing a decision tree enhancer with LLM-derived rule for tabular prediction, DeLTa. The proposed DeLTa avoids tabular data serialization, and can be applied to full data learning setting without LLM fine-tuning. Specifically, we leverage the reasoning ability of LLMs to redesign an improved rule given a set of decision tree rules. Furthermore, we provide a calibration method for original decision trees via new generated rule by LLM, which approximates the error correction vector to steer the original decision tree predictions in the direction of ``errors'' reducing. Finally, extensive experiments on diverse tabular benchmarks show that our method achieves state-of-the-art performance.

Recent deep sequential recommendation models often struggle to effectively model key characteristics of user behaviors, particularly in handling sequence length variations and capturing diverse interaction patterns. We propose STAR-Rec, a novel architecture that synergistically combines preference-aware attention and state-space modeling through a sequence-level mixture-of-experts framework. STAR-Rec addresses these challenges by: (1) employing preference-aware attention to capture both inherently similar item relationships and diverse preferences (2) utilizing state-space modeling to efficiently process variable-length sequences with linear complexity, and (3) incorporating a mixture-of-experts component that adaptively routes different behavioral patterns to specialized experts, handling both focused category-specific browsing and diverse category exploration patterns. We theoretically demonstrate how the state space model and attention mechanisms can be naturally unified in recommendation scenarios, where SSM captures temporal dynamics through state compression while attention models both similar and diverse item relationships. Extensive experiments on four real-world datasets demonstrate that STAR-Rec consistently outperforms state-of-the-art sequential recommendation methods, particularly in scenarios involving diverse user behaviors and varying sequence lengths. The implementation code is available anonymously online for easy reproducibility.

A question-answering (QA) simulator is a model that simulates human students QA behaviors. By leveraging QA history to estimate the probability of correctly answering a newly recommended question, the simulator enables the educational recommender systems to be trained in a simulated environment, protecting human students from the potential negative impact of low-quality recommendations. Despite its significant importance, the construction of QA simulators has not been thoroughly explored in the research domain of AI. Previous methods mainly rely on existing knowledge tracing (KT) models to construct such a simulator. However, due to the discrepancy between the KT task and the simulation task, those KT-based simulators suffer from severe bias accumulation, which limits the effectiveness of the simulation. In this paper, we propose a method called Diffusion-based Simulator (DSim), which takes advantage of diffusion to alleviate the bias accumulation. To our knowledge, DSim is the first to focus on building a QA simulator.

Imaging-based spatial transcriptomics (iST), such as MERFISH, CosMx SMI, and Xenium, quantify gene expression level across cells in space, but more importantly, they directly reveal the subcellular distribution of RNA transcripts at the single-molecule resolution. The subcellular localization of RNA molecules plays a crucial role in the compartmentalization-dependent regulation of genes within individual cells. Understanding the intracellular spatial distribution of RNA for a particular cell type thus not only improves the characterization of cell identity but also is of paramount importance in elucidating unique subcellular regulatory mechanisms specific to the cell type. However, current cell type annotation approaches of iST primarily utilize gene expression information while neglecting the spatial distribution of RNAs within cells. In this work, we introduce a semi-supervised graph contrastive learning method called Focus, the first method, to the best of our knowledge, that explicitly models RNA's subcellular distribution and community to improve cell type annotation. Focus demonstrates significant improvements over state-of-the-art algorithms across a range of spatial transcriptomics platforms, achieving improvements up to 27.8% in terms of accuracy and 51.9% in terms of F1-score for cell type annotation. Furthermore, Focus enjoys the advantages of intricate cell type-specific subcellular spatial gene patterns and providing interpretable subcellular gene analysis, such as defining the gene importance score. Importantly, with the importance score, Focus identifies genes harboring strong relevance to cell type-specific pathways, indicating its potential in uncovering novel regulatory programs across numerous biological systems.

Prompt tuning has been successfully used in leveraging the knowledge of Large-scale Vision-Language Pre-trained (VLP) models on downstream tasks. Most existing prompt tuning approaches learn prompts by maximizing the pairwise similarity. Although samples in different modalities might be relatively aligned pairwisely, such alignment does not fully utilize the information between samples, which can be less consistent on the modality level. In this paper, we propose a novel prompt tuning strategy by distributionally matching different modalities. Specifically, we minimize the distribution-wise distance between the image and text modalities with optimal transport (OT) theory. Simultaneously, we add a constraint on the learned transport plan during the modality matching to enhance the learning of vision and text prompts. Our proposed one can be applied to improve existing uni-modal and multi-modal prompt learning methods for being a plug-and-play method, which can generate modality-consistent representations. Experiments on eleven public datasets demonstrate that our proposed method has excellent performance, achieving substantial improvements on both uni-modal and multi-modal prompt tuning methods.

The cooperation in mixed cooperative-competitive tasks has drawn significant attention in multi-agent deep reinforcement learning. Agents need to cooperate with their teammates while competing against their opponents. However, most existing works treat the cooperative agents and competitive agents equally as they perform the same operation on all the agents. As a result, without distinguishing between cooperative and competitive agents, they may suffer from information disorder in learning an optimally cooperative policy and struggle to decide on the next step action. To address the above issues, we decompose the final Q-value into a weighted combination of three parts: the Q-values of the cooperative agents, the competitive group, and the current agent. A theoretical proof of the correctness of the decomposition is provided. With this decomposition, we are able to consider cooperative and competitive agents separately. Accordingly, we propose a multi-agent actor-critic algorithm called actor-double-attention-critic (ADAC) under centralized training and decentralized execution according to the decomposition. In ADAC, networks with group-specific attention and an attentional weighting network are specially designed. With the designed double-attention structure, ADAC can capture the distributions from different agents and improve cooperation performance. Extensive experiments are conducted in three scenarios with nine settings against six representative methods. The results demonstrate the superiority of the proposed ADAC model against state-of-the-art methods in various mixed cooperative-competitive tasks. The code is available at https://github.com/CrazyBayes/ADAC

Semi-supervised support vector machine (S3VM) is important because it can use plentiful unlabeled data to improve the generalization accuracy of traditional SVMs. In order to achieve good performance, it is necessary for S3VM to take some effective measures to select hyperparameters. However, model selection for semi-supervised models is still a key open problem. Existing methods for semi-supervised models to search for the optimal parameter values are usually computationally demanding, especially those ones with grid search. To address this challenging problem, in this article, we first propose solution paths of S3VM (SPS3VM), which can track the solutions of the nonconvex S3VM with respect to the hyperparameters. Specifically, we apply incremental and decremental learning methods to update the solution and let it satisfy the Karush-Kuhn-Tucker (KKT) conditions. Based on the SPS3VM and the piecewise linearity of model function, we can find the model with the minimum cross-validation (CV) error for the entire range of candidate hyperparameters by computing the error path of S3VM. Our SPS3VM is the first solution path algorithm for nonconvex optimization problem of semi-supervised learning models. We also provide the finite convergence analysis and computational complexity of SPS3VM. Experimental results on a variety of benchmark datasets not only verify that our SPS3VM can globally search the hyperparameters (regularization and ramp loss parameters) but also show a huge reduction of computational time while retaining similar or slightly better generalization performance compared with the grid search approach.

Artificial neural networks, especially recent diffusion-based models, have shown remarkable superiority in gaming, control, and QA systems, where the training tasks' datasets are usually static. However, in real-world applications, such as robotic control of reinforcement learning (RL), the tasks are changing, and new tasks arise in a sequential order. This situation poses the new challenge of plasticity-stability tradeoff for training an agent who can adapt to task changes and retain acquired knowledge. In view of this, we propose a rehearsal-based continual diffusion model, called continual diffuser (CoD), to endow the diffuser with the capabilities of quick adaptation (plasticity) and lasting retention (stability). Specifically, we first construct an offline benchmark that contains 90 tasks from multiple domains. Then, we train the CoD on each task with sequential modeling and conditional generation for making decisions. Next, we preserve a small portion of previous datasets as the rehearsal buffer and replay it to retain the acquired knowledge. Extensive experiments on a series of tasks show that CoD can achieve a promising plasticity-stability tradeoff and outperform existing diffusion-based methods and other representative baselines on most tasks. The source code is available at https://github.com/JF-Hu/Continual_Diffuser.

Multiagent policy gradients (MAPGs), an essential branch of reinforcement learning (RL), have made great progress in both industry and academia. However, existing models do not pay attention to the inadequate training of individual policies, thus limiting the overall performance. We verify the existence of imbalanced training in multiagent tasks and formally define it as an imbalance between policies (IBPs). To address the IBP issue, we propose a dynamic policy balance (DPB) model to balance the learning of each policy by dynamically reweighting the training samples. In addition, current methods for better performance strengthen the exploration of all policies, which leads to disregarding the training differences in the team and reducing learning efficiency. To overcome this drawback, we derive a technique named weighted entropy regularization (WER), a team-level exploration with additional incentives for individuals who exceed the team. DPB and WER are evaluated in homogeneous and heterogeneous tasks, effectively alleviating the imbalanced training problem and improving exploration efficiency. Furthermore, the experimental results show that our models can outperform the state-of-the-art MAPG methods and boast over 12.1% performance gain on average.

SDU learning, a weakly supervised learning problem with only pairwise similarities, dissimilarities data points and unlabeled data available, has many practical applications. However, it is still lacking in defense against adversarial samples, and its learning process can be expensive. To address this gap, we propose a novel adversarial training framework for SDU learning. Our approach reformulates the conventional minimax problem as an equivalent minimization problem based on the kernel perspective, departing from traditional confrontational training methods. Additionally, we employ the random gradient method and random features to accelerate the training process. Theoretical analysis shows that our method can converge to a stationary point at a rate of $\mathcal{O}(1/T^{1/4})$. Our experimental results show that our algorithm is superior to other adversarial training methods in terms of generalization, efficiency and scalability against various adversarial attacks.

Zeroth-order (ZO) methods, which use the finite difference of two function evaluations (also called ZO gradient) to approximate first-order gradient, have attracted much attention recently in machine learning because of their broad applications. The accuracy of the ZO gradient highly depends on how many finite differences are averaged, which are intrinsically determined by the number of perturbations randomly drawn from a distribution. Existing ZO methods try to learn a data-driven distribution for sampling the perturbations to improve the efficiency of ZO optimization (ZOO) algorithms. In this paper, we explore a new and parallel direction, \textit{i.e.}, learn an optimal sampling policy instead of using a totally random strategy to generate perturbations based on the techniques of reinforcement learning (RL), which makes it possible to approximate the gradient with only two function evaluations. Specifically, we first formulate the problem of learning a sampling policy as a Markov decision process. Then, we propose our ZO-RL algorithm, \textit{i.e.}, using deep deterministic policy gradient, an actor-critic RL algorithm to learn a sampling policy that can guide the generation of perturbed vectors in getting ZO gradients as accurately as possible. Importantly, the existing ZOO algorithms for learning a distribution can be plugged in to improve the exploration of ZO-RL. Experimental results with different ZO estimators show that our ZO-RL algorithm can effectively reduce the query complexity of ZOO algorithms and converge faster than existing ZOO algorithms, especially in the later stage of the optimization process.

Tabular data have been playing a mostly important role in diverse real-world fields, such as healthcare, engineering, finance, etc. With the recent success of deep learning, many tabular machine learning (ML) methods based on deep networks (e.g., Transformer, ResNet) have achieved competitive performance on tabular benchmarks. However, existing deep tabular ML methods suffer from the representation entanglement and localization, which largely hinders their prediction performance and leads to performance inconsistency on tabular tasks. To overcome these problems, we explore a novel direction of applying prototype learning for tabular ML and propose a prototype-based tabular representation learning framework, PTaRL, for tabular prediction tasks. The core idea of PTaRL is to construct prototype-based projection space (P-Space) and learn the disentangled representation around global data prototypes. Specifically, PTaRL mainly involves two stages: (i) Prototype Generating, that constructs global prototypes as the basis vectors of P-Space for representation, and (ii) Prototype Projecting, that projects the data samples into P-Space and keeps the core global data information via Optimal Transport. Then, to further acquire the disentangled representations, we constrain PTaRL with two strategies: (i) to diversify the coordinates towards global prototypes of different representations within P-Space, we bring up a diversifying constraint for representation calibration; (ii) to avoid prototype entanglement in P-Space, we introduce a matrix orthogonalization constraint to ensure the independence of global prototypes. Finally, we conduct extensive experiments in PTaRL coupled with state-of-the-art deep tabular ML models on various tabular benchmarks and the results have shown our consistent superiority.

In this work, we investigate the potential of large language models (LLMs) based agents to automate data science tasks, with the goal of comprehending task requirements, then building and training the best-fit machine learning models. Despite their widespread success, existing LLM agents are hindered by generating unreasonable experiment plans within this scenario. To this end, we present DS-Agent, a novel automatic framework that harnesses LLM agent and case-based reasoning (CBR). In the development stage, DS-Agent follows the CBR framework to structure an automatic iteration pipeline, which can flexibly capitalize on the expert knowledge from Kaggle, and facilitate consistent performance improvement through the feedback mechanism. Moreover, DS-Agent implements a low-resource deployment stage with a simplified CBR paradigm to adapt past successful solutions from the development stage for direct code generation, significantly reducing the demand on foundational capabilities of LLMs. Empirically, DS-Agent with GPT-4 achieves 100% success rate in the development stage, while attaining 36% improvement on average one pass rate across alternative LLMs in the deployment stage. In both stages, DS-Agent achieves the best rank in performance, costing $1.60 and \$0.13 per run with GPT-4, respectively. Our data and code are open-sourced at https://github.com/guosyjlu/DS-Agent.

Counterfactual estimation from observations represents a critical endeavor in numerous application fields, such as healthcare and finance, with the primary challenge being the mitigation of treatment bias. The balancing strategy aimed at reducing covariate disparities between different treatment groups serves as a universal solution. However, when it comes to the time series data, the effectiveness of balancing strategies remains an open question, with a thorough analysis of the robustness and applicability of balancing strategies still lacking. This paper revisits counterfactual estimation in the temporal setting and provides a brief overview of recent advancements in balancing strategies. More importantly, we conduct a critical empirical examination for the effectiveness of the balancing strategies within the realm of temporal counterfactual estimation in various settings on multiple datasets. Our findings could be of significant interest to researchers and practitioners and call for a reexamination of the balancing strategy in time series settings.

Bilevel optimization (BO) has recently gained prominence in many machine learning applications due to its ability to capture the nested structure inherent in these problems. Recently, many hypergradient methods have been proposed as effective solutions for solving large-scale problems. However, current hypergradient methods for the lower-level constrained bilevel optimization (LCBO) problems need very restrictive assumptions, namely, where optimality conditions satisfy the differentiability and invertibility conditions, and lack a solid analysis of the convergence rate. What's worse, existing methods require either double-loop updates, which are sometimes less efficient. To solve this problem, in this paper, we propose a new hypergradient of LCBO leveraging the theory of nonsmooth implicit function theorem instead of using the restrive assumptions. In addition, we propose a *single-loop single-timescale* algorithm based on the double-momentum method and adaptive step size method and prove it can return a $(\delta, \epsilon)$-stationary point with $\tilde{\mathcal{O}}(d_2^2\epsilon^{-4})$ iterations. Experiments on two applications demonstrate the effectiveness of our proposed method.

Estimating individual-level causal effects is crucial for decision-making in various domains, such as personalized healthcare, social marketing, and public policy. Addressing confounding bias is a critical step in accurately estimating the causal effects of treatments on outcomes. However, many current causal inference approaches consider all observed variables as confounders without distinguishing them from colliders or indirect (two-order) colliders. This may lead to M-bias when improperly eliminating confounding bias. In this study, we propose a new framework to accurately estimate individual-level treatment effects by considering a causal structure that includes both confounding variables and indirect colliders. Specifically, we first perform a sample reweighting to approximately eliminate confounding bias. Then, we restore the covariate’ potential latent parents and extract the modules solely related to the outcome. Finally, we take both these modules with the treatment variables to infer counterfactuals for causal inference. To validate the effectiveness of our proposed approach, we conduct extensive experiments on synthetic and commonly used semi-synthetic benchmark datasets. The experimental results demonstrate that our method outperforms current state-of-the-art methods.

Graph neural networks (GNNs) have exhibited remarkable performance under the assumption that test data comes from the same distribution of training data. However, in real-world scenarios, this assumption may not always be valid. Consequently, there is a growing focus on exploring the Out-of-Distribution (OOD) problem in the context of graphs. Most existing efforts have primarily concentrated on improving graph OOD generalization from two model-agnostic perspectives: data-driven methods and strategy-based learning. However, there has been limited attention dedicated to investigating the impact of well-known GNN model architectures on graph OOD generalization, which is orthogonal to existing research. In this work, we provide the first comprehensive investigation of OOD generalization on graphs from an architecture perspective, by examining the common building blocks of modern GNNs. Through extensive experiments, we reveal that both the graph self-attention mechanism and the decoupled architecture contribute positively to graph OOD generalization. In contrast, we observe that the linear classification layer tends to compromise graph OOD generalization capability. Furthermore, we provide in-depth theoretical insights and discussions to underpin these discoveries. These insights have empowered us to develop a novel GNN backbone model, DGat, designed to harness the robust properties of both graph self-attention mechanism and the decoupled architecture. Extensive experimental results demonstrate the effectiveness of our model under graph OOD, exhibiting substantial and consistent enhancements across various training strategies. Our codes are available at https://github.com/KaiGuo20/DGAT **REMOVE 2nd URL**://github.com/KaiGuo20/DGAT.

In this paper, we introduce a novel framework for black-box prompt tuning with a subspace learning and selection strategy, leveraging derivative-free optimization algorithms. This approach is crucial for scenarios where user interaction with language models is restricted to API usage, without direct access to their internal structures or gradients, a situation typical in Language-Model-as-a-Service (LMaaS). Our framework focuses on exploring the low-dimensional subspace of continuous prompts. Previous work on black-box prompt tuning necessitates a substantial number of API calls due to the random choice of the subspace. To tackle this problem, we propose to use a simple zeroth-order optimization algorithm to tackle nonconvex optimization challenges with nonsmooth nonconvex regularizers: the Zeroth-Order Mini-Batch Stochastic Proximal Gradient method (ZO-MB-SPG). A key innovation is the incorporation of nonsmooth nonconvex regularizers, including the indicator function of the l0 constraint, which enhances our ability to select optimal subspaces for prompt optimization. The experimental results show that our proposed black-box prompt tuning method on a few labeled samples can attain similar performance to the methods applicable to LMaaS with much fewer API calls.

Most fake news detection methods learn latent feature representations based on neural networks, which makes them black boxes to classify a piece of news without giving any justification. Existing explainable systems generate veracity justifications from investigative journalism, which suffer from debunking delayed and low efficiency. Recent studies simply assume that the justification is equivalent to the majority opinions expressed in the wisdom of crowds. However, the opinions typically contain some inaccurate or biased information since the wisdom of crowds is uncensored. To detect fake news from a sea of diverse, crowded and even competing narratives, in this paper, we propose a novel defense-based explainable fake news detection framework. Specifically, we first propose an evidence extraction module to split the wisdom of crowds into two competing parties and respectively detect salient evidences. To gain concise insights from evidences, we then design a prompt-based module that utilizes a large language model to generate justifications by inferring reasons towards two possible veracities. Finally, we propose a defense-based inference module to determine veracity via modeling the defense among these justifications. Extensive experiments conducted on two real-world benchmarks demonstrate that our proposed method outperforms state-of-the-art baselines in terms of fake news detection and provides high-quality justifications.

Along with the vast application of Internet of Things (IoT) and the ever-growing concerns about data protection, a novel type of learning, named incremental federated learning (IFL), is rising to further elevate the intelligence and quality of various IoT systems and services by consistently learning and updating their models, e.g., deep neural networks, in dynamic contexts, where clients and data can increase and accumulate gradually. Since IFL is still in its infancy, to overcome its emerging challenges as represented in 1) periodic learning about how to initialize the model update rationally to avoid catastrophic performance dropping, and 2) iterative learning about how to update the model cost-efficiently to remedy overlearning on duplicated information, this paper proposes a stage-based and layer-wise mechanism for IFL, called SLMFed, in which, the periodic learning is managed by a stage transition and client selection strategy to trigger model update according to the quantitative and qualitative changes on clients, data, and user experience, and the iterative learning is enhanced by an adaptive layer uploading and aggregation strategy to update the global model by measuring representational consistencies and information richness of local model layers. As shown by the evaluation results, SLMFed can not only stabilize the learning across various learning stages but also boost the performance in terms of learning accuracy, communication cost, and stage contribution by about 32.09%, 105.94%, and 22.02%, respectively.

Policy diversity, encompassing the variety of policies an agent can adopt, enhances reinforcement learning (RL) success by fostering more robust, adaptable, and innovative problem-solving in the environment. The environment in which standard RL operates is usually modeled with a Markov Decision Process (MDP) as the theoretical foundation. However, in many real-world scenarios, the rewards depend on an agent's history of states and actions leading to a non-MDP. Under the premise of policy diffusion initialization, non-MDPs may have unstructured expanding solution space due to varying historical information and temporal dependencies. This results in solutions having non-equivalent closed forms in non-MDPs. In this paper, deriving diverse solutions for non-MDPs requires policies to break through the boundaries of the current solution space through gradual dispersion. The goal is to expand the solution space, thereby obtaining more diverse policies. Specifically, we first model the sequences of states and actions by a transformer-based method to learn policy embeddings for dispersion in the solution space, since the transformer has advantages in handling sequential data and capturing long-range dependencies for non-MDP. Then, we stack the policy embeddings to construct a dispersion matrix as the policy diversity measure to induce the policy dispersion in the solution space and obtain a set of diverse policies. Finally, we prove that if the dispersion matrix is positive definite, the dispersed embeddings can effectively enlarge the disagreements across policies, yielding a diverse expression for the original policy embedding distribution. Experimental results of both non-MDP and MDP environments show that this dispersion scheme can obtain more expressive diverse policies via expanding the solution space, showing more robust performance than the recent learning baselines.

Large language models (LLMs) are gaining increasing popularity in both academia and industry, owing to their unprecedented performance in various applications. As LLMs continue to play a vital role in both research and daily use, their evaluation becomes increasingly critical, not only at the task level, but also at the society level for better understanding of their potential risks. Over the past years, significant efforts have been made to examine LLMs from various perspectives. This paper presents a comprehensive review of these evaluation methods for LLMs, focusing on three key dimensions: what to evaluate , where to evaluate , and how to evaluate . Firstly, we provide an overview from the perspective of evaluation tasks, encompassing general natural language processing tasks, reasoning, medical usage, ethics, education, natural and social sciences, agent applications, and other areas. Secondly, we answer the ‘where’ and ‘how’ questions by diving into the evaluation methods and benchmarks, which serve as crucial components in assessing the performance of LLMs. Then, we summarize the success and failure cases of LLMs in different tasks. Finally, we shed light on several future challenges that lie ahead in LLMs evaluation. Our aim is to offer invaluable insights to researchers in the realm of LLMs evaluation, thereby aiding the development of more proficient LLMs. Our key point is that evaluation should be treated as an essential discipline to better assist the development of LLMs. We consistently maintain the related open-source materials at: https://github.com/MLGroupJLU/LLM-eval-survey

Artificial Intelligence (AI) has achieved a wide range of successes in autonomous air combat decision-making recently. Previous research demonstrated that AI-enabled air combat approaches could even acquire beyond human-level capabilities. However, there remains a lack of evidence regarding two major difficulties. First, the existing methods with fixed decision intervals are mostly devoted to solving what to act but merely pay attention to when to act, which occasionally misses optimal decision opportunities. Second, the method of an expert-crafted finite maneuver library leads to a lack of tactics diversity, which is vulnerable to an opponent equipped with new tactics. In view of this, we propose a novel Deep Reinforcement Learning (DRL) and prior knowledge hybrid autonomous air combat tactics discovering algorithm, namely deep E xcitatory-i N hibitory f ACT or I zed maneu VE r ( ENACTIVE ) learning. The algorithm consists of two key modules, i.e., ENHANCE and FACTIVE. Specifically, ENHANCE learns to adjust the air combat decision-making intervals and appropriately seize key opportunities. FACTIVE factorizes maneuvers and then jointly optimizes them with significant tactics diversity increments. Extensive experimental results reveal that the proposed method outperforms state-of-the-art algorithms with a 62% winning rate and further obtains a margin of a 2.85-fold increase in terms of global tactic space coverage. It also demonstrates that a variety of discovered air combat tactics are comparable to human experts’ knowledge.

Estimating individual treatment effect (ITE) from observational data has attracted great interest in recent years, which plays a crucial role in decision-making across many high-impact domains such as economics, medicine, and e-commerce. Most existing studies of ITE estimation assume that different units at play are independent and do not influence each other. However, many social science experiments have shown that there often exist different levels of interactions between units in observational data, especially in a networked environment. As a result, the treatment assignment of one unit can affect the outcome of other units connected to it in the network, which is referred to as the interference or spillover effect . In this article, we study an important problem of ITE estimation from networked observational data by modeling the interference between different units and provide a principled framework to support such study. Methodologically, we propose a novel framework, SPNet , that first captures the influence of hidden confounders with the aid of graph convolutional network and then models the interference by introducing an environment summary variable and developing a masked attention mechanism. Experimental evaluations on several semi-synthetic datasets based on real-world networks corroborate the superiority of our proposed framework over state-of-the-art individual treatment effect estimation methods.

Offline reinforcement learning (RL) makes it possible to train the agents entirely from a previously collected dataset. However, constrained by the quality of the offline dataset, offline RL agents typically have limited performance and cannot be directly deployed. Thus, it is desirable to further finetune the pretrained offline RL agents via online interactions with the environment. Existing offline-to-online RL algorithms suffer from the low sample efficiency issue, due to two inherent challenges, i.e., exploration limitation and distribution shift. To this end, we propose a sample-efficient offline-to-online RL algorithm via Optimistic Exploration and Meta Adaptation (OEMA). Specifically, we first propose an optimistic exploration strategy according to the principle of optimism in the face of uncertainty. This allows agents to sufficiently explore the environment in a stable manner. Moreover, we propose a meta learning based adaptation method, which can reduce the distribution shift and accelerate the offline-to-online adaptation process. We empirically demonstrate that OEMA improves the sample efficiency on D4RL benchmark. Besides, we provide in-depth analyses to verify the effectiveness of both optimistic exploration and meta adaptation.

In this study, we present a transductive inference approach on that reward information propagation graph, which enables the effective estimation of rewards for unlabelled data in offline reinforcement learning. Reward inference is the key to learning effective policies in practical scenarios, while direct environmental interactions are either too costly or unethical and the reward functions are rarely accessible, such as in healthcare and robotics. Our research focuses on developing a reward inference method based on the contextual properties of information propagation on graphs that capitalizes on a constrained number of human reward annotations to infer rewards for unlabelled data. We leverage both the available data and limited reward annotations to construct a reward propagation graph, wherein the edge weights incorporate various influential factors pertaining to the rewards. Subsequently, we employ the constructed graph for transductive reward inference, thereby estimating rewards for unlabelled data. Furthermore, we establish the existence of a fixed point during several iterations of the transductive inference process and demonstrate its at least convergence to a local optimum. Empirical evaluations on locomotion and robotic manipulation tasks validate the effectiveness of our approach. The application of our inferred rewards improves the performance in offline reinforcement learning tasks.

Graph neural networks (GNNs) are designed to perform inference on data described by graph-structured node features and topology information. From the perspective of graph signal denoising, the typical message passing schemes of GNNs act as a globally uniform smoothing that minimizes disagreements between embeddings of connected nodes. However, the level of smoothing over different regions of the graph should be different, especially for those inter-class regions. This deviation limits the expressiveness of GNNs, and then renders them fragile to over-smoothing, long-range dependencies, and non-homophily settings. In this paper, we find that the node disagreements of initial graph features can present more trustworthy constraints on node embeddings, thereby enhancing the locally adaptive smoothing of GNNs. To spread the inherent disagreements of nodes, we propose the Laplacian node disagreement to jointly measure the initial features and output embeddings. With such a measurement, we then present a new graph signal denoising objective deriving a more effective message passing scheme and further incorporate it into the GNN architecture, named Laplacian node disagreement-based GNN (LND-GNN). Learning from its output node representations, we integrate an auxiliary disagreement constraint into the overall classification loss. Experiments demonstrate the expressive ability of LND-GNN in the downstream semi-supervised node classification task.

With the growing concerns on data security and user privacy, a decentralized mechanism is implemented for federated data mining (FDM), which can bridge data silos and collaborate diverse devices in ubiquitous IoT (Internet of Things) systems and services to extract global and shareable knowledge, i.e., encoded in deep neural networks (DDNs). Moreover, compared with FDM in synchronous mode, asynchronous FDM (AFDM) is more suitable to accommodate devices with diversified computing resources and distinguishable working statuses. However, as AFDM is still in its infancy, how to harness heterogeneous resources and biased knowledge of learning participants within the asynchronous context remains to be addressed. Such that, this paper proposes an adaptive and integrated mechanism, named AiFed, in which, a layer-wise optimization of AFDM is implemented based on the integration of two dedicated strategies, i.e., an adaptive local model uploading strategy (ALMU), and an adaptive global model aggregation strategy (AGMA). As shown by the evaluation results, AiFed can outperform five state-of-the-art methods to reduce communication costs by about 61.76% and 56.88%, improve learning accuracy by about 1.66% and 3.05%, and accelerate learning speed by about 22.16% and 37.81% under IID (independent and identically distributed) and Non-IID settings of four standard datasets, respectively.

Document distance is a fundamental yet significant research topic in the information retrieval community, and its accuracy dominates the performance of many text retrieval applications. Beyond the Bag-of-Words (BoW) model, the Word Mover's Distance (WMD) semantically defines the distance between documents as the minimum cost (i.e., measured by word similarities of embeddings) required to transport the words from one document to another, and it has been proven to be superior by k -nearest neighbor classification. In this article, we thoroughly study the characteristics of WMD and its relaxed versions, e.g., Relaxed WMD (RWMD) and Iterative Constrained Transfers (ICT), in various scenarios. Specifically, we concentrate on the problem of negative word similarity: the WMD family leverages all word similarities, however, most of them are meaningless, resulting in negative effects for measuring document distances. To remedy this problem, we propose Informative Similarity Filter (ISF), which retains a very small percentage of top word similarities and fixes the others as the same lower similarity. Built on it, we propose a greedy optimization (GOM) for WMD, an accurate approximation to WMD. We theoretically analyze that ISF-GOM is more applicable for relatively longer documents. Extensive experiments have been conducted to validate: 1) the problem of RWMD; 2) the effectiveness of ISF-GOM; and 3) the consistence of our analysis of ISF-GOM. Our codes and datasets are available at https://github.com/BoCheng-96/ISF-GOM.

Many graph neural networks (GNNs) are inapplicable when the graph structure representing the node relations is unavailable. Recent studies have shown that this problem can be effectively solved by jointly learning the graph structure and the parameters of GNNs. However, most of these methods learn graphs by using either a Euclidean or hyperbolic metric, which means that the space curvature is assumed to be either constant zero or constant negative. Graph embedding spaces usually have nonconstant curvatures, and thus, such an assumption may produce some obfuscatory nodes, which are improperly embedded and close to multiple categories. In this article, we propose a joint-space graph learning (JSGL) method for GNNs. JSGL learns a graph based on Euclidean embeddings and identifies Euclidean obfuscatory nodes. Then, the graph topology near the identified obfuscatory nodes is refined in hyperbolic space. We also present a theoretical justification of our method for identifying obfuscatory nodes and conduct a series of experiments to test the performance of JSGL. The results show that JSGL outperforms many baseline methods. To obtain more insights, we analyze potential reasons for this superior performance.

Graph neural networks (GNNs) have achieved great success in many fields due to their powerful capabilities of processing graph-structured data. However, most GNNs can only be applied to scenarios where graphs are known, but real-world data are often noisy or even do not have available graph structures. Recently, graph learning has attracted increasing attention in dealing with these problems. In this article, we develop a novel approach to improving the robustness of the GNNs, called composite GNN. Different from existing methods, our method uses composite graphs (C-graphs) to characterize both sample and feature relations. The C-graph is a unified graph that unifies these two kinds of relations, where edges between samples represent sample similarities, and each sample has a tree-based feature graph to model feature importance and combination preference. By jointly learning multiaspect C-graphs and neural network parameters, our method improves the performance of semisupervised node classification and ensures robustness. We conduct a series of experiments to evaluate the performance of our method and the variants of our method that only learn sample relations or feature relations. Extensive experimental results on nine benchmark datasets demonstrate that our proposed method achieves the best performance on almost all the datasets and is robust to feature noises.

In the era of information explosion, named entity recognition (NER) has attracted widespread attention in the field of natural language processing, as it is fundamental to information extraction. Recently, methods of NER based on representation learning, e.g., character embedding and word embedding, have demonstrated promising recognition results. However, existing models only consider partial features derived from words or characters while failing to integrate semantic and syntactic information, e.g., capitalization, inter-word relations, keywords, and lexical phrases, from multilevel perspectives. Intuitively, multilevel features can be helpful when recognizing named entities from complex sentences. In this study, we propose a novel attentive multilevel feature fusion (AMFF) model for NER, which captures the multilevel features in the current context from various perspectives. It consists of four components to, respectively, capture the local character-level (CL), global character-level (CG), local word-level (WL), and global word-level (WG) features in the current context. In addition, we further define document-level features crafted from other sentences to enhance the representation learning of the current context. To this end, we introduce a novel context-aware attentive multilevel feature fusion (CAMFF) model based on AMFF, to fully leverage document-level features from all the previous inputs. The obtained multilevel features are then fused and fed into a bidirectional long short-term memory (BiLSTM)-conditional random field (CRF) network for the final sequence labeling. Extensive experiments on four benchmark datasets demonstrate that our proposed AMFF and CAMFF models outperform a set of state-of-the-art baseline methods and the features learned from multiple levels are complementary.

The popular Proximal Policy Optimization (PPO) algorithm approximates the solution in a clipped policy space. Does there exist better policies outside of this space? By using a novel surrogate objective that employs the sigmoid function (which provides an interesting way of exploration), we found that the answer is "YES", and the better policies are in fact located very far from the clipped space. We show that PPO is insufficient in "off-policyness", according to an off-policy metric called DEON. Our algorithm explores in a much larger policy space than PPO, and it maximizes the Conservative Policy Iteration (CPI) objective better than PPO during training. To the best of our knowledge, all current PPO methods have the clipping operation and optimize in the clipped policy space. Our method is the first of this kind, which advances the understanding of CPI optimization and policy gradient methods. Code is available at https://github.com/raincchio/P3O.

This paper studies learning meaningful node representations for signed graphs, where both positive and negative links exist. This problem has been widely studied by meticulously designing expressive signed graph neural networks, as well as capturing the structural information of the signed graph through traditional structure decomposition methods, e.g., spectral graph theory. In this paper, we propose a novel signed graph representation learning framework, called Signed Laplacian Graph Neural Network (SLGNN), which combines the advantages of both. Specifically, based on spectral graph theory and graph signal processing, we first design different low-pass and high-pass graph convolution filters to extract low-frequency and high-frequency information on positive and negative links, respectively, and then combine them into a unified message passing framework. To effectively model signed graphs, we further propose a self-gating mechanism to estimate the impacts of low-frequency and high-frequency information during message passing. We mathematically establish the relationship between the aggregation process in SLGNN and signed Laplacian regularization in signed graphs, and theoretically analyze the expressiveness of SLGNN. Experimental results demonstrate that SLGNN outperforms various competitive baselines and achieves state-of-the-art performance.

Unsupervised Anomaly Detection (UAD) is a key data mining problem owing to its wide real-world applications. Due to the complete absence of supervision signals, UAD methods rely on implicit assumptions about anomalous patterns (e.g., scattered/sparsely/densely clustered) to detect anomalies. However, real-world data are complex and vary significantly across different domains. No single assumption can describe such complexity and be valid in all scenarios. This is also confirmed by recent research that shows no UAD method is omnipotent [1]. Based on above observations, instead of searching for a magic universal winner assumption, we seek to design a general UAD Booster (UADB) that empowers any UAD models with adaptability to different data. This is a challenging task given the heterogeneous model structures and assumptions adopted by existing UAD methods. To achieve this, we dive deep into the UAD problem and find that compared to normal data, anomalies (i) lack clear structure/pattern in feature space, thus (ii) harder to learn by model without a suitable assumption, and finally, leads to (iii) high variance between different learners. In light of these findings, we propose to (i) distill the knowledge of the source UAD model to an imitation learner (booster) that holds no data assumption, then (ii) exploit the variance between them to perform automatic correction, and thus (iii) improve the booster over the original UAD model. We use a neural network as the booster for its strong expressive power as a universal approximator and ability to perform flexible posthoc tuning. Note that UADB is a model-agnostic framework that can enhance heterogeneous UAD models in a unified way. Extensive experiments on over 80 tabular datasets demonstrate the effectiveness of UADB. To facilitate further research, code, figures, and datasets are available at UADB’s Github repository1.

Out-of-distribution generalization is challenging for non-participating clients of federated learning under distribution shifts. A proven strategy is to explore those invariant relationships between input and target variables, working equally well for non-participating clients. However, learning invariant relationships is often in an explicit manner from data, representation, and distribution, which violates the federated principles of privacy-preserving and limited communication. In this paper, we propose FedIIR, which implicitly learns invariant relationships from parameter for out-of-distribution generalization, adhering to the above principles. Specifically, we utilize the prediction disagreement to quantify invariant relationships and implicitly reduce it through inter-client gradient alignment. Theoretically, we demonstrate the range of non-participating clients to which FedIIR is expected to generalize and present the convergence results for FedIIR in the massively distributed with limited communication. Extensive experiments show that FedIIR significantly outperforms relevant baselines in terms of out-of-distribution generalization of federated learning.

The aging of global population is witnessing increasing prevalence of spinal disorders. According to latest statistics, nearly five percent of the global population is suffering from spinal disorders. To relieve the pain, many spine patients tend to choose surgeries. However, recent evidences reveal that some spine patients can self-heal over time with nonoperative treatment and even surgeries may not ease the pain for some others, which raises a critical question regarding the appropriateness of such surgeries. Furthermore, the complex and time-consuming diagnostic process places a great burden on both clinicians and patients. Due to the development of web technology, it is possible for spine patients to obtain decision making suggestions on the Internet. The uniqueness of web technology, including its popularity, convenience, and immediacy, makes intelligent healthcare techniques, especially Treatment Outcome Forecasting (TOF), able to support clinical decision-making for doctors and healthcare providers. Despite a few machine-learning-based methods have been proposed for TOF, their performance and feasibility are mostly unsatisfactory due to the neglect of a few practical challenges (caused by applying on the Internet), including biased data selection, noisy supervision, and patient noncompliance. In light of this, we propose DeepTOF, a novel end-to-end deep learning model to cope with the unique challenges in web-based long-term continuous spine TOF. In particular, we combine different patient groups and train a unified predictive model to eliminate the data selection bias. Towards robust learning, we further take advantage of indirect but fine-grained supervision signals to mutually calibrate with the noisy training labels. Additionally, a feature selector was co-trained with DeepTOF to select the most important features (i.e., answers/indicators that need to be collected) for inference, thus easing the use of DeepTOF during web-based real-world application. The proposed DeepTOF could bring great benefits to the rehabilitation of spine patients. Comprehensive experiments and analysis show that DeepTOF outperforms conventional solutions by a large margin.

In visual-based Reinforcement Learning (RL), agents often struggle to generalize well to environmental variations in the state space that were not observed during training. The variations can arise in both task-irrelevant features, such as background noise, and task-relevant features, such as robot configurations, that are related to the optimal decisions. To achieve generalization in both situations, agents are required to accurately understand the impact of changed features on the decisions, i.e., establishing the true associations between changed features and decisions in the policy model. However, due to the inherent correlations among features in the state space, the associations between features and decisions become entangled, making it difficult for the policy to distinguish them. To this end, we propose Saliency-Guided Features Decorrelation (SGFD) to eliminate these correlations through sample reweighting. Concretely, SGFD consists of two core techniques: Random Fourier Functions (RFF) and the saliency map. RFF is utilized to estimate the complex non-linear correlations in high-dimensional images, while the saliency map is designed to identify the changed features. Under the guidance of the saliency map, SGFD employs sample reweighting to minimize the estimated correlations related to changed features, thereby achieving decorrelation in visual RL tasks. Our experimental results demonstrate that SGFD can generalize well on a wide range of test environments and significantly outperforms state-of-the-art methods in handling both task-irrelevant variations and task-relevant variations.

Recently, low-rank tensor recovery methods based on subspace representation have received increased attention in the field of hyperspectral image (HSI) denoising. Unfortunately, those methods usually analyze the prior structural information within different dimensions indiscriminately, ignoring the differences between modes, leaving substantial room for improvement. In this article, we first consider the low-rank properties in the subspace and prove that the structure correlation across the nonlocal self-similarity mode is much stronger than in the spatial sparsity and spectral correlation modes. On that basis, we introduce a new multidirectional low-rank regularization, in which each mode is assigned a different weight to characterize its contribution to estimating the tensor rank. After that, integrating the proposed regularization with the subspace-based tensor recovery framework, an optimization model for HSI mixed noise removal is developed. The proposed model can be addressed efficiently via the alternating minimization algorithm. Extensive experiments implemented with synthetic and real data demonstrate that the proposed method significantly outperforms other state-of-the-art HSI denoising methods, which clearly indicates the effectiveness of the proposed approach in HSI denoising.

Graph structured data has wide applicability in various domains such as physics, chemistry, biology, computer vision, and social networks, to name a few. Recently, graph neural networks (GNN) were shown to be successful in effectively representing graph structured data because of their good performance and generalization ability. However, explaining the effectiveness of GNN models is a challenging task because of the complex nonlinear transformations made over the iterations. In this paper, we propose GraphLIME, a local interpretable model explanation for graphs using the Hilbert-Schmidt Independence Criterion (HSIC) Lasso, which is a nonlinear feature selection method. GraphLIME is a generic GNN-model explanation framework that learns a nonlinear interpretable model locally in the subgraph of the node being explained. Through experiments on two real-world datasets, the explanations of GraphLIME are found to be of extraordinary degree and more descriptive in comparison to the existing explanation methods.

Graph neural networks (GNNs) have received tremendous attention due to their superiority in learning node representations. These models rely on message passing and feature transformation functions to encode the structural and feature information from neighbors. However, stacking more convolutional layers significantly decreases the performance of GNNs. Most recent studies attribute this limitation to the over-smoothing issue, where node embeddings converge to indistinguishable vectors. Through a number of experimental observations, we argue that the main factor degrading the performance is the unstable forward normalization and backward gradient resulted from the improper design of the feature transformation, especially for shallow GNNs where the over-smoothing has not happened. Therefore, we propose a novel orthogonal feature transformation, named Ortho-GConv, which could generally augment the existing GNN backbones to stabilize the model training and improve the model's generalization performance. Specifically, we maintain the orthogonality of the feature transformation comprehensively from three perspectives, namely hybrid weight initialization, orthogonal transformation, and orthogonal regularization. By equipping the existing GNNs (e.g. GCN, JKNet, GCNII) with Ortho-GConv, we demonstrate the generality of the orthogonal feature transformation to enable stable training, and show its effectiveness for node and graph classification tasks.

Existing fake news detection methods aim to classify a piece of news as true or false and provide veracity explanations, achieving remarkable performances. However, they often tailor automated solutions on manual fact-checked reports, suffering from limited news coverage and debunking delays. When a piece of news has not yet been fact-checked or debunked, certain amounts of relevant raw reports are usually disseminated on various media outlets, containing the wisdom of crowds to verify the news claim and explain its verdict. In this paper, we propose a novel Coarse-to-fine Cascaded Evidence-Distillation (CofCED) neural network for explainable fake news detection based on such raw reports, alleviating the dependency on fact-checked ones. Specifically, we first utilize a hierarchical encoder for web text representation, and then develop two cascaded selectors to select the most explainable sentences for verdicts on top of the selected top-K reports in a coarse-to-fine manner. Besides, we construct two explainable fake news datasets, which is publicly available. Experimental results demonstrate that our model significantly outperforms state-of-the-art detection baselines and generates high-quality explanations from diverse evaluation perspectives.

Weakly-supervised text classification aims to train a classifier using only class descriptions and unlabeled data.Recent research shows that keyword-driven methods can achieve state-ofthe-art performance on various tasks.However, these methods not only rely on carefullycrafted class descriptions to obtain classspecific keywords but also require substantial amount of unlabeled data and takes a long time to train.This paper proposes FastClass, an efficient weakly-supervised classification approach.It uses dense text representation to retrieve class-relevant documents from external unlabeled corpus and selects an optimal subset to train a classifier.Compared to keyworddriven methods, our approach is less reliant on initial class descriptions as it no longer needs to expand each class description into a set of class-specific keywords.Experiments on a wide range of classification tasks show that the proposed approach frequently outperforms keyword-driven models in terms of classification accuracy and often enjoys orders-ofmagnitude faster training speed.

Recently, discrete latent variable models have received a surge of interest in both Natural Language Processing (NLP) and Computer Vision (CV), attributed to their comparable performance to the continuous counterparts in representation learning, while being more interpretable in their predictions.In this paper, we develop a topic-informed discrete latent variable model for semantic textual similarity, which learns a shared latent space for sentence-pair representation via vector quantization.Compared with previous models limited to local semantic contexts, our model can explore richer semantic information via topic modeling.We further boost the performance of semantic similarity by injecting the quantized representation into a transformer-based language model with a well-designed semanticdriven attention mechanism.We demonstrate, through extensive experiments across various English language datasets, that our model is able to surpass several strong neural baselines in semantic textual similarity tasks.

Multi-agent reinforcement learning has drawn increasing attention in practice, e.g., robotics and automatic driving, as it can explore optimal policies using samples generated by interacting with the environment. However, high reward uncertainty still remains a problem when we want to train a satisfactory model, because obtaining high-quality reward feedback is usually expensive and even infeasible. To handle this issue, previous methods mainly focus on passive reward correction. At the same time, recent active reward estimation methods have proven to be a recipe for reducing the effect of reward uncertainty. In this paper, we propose a novel Distributional Reward Estimation framework for effective Multi-Agent Reinforcement Learning (DRE-MARL). Our main idea is to design the multi-action-branch reward estimation and policy-weighted reward aggregation for stabilized training. Specifically, we design the multi-action-branch reward estimation to model reward distributions on all action branches. Then we utilize reward aggregation to obtain stable updating signals during training. Our intuition is that consideration of all possible consequences of actions could be useful for learning policies. The superiority of the DRE-MARL is demonstrated using benchmark multi-agent scenarios, compared with the SOTA baselines in terms of both effectiveness and robustness.

Federated Learning in asynchronous mode (AFL) is attracting much attention from both industry and academia to build intelligent cores for various Internet of Things (IoT) systems and services by harnessing sensitive data and idle computing resources dispersed at massive IoT devices in a privacy-preserving and interaction-unblocking manner. Since AFL is still in its infancy, it encounters three challenges that need to be resolved jointly, namely: 1) how to rationally utilize AFL clients, whose local data grow gradually, to avoid overlearning issues; 2) how to properly manage the client-server interaction with both communication cost reduced and model performance improved; and finally, 3) how to effectively and efficiently aggregate heterogeneous parameters received at the server to build a global model. To fill the gap, this article proposes a triple-step asynchronous federated learning mechanism (TrisaFed), which can: 1) activate clients with rich information according to an informative client activating strategy (ICA); 2) optimize the client–server interaction by a multiphase layer updating strategy (MLU); and 3) enhance the model aggregation function by a temporal weight fading strategy (TWF), and an informative weight enhancing strategy (IWE). Moreover, based on four standard data sets, TrisaFed is evaluated. As shown by the result, compared with four state-of-the-art baselines, TrisaFed can not only dramatically reduce the communication cost by over 80% but also can significantly improve the learning performance in terms of model accuracy and training speed by over 8% and 70%, respectively.

Patient stratification has been studied widely to tackle subtype diagnosis problems for effective treatment. Due to the dimensionality curse and poor interpretability of data, there is always a long-lasting challenge in constructing a stratification model with high diagnostic ability and good generalization. To address these problems, this article proposes two novel evolutionary multiobjective clustering algorithms with ensemble (NSGA-II-ECFE and MOEA/D-ECFE) with four cluster validity indices used as the objective functions. First, an effective ensemble construction method is developed to enrich the ensemble diversity. After that, an ensemble clustering fitness evaluation (ECFE) method is proposed to evaluate the ensembles by measuring the consensus clustering under those four objective functions. To generate the consensus clustering, ECFE exploits the hybrid co-association matrix from the ensembles and then dynamically selects the suitable clustering algorithm on that matrix. Multiple experiments have been conducted to demonstrate the effectiveness of the proposed algorithm in comparison with seven clustering algorithms, twelve ensemble clustering approaches, and two multiobjective clustering algorithms on 55 synthetic datasets and 35 real patient stratification datasets. The experimental results demonstrate the competitive edges of the proposed algorithms over those compared methods. Furthermore, the proposed algorithm is applied to extend its advantages by identifying cancer subtypes from five cancer-related single-cell RNA-seq datasets.

With the boosting of neural networks, recommendation methods become significantly improved by their powerful ability of prediction and inference. Existing neural-network based recommender systems (NN-RSs) usually first employ matrix embedding (ME) as a pre-process to learn users’ and items’ representations (latent vectors), then input these representations to a specific modified neural network framework to make accurate Top-k recommendations. Obviously, the performance of ME has a significant effect on RS models. However, most NN-RSs focus on accuracy by building representations from the direct user-item interactions (e.g., user-item rating matrix), while ignoring the underlying relatedness between users and items (e.g., users who rate the same ratings for the same items should be embedded into similar representations), which is an ideological disadvantage. On the other hand, ME models directly employ inner products as a default loss function metric that cannot project users and items into a proper latent space, which is a methodological disadvantage. In this paper, we propose a supervised collaborative representation learning model - Magnetic Metric Learning (MML) - to map users and items into a unified latent vector space, enhancing the representation learning for NN-RSs. First, MML utilizes dual triplets to model not only the observed relationships between users and items, but also the underlying relationships between users as well as items to overcome the ideological disadvantage. Specifically, a modified metric-based dual loss function is proposed in MML to gather similar entities and disperse the dissimilar ones. With MML, we can easily compare all the relationships (user to user, item to item, user to item) according to the weighted metric, which overcomes the methodological disadvantage. We conduct extensive experiments on four real-world datasets with large item space. The results demonstrate that MML can learn a proper unified latent space for representations from the user-item matrix with high accuracy and effectiveness, and lead to a performance gain over the state-of-the-art RS models by an average of 17 percent.

In this article, we propose a Latent Dirichlet Allocation– (LDA) based topic-graph probabilistic personalization model for Web search. This model represents a user graph in a latent topic graph and simultaneously estimates the probabilities that the user is interested in the topics, as well as the probabilities that the user is not interested in the topics. For a given query issued by the user, the webpages that have higher relevancy to the interested topics are promoted, and the webpages more relevant to the non-interesting topics are penalized. In particular, we simulate a user’s search intent by building two profiles: A positive user profile for the probabilities of the user is interested in the topics and a corresponding negative user profile for the probabilities of being not interested in the the topics. The profiles are estimated based on the user’s search logs. A clicked webpage is assumed to include interesting topics. A skipped (viewed but not clicked) webpage is assumed to cover some non-interesting topics to the user. Such estimations are performed in the latent topic space generated by LDA. Moreover, a new approach is proposed to estimate the correlation between a given query and the user’s search history so as to determine how much personalization should be considered for the query. We compare our proposed models with several strong baselines including state-of-the-art personalization approaches. Experiments conducted on a large-scale real user search log collection illustrate the effectiveness of the proposed models.

The task of reasoning over knowledge graph for factoid questions has received significant interest from the research community of natural language processing. Performing this task inevitably faces the issues of question complexity and reasoning efficiency. In this paper, we investigate modern reasoning approaches over knowledge graph to tackle complex factoid questions of diverse reasoning schemas with attractive speedup in computational efficiency. To this end, we propose two evidence retrieval strategies to generate concise and informative evidence graph of high semantic-relevance and factual coverage to the question. Then, we adopt DELFT, a graph neural networks based framework that takes the linguistic structure representation of a question and the evidence graph as input, to predict the answer by reasoning over the evidence graph. We evaluate the performance across several baselines in terms of effectiveness and efficiency on two real-world datasets, MOOCQA and MetaQA. The results show the superiority of message passing paradigm in delivering a robust reasoner with better answer quality and significantly improved computational efficiency.

Post-click conversion, as a strong signal indicating the user preference, is salutary for building recommender systems. However, accurately estimating the post-click conversion rate (CVR) is challenging due to the selection bias, i.e., the observed clicked events usually happen on users' preferred items. Currently, most existing methods utilize counterfactual learning to debias recommender systems. Among them, the doubly robust (DR) estimator has achieved competitive performance by combining the error imputation based (EIB) estimator and the inverse propensity score (IPS) estimator in a doubly robust way. However, inaccurate error imputation may result in its higher variance than the IPS estimator. Worse still, existing methods typically use simple model-agnostic methods to estimate the imputation error, which are not sufficient to approximate the dynamically changing model-correlated target (i.e., the gradient direction of the prediction model). To solve these problems, we first derive the bias and variance of the DR estimator. Based on it, a more robust doubly robust (MRDR) estimator has been proposed to further reduce its variance while retaining its double robustness. Moreover, we propose a novel double learning approach for the MRDR estimator, which can convert the error imputation into the general CVR estimation. Besides, we empirically verify that the proposed learning scheme can further eliminate the high variance problem of the imputation learning. To evaluate its effectiveness, extensive experiments are conducted on a semi-synthetic dataset and two real-world datasets. The results demonstrate the superiority of the proposed approach over the state-of-the-art methods. The code is available at https://github.com/guosyjlu/MRDR-DL.

In feeds recommendation, users are able to constantly browse items generated by never-ending feeds using mobile phones. The implicit feedback from users is an important resource for learning to rank, however, building ranking functions from such observed data is recognized to be biased. The presentation of the items will influence the user's judgements and therefore introduces biases. Most previous works in the unbiased learning to rank literature focus on position bias (i.e., an item ranked higher has more chances of being examined and interacted with). By analyzing user behaviors in product feeds recommendation, in this paper, we identify and introduce context bias, which refers to the probability that a user interacting with an item is biased by its surroundings, to unbiased learning to rank. We propose an Unbiased Learning to Rank with Combinational Propensity (ULTR-CP) framework to remove the inherent biases jointly caused by multiple factors. Under this framework, a context-aware position bias model is instantiated to estimate the unified bias considering both position and context biases. In addition to evaluating propensity score estimation approaches by the ranking metrics, we also discuss the evaluation of the propensities directly by checking their balancing properties. Extensive experiments performed on a real e-commerce data set collected from JD.com verify the effectiveness of context bias and illustrate the superiority of ULTR-CP against the state-of-the-art methods.

Human-curated knowledge graphs provide critical supportive information to various natural language processing tasks, but these graphs are usually incomplete, urging auto-completion of them (a.k.a. knowledge graph completion). Prevalent graph embedding approaches, e.g., TransE, learn structured knowledge via representing graph elements (i.e., entities/relations) into dense embeddings and capturing their triple-level relationship with spatial distance. However, they are hardly generalizable to the elements never visited in training and are intrinsically vulnerable to graph incompleteness. In contrast, textual encoding approaches, e.g., KG-BERT, resort to graph triple’s text and triple-level contextualized representations. They are generalizable enough and robust to the incompleteness, especially when coupled with pre-trained encoders. But two major drawbacks limit the performance: (1) high overheads due to the costly scoring of all possible triples in inference, and (2) a lack of structured knowledge in the textual encoder. In this paper, we follow the textual encoding paradigm and aim to alleviate its drawbacks by augmenting it with graph embedding techniques – a complementary hybrid of both paradigms. Specifically, we partition each triple into two asymmetric parts as in translation-based graph embedding approach, and encode both parts into contextualized representations by a Siamese-style textual encoder. Built upon the representations, our model employs both deterministic classifier and spatial measurement for representation and structure learning respectively. It thus reduces the overheads by reusing graph elements’ embeddings to avoid combinatorial explosion, and enhances structured knowledge by exploring the spatial characteristics. Moreover, we develop a self-adaptive ensemble scheme to further improve the performance by incorporating triple scores from an existing graph embedding model. In experiments, we achieve state-of-the-art performance on three benchmarks and a zero-shot dataset for link prediction, with highlights of inference costs reduced by 1-2 orders of magnitude compared to a sophisticated textual encoding method.

We study the problem of incorporating prior knowledge into a deep Transformer-based model, i.e., Bidirectional Encoder Representations from Transformers (BERT), to enhance its performance on semantic textual matching tasks. By probing and analyzing what BERT has already known when solving this task, we obtain better understanding of what task-specific knowledge BERT needs the most and where it is most needed. The analysis further motivates us to take a different approach than most existing works. Instead of using prior knowledge to create a new training task for fine-tuning BERT, we directly inject knowledge into BERT’s multi-head attention mechanism. This leads us to a simple yet effective approach that enjoys fast training stage as it saves the model from training on additional data or tasks other than the main task. Extensive experiments demonstrate that the proposed knowledge-enhanced BERT is able to consistently improve semantic textual matching performance over the original BERT model, and the performance benefit is most salient when training data is scarce.

With the rapid development of social media services in recent years, relational data are explosively growing. The signed network, which consists of a mixture of positive and negative links, is an effective way to represent the friendly and hostile relations among nodes, which can represent users or items. Because the features associated with a node of a signed network are usually incomplete, noisy, unlabeled, and high-dimensional, feature selection is an important procedure to eliminate irrelevant features. However, existing network-based feature selection methods are linear methods, which means they can only select features that having the linear dependency on the output values. Moreover, in many social data, most nodes are unlabeled; therefore, selecting features in an unsupervised manner is generally preferred. To this end, in this paper, we propose a nonlinear unsupervised feature selection method for signed networks, called SignedLasso. This method can select a small number of important features with nonlinear associations between inputs and output from a high-dimensional data. More specifically, we formulate unsupervised feature selection as a nonlinear feature selection problem with the Hilbert-Schmidt Independence Criterion Lasso (HSIC Lasso), which can find a small number of features in a nonlinear manner. Then, we propose the use of a deep learning-based node embedding to represent node similarity without label information and incorporate the node embedding into the HSIC Lasso. Through experiments on two real world datasets, we show that the proposed algorithm is superior to existing linear unsupervised feature selection methods.

Learning the low-dimensional representations of graphs (i.e., network embedding) plays a critical role in network analysis and facilitates many downstream tasks. Recently graph convolutional networks (GCNs) have revolutionized the field of network embedding, and led to state-of-the-art performance in network analysis tasks such as link prediction and node classification. Nevertheless, most of the existing GCN-based network embedding methods are proposed for unsigned networks. However, in the real world, some of the networks are signed, where the links are annotated with different polarities, e.g., positive vs. negative. Since negative links may have different properties from the positive ones and can also significantly affect the quality of network embedding. Thus in this paper, we propose a novel network embedding framework SNEA to learn Signed Network Embedding via graph Attention. In particular, we propose a masked self-attentional layer, which leverages self-attention mechanism to estimate the importance coefficient for pair of nodes connected by different type of links during the embedding aggregation process. Then SNEA utilizes the masked self-attentional layers to aggregate more important information from neighboring nodes to generate the node embeddings based on balance theory. Experimental results demonstrate the effectiveness of the proposed framework through signed link prediction task on several real-world signed network datasets.

Extracting relational triples from unstructured text is crucial for large-scale knowledge graph construction.However, few existing works excel in solving the overlapping triple problem where multiple relational triples in the same sentence share the same entities.In this work, we introduce a fresh perspective to revisit the relational triple extraction task and propose a novel cascade binary tagging framework (CASREL) derived from a principled problem formulation.Instead of treating relations as discrete labels as in previous works, our new framework models relations as functions that map subjects to objects in a sentence, which naturally handles the overlapping problem.Experiments show that the CAS-REL framework already outperforms state-ofthe-art methods even when its encoder module uses a randomly initialized BERT encoder, showing the power of the new tagging framework.It enjoys further performance boost when employing a pre-trained BERT encoder, outperforming the strongest baseline by 17.5 and 30.2 absolute gain in F1-score on two public datasets NYT and WebNLG, respectively.In-depth analysis on different scenarios of overlapping triples shows that the method delivers consistent performance gain across all these scenarios.The source code and data are released online 1 .

Distantly Supervised Relation Extraction (DSRE) has proven to be effective to find relational facts from texts, but it still suffers from two main problems: the wrong labeling problem and the long-tail problem.Most of the existing approaches address these two problems through flat classification, which lacks hierarchical information of relations.To leverage the informative relation hierarchies, we formulate DSRE as a hierarchical classification task and propose a novel hierarchical classification framework, which extracts the relation in a top-down manner.Specifically, in our proposed framework, 1) we use a hierarchically-refined representation method to achieve hierarchy-specific representation; 2) a top-down classification strategy is introduced instead of training a set of local classifiers.The experiments on NYT dataset demonstrate that our approach significantly outperforms other state-of-the-art approaches, especially for the long-tail problem.

Many real-world applications reveal difficulties in learning classifiers from imbalanced data. The rising big data era has been witnessing more classification tasks with large-scale but extremely imbalance and low-quality datasets. Most of existing learning methods suffer from poor performance or low computation efficiency under such a scenario. To tackle this problem, we conduct deep investigations into the nature of class imbalance, which reveals that not only the disproportion between classes, but also other difficulties embedded in the nature of data, especially, noises and class overlapping, prevent us from learning effective classifiers. Taking those factors into consideration, we propose a novel framework for imbalance classification that aims to generate a strong ensemble by self-paced harmonizing data hardness via under-sampling. Extensive experiments have shown that this new framework, while being very computationally efficient, can lead to robust performance even under highly overlapping classes and extremely skewed distribution. Note that, our methods can be easily adapted to most of existing learning methods (e.g., C4.5, SVM, GBDT and Neural Network) to boost their performance on imbalanced data.

The explosive growth of fake news along with destructive effects on politics, economy, and public safety has increased the demand for fake news detection. Fake news on social media does not exist independently in the form of an article. Many other entities, such as news creators, news subjects, and so on, exist on social media and have relationships with news articles. Different entities and relationships can be modeled as a heterogeneous information network (HIN). In this paper, we attempt to solve the fake news detection problem with the support of a news-oriented HIN. We propose a novel fake news detection framework, namely Adversarial Active Learning-based Heterogeneous Graph Neural Network (AA-HGNN) which employs a novel hierarchical attention mechanism to perform node representation learning in the HIN. AA-HGNN utilizes an active learning framework to enhance learning performance, especially when facing the paucity of labeled data. An adversarial selector will be trained to query high-value candidates for the active learning framework. When the adversarial active learning is completed, AA-HGNN detects fake news by classifying news article nodes. Experiments with two real-world fake news datasets show that our model can outperform text-based models and other graph-based models when using less labeled data benefiting from the adversarial active learning. As a model with generalizability, AA-HGNN also has the ability to be widely used in other node classification-related applications on heterogeneous graphs.

Named entity recognition (NER) is a fundamental task in the natural language processing (NLP) area. Recently, representation learning methods (e.g., character embedding and word embedding) have achieved promising recognition results. However, existing models only consider partial features derived from words or characters while failing to integrate semantic and syntactic information (e.g., capitalization, inter-word relations, keywords, lexical phrases, etc.) from multi-level perspectives. Intuitively, multi-level features can be helpful when recognizing named entities from complex sentences. In this study, we propose a novel framework called attention-based multi-level feature fusion (AMFF), which is used to capture the multi-level features from different perspectives to improve NER. Our model consists of four components to respectively capture the local character-level, global character-level, local word-level, and global word-level features, which are then fed into a BiLSTM-CRF network for the final sequence labeling. Extensive experimental results on four benchmark datasets show that our proposed model outperforms a set of state-of-the-art baselines.

Imbalanced learning (IL), i.e., learning unbiased models from class-imbalanced data, is a challenging problem. Typical IL methods including resampling and reweighting were designed based on some heuristic assumptions. They often suffer from unstable performance, poor applicability, and high computational cost in complex tasks where their assumptions do not hold. In this paper, we introduce a novel ensemble IL framework named MESA. It adaptively resamples the training set in iterations to get multiple classifiers and forms a cascade ensemble model. MESA directly learns the sampling strategy from data to optimize the final metric beyond following random heuristics. Moreover, unlike prevailing meta-learning-based IL solutions, we decouple the model-training and meta-training in MESA by independently train the meta-sampler over task-agnostic meta-data. This makes MESA generally applicable to most of the existing learning models and the meta-sampler can be efficiently applied to new tasks. Extensive experiments on both synthetic and real-world tasks demonstrate the effectiveness, robustness, and transferability of MESA. Our code is available at https://github.com/ZhiningLiu1998/mesa.

Cancer staging, grading and subtyping all represent important problems for precision diagnosis, treatment and mechanistic studies of cancer. The majority of the existing computational methods solve this problem via multi-classification of differential gene-expressions of cancer samples of specific classes (Stages, Grades and subtypes) vs. controls. However, the performance of such classification techniques is generally not satisfactory since the discerning power of differential expression patterns in such classifications is limited. We present here a multi-classification technique, based on co-expression patterns specific to individual subclasses in provided training data as co-expression patterns tend to be more conserved than differential expressions within each subclass. A challenge in implementing this strategy lies in how to effectively derive co-expression patterns in individual samples, which is solved through comparing co-expression patterns within a subclass and those in the subclass plus a new sample. Compared with the state-of-the-art gene expression-based classification methods, our method outperforms them in cancer staging, grading and subtyping of cancer samples from TCGA in almost all the measures used. In addition, the co-expressed genes computationally selected for classifications are biologically meaningful, which will prove important for diagnostic biomarker design, treatment plan selection and possibly mechanistic studies of cancer.

A large portion of today's big data can be represented as networks. However, not all networks are the same, and in fact, for many that have additional complexities to their structure, traditional general network analysis methods are no longer applicable. For example, signed networks contain both positive and negative links, and thus dedicated theories and algorithms have been developed. However, previous work mainly focuses on the unipartite setting where signed links connect any pair of nodes. Signed bipartite networks on the one hand, are commonly found, but have primarily been overlooked. Their complexities of having two node types where signed links can only form across the two sets introduce challenges that prevent most existing literature on unipartite signed and unsigned bipartite networks from being applied. On the other hand, balance theory, a key signed social theory, has been generally defined for cycles of any length and is being used in the form of triangles for numerous unipartite signed network tasks. However, in bipartite networks there are no triangles and furthermore there exist two types of nodes. Therefore, in this work, we conduct the first comprehensive analysis and validation of balance theory using the smallest cycle in signed bipartite networks - signed butterflies (i.e., cycles of length 4 containing the two node types). Then, to investigate the applicability of balance theory aiding signed bipartite network tasks, we develop multiple sign prediction methods that utilize balance theory in the form of signed butterflies. Our sign prediction experiment on three real-world signed bipartite networks demonstrates the effectiveness of using these signed butterflies for not only sign prediction, but paves the way for improvements in other signed bipartite network analysis tasks.

In real-world question-answering (QA) systems, ill-formed questions, such as wrong words, ill word order and noisy expressions, are common and may prevent the QA systems from understanding and answering them accurately. In order to eliminate the effect of ill-formed questions, we approach the question refinement task and propose a unified model, QREFINE, to refine the ill-formed questions to well-formed question. The basic idea is to learn a Seq2Seq model to generate a new question from the original one. To improve the quality and retrieval performance of the generated questions, we make two major improvements: 1) To better encode the semantics of ill-formed questions, we enrich the representation of questions with character embedding and the recent proposed contextual word embedding such as BERT, besides the traditional context-free word embeddings; 2) To make it capable to generate desired questions, we train the model with deep reinforcement learning techniques that considers an appropriate wording of the generation as an immediate reward and the correlation between generated question and answer as time-delayed long-term rewards. Experimental results on real-world datasets show that the proposed QREFINE method can generate refined questions with more readability but fewer mistakes than the original questions provided by users. Moreover, the refined questions also significantly improve the accuracy of answer retrieval.

Network embedding is an effective approach to learn the low-dimensional representations of vertices in networks, aiming to capture and preserve the structure and inherent properties of networks. The vast majority of existing network embedding methods exclusively focus on vertex proximity of networks, while ignoring the network internal community structure. However, the homophily principle indicates that vertices within the same community are more similar to each other than those from different communities, thus vertices within the same community should have similar vertex representations. Motivated by this, we propose a novel network embedding framework NECS to learn the Network Embedding with Community Structural information, which preserves the high-order proximity and incorporates the community structure in vertex representation learning. We formulate the problem into a principled optimization framework and provide an effective alternating algorithm to solve it. Extensive experimental results on several benchmark network datasets demonstrate the effectiveness of the proposed framework in various network analysis tasks including network reconstruction, link prediction and vertex classification.

Ad-hoc retrieval models with implicit feedback often have problems, e.g., the imbalanced classes in the data set. Too few clicked documents may hurt generalization ability of the models, whereas too many non-clicked documents may harm effectiveness of the models and efficiency of training. In addition, recent neural network-based models are vulnerable to adversarial examples due to the linear nature in them. To solve the problems at the same time, we propose an adversarial sampling and training framework to learn ad-hoc retrieval models with implicit feedback. Our key idea is (i) to augment clicked examples by adversarial training for better generalization and (ii) to obtain very informational non-clicked examples by adversarial sampling and training. Experiments are performed on benchmark data sets for common ad-hoc retrieval tasks such as Web search, item recommendation, and question answering. Experimental results indicate that the proposed approaches significantly outperform strong baselines especially for high-ranked documents, and they outperform IRGAN in [email protected] using only 5% of labeled data for the Web search task.

Aspect sentiment classification (ASC) is a fundamental task in sentiment analysis.Given an aspect/target and a sentence, the task classifies the sentiment polarity expressed on the target in the sentence.Memory networks (MNs) have been used for this task recently and have achieved state-of-the-art results.In MNs, attention mechanism plays a crucial role in detecting the sentiment context for the given target.However, we found an important problem with the current MNs in performing the ASC task.Simply improving the attention mechanism will not solve it.The problem is referred to as target-sensitive sentiment, which means that the sentiment polarity of the (detected) context is dependent on the given target and it cannot be inferred from the context alone.To tackle this problem, we propose the targetsensitive memory networks (TMNs).Several alternative techniques are designed for the implementation of TMNs and their effectiveness is experimentally evaluated.

Shashi Narayan, Ronald Cardenas, Nikos Papasarantopoulos, Shay B. Cohen, Mirella Lapata, Jiangsheng Yu, Yi Chang. Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2018.

Mobile Applications (or Apps) are becoming more and more popular in recent years, which has attracted increasing attention on mobile App recommendations. The majority of existing App recommendation algorithms focus on mining App functionality or user usage data for discovering user preferences; while actions taken by a user when he/she decides to download an App or not are ignored. In realistic scenarios, a user will first view the description of the App and then decide if he/she wants to download it or not. The actions such as viewing or downloading provide rich information about users' preferences and tastes for Apps, which have great potentials to advance App recommendations. However, the work on exploring action data for App recommendations is rather limited. Therefore, in this paper we study the novel problem of exploiting user actions for App recommendations. We propose a new framework ActionRank, which simultaneously captures various signals from user actions for App recommendations. Experimental results on real-world datasets demonstrate the effectiveness of the proposed framework.

Previous work has shown that popular trending events are important external factors which pose significant influence on user search behavior and also provided a way to computationally model this influence. However, their problem formulation was based on the strong assumption that each event poses its influence independently. This assumption is unrealistic as there are many correlated events in the real world which influence each other and thus, would pose a joint influence on the user search behavior rather than posing influence independently. In this paper, we study this novel problem of Modeling the Joint Influences posed by multiple correlated events on user search behavior. We propose a Joint Influence Model based on the Multivariate Hawkes Process which captures the inter-dependency among multiple events in terms of their influence upon user search behavior. We evaluate the proposed Joint Influence Model using two months query-log data from https://search.yahoo.com/. Experimental results show that the model can indeed capture the temporal dynamics of the joint influence over time and also achieves superior performance over different baseline methods when applied to solve various interesting prediction problems as well as real-word application scenarios, e.g., query auto-completion.

User intent detection plays a critical role in question-answering and dialog systems.Most previous works treat intent detection as a classification problem where utterances are labeled with predefined intents.However, it is labor-intensive and time-consuming to label users' utterances as intents are diversely expressed and novel intents will continually be involved.Instead, we study the zero-shot intent detection problem, which aims to detect emerging user intents where no labeled utterances are currently available.We propose two capsule-based architectures: INTENT-CAPSNET that extracts semantic features from utterances and aggregates them to discriminate existing intents, and INTENTCAPSNET-ZSL which gives INTENTCAPSNET the zero-shot learning ability to discriminate emerging intents via knowledge transfer from existing intents.Experiments on two real-world datasets show that our model not only can better discriminate diversely expressed existing intents, but is also able to discriminate emerging intents when no labeled utterances are available.

Fuad Issa, Marco Damonte, Shay B. Cohen, Xiaohui Yan, Yi Chang. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.

Reciprocity in directed networks points to user»s willingness to return favors in building mutual interactions. High reciprocity has been widely observed in many directed social media networks such as following relations in Twitter and Tumblr. Therefore, reciprocal relations between users are often regarded as a basic mechanism to create stable social ties and play a crucial role in the formation and evolution of networks. Each reciprocity relation is formed by two parasocial links in a back-and-forth manner with a time delay. Hence, understanding the delay can help us gain better insights into the underlying mechanisms of network dynamics. Meanwhile, the accurate prediction of delay has practical implications in advancing a variety of real-world applications such as friend recommendation and marketing campaign. For example, by knowing when will users follow back, service providers can focus on the users with a potential long reciprocal delay for effective targeted marketing. This paper presents the initial investigation of the time delay in reciprocal relations. Our study is based on a large-scale directed network from Tumblr that consists of 62.8 million users and 3.1 billion user following relations with a timespan of multiple years (from 31 Oct 2007 to 24 Jul 2013). We reveal a number of interesting patterns about the delay that motivate the development of a principled learning model to predict the delay in reciprocal relations. Experimental results on the above mentioned dynamic networks corroborate the effectiveness of the proposed delay prediction model.

Ad-hoc entity search, which is to retrieve a ranked list of relevant entities in response to a query of natural language question, has been widely studied. It has been shown that category matching of entities, especially when matching to fine-grained entity types/categories, is critical to the performance of entity search. However, the potentials of the fine-grained Wikipedia entity categories, has not been well exploited by existing studies. Based on the observation of how people describe entities of a specific type, we propose a headword-and-modifier model to deeply interpret both queries and fine-grained entity types/categories. Probabilistic generative models are designed to effectively estimate the relevance of headwords and modifiers as a pattern-based matching problem, taking the Wikipedia type taxonomy as an important input to address the ad-hoc representations of concepts/entities in queries. Extensive experimental results on three widely-used test sets: INEX-XER 2009, SemSearch-LS and TREC-Entity, show that our method achieves a significant improvement of the entity search performance over the state-of-the-art methods.

Stripe noise removal is a fundamental problem in remote sensing image processing. Many efforts have been made to resolve this problem. Recently, a state-of-the-art method was proposed from image-decomposition perspective. This method argued that the stripe and clear image can be simultaneously estimated by modeling the directional structure of stripes and the local smoothness of remote sensing images. However, the potential of this method cannot be fully delivered when confronting with dense stripes with high intensity. In this letter, we further consider the nonlocal self-similarity of image patches in the spatiospectral volume in terms of nonlocal total variation and propose a method of better robustness to dense stripes. Experimental results on both synthetic and real multispectral data show that the proposed method outperforms other competing methods in the remote sensing image destriping task.

Machine learning methods are used to discover complex nonlinear relationships in biological and medical data. However, sophisticated learning models are computationally unfeasible for data with millions of features. Here, we introduce the first feature selection method for nonlinear learning problems that can scale up to large, ultra-high dimensional biological data. More specifically, we scale up the novel Hilbert-Schmidt Independence Criterion Lasso (HSIC Lasso) to handle millions of features with tens of thousand samples. The proposed method is guaranteed to find an optimal subset of maximally predictive features with minimal redundancy, yielding higher predictive power and improved interpretability. Its effectiveness is demonstrated through applications to classify phenotypes based on module expression in human prostate cancer patients and to detect enzymes among protein structures. We achieve high accuracy with as few as 20 out of one million features-a dimensionality reduction of 99.998 percent. Our algorithm can be implemented on commodity cloud computing platforms. The dramatic reduction of features may lead to the ubiquitous deployment of sophisticated prediction models in mobile health care applications.

Modern search engines aggregate results from different verticals : webpages, news, images, video, shopping, knowledge cards, local maps, and so on. Unlike “ten blue links,” these search results are heterogeneous in nature and not even arranged in a list on the page. This revolution directly challenges the conventional “ranked list” formulation in ad hoc search. Therefore, finding proper presentation for a gallery of heterogeneous results is critical for modern search engines. We propose a novel framework that learns the optimal page presentation to render heterogeneous results onto search result page (SERP). Page presentation is broadly defined as the strategy to present a set of items on SERP, much more expressive than a ranked list. It can specify item positions, image sizes, text fonts, and any other styles as long as variations are within business and design constraints. The learned presentation is content aware, i.e., tailored to specific queries and returned results. Simulation experiments show that the framework automatically learns eye-catchy presentations for relevant results. Experiments on real data show that simple instantiations of the framework already outperform leading algorithm in federated search result presentation. It means the framework can learn its own result presentation strategy purely from data, without even knowing the “probability ranking principle.”

Multi-Armed Bandit (MAB) framework has been successfully applied in many web applications. However, many complex real-world applications that involve multiple content recommendations cannot fit into the traditional MAB setting. To address this issue, we consider an ordered combinatorial semi-bandit problem where the learner recommends S actions from a base set of K actions, and displays the results in S (out of M) different positions. The aim is to maximize the cumulative reward with respect to the best possible subset and positions in hindsight. By the adaptation of a minimum-cost maximum-flow network, a practical algorithm based on Thompson sampling is derived for the (contextual) combinatorial problem, thus resolving the problem of computational intractability.With its potential to work with whole-page recommendation and any probabilistic models, to illustrate the effectiveness of our method, we focus on Gaussian process optimization and a contextual setting where click-through rate is predicted using logistic regression. We demonstrate the algorithms’ performance on synthetic Gaussian process problems and on large-scale news article recommendation datasets from Yahoo! Front Page Today Module.

We propose a new random pruning method (called “submodular sparsification (SS)”) to reduce the cost of submodular maximization. The pruning is applied via a “submodularity graph” over the $n$ ground elements, where each directed edge is associated with a pairwise dependency defined by the submodular function. In each step, SS prunes a $1-1/\sqrt{c}$ (for $c>1$) fraction of the nodes using weights on edges computed based on only a small number ($O(\log n)$) of randomly sampled nodes. The algorithm requires $\log_\sqrt{c}n$ steps with a small and highly parallelizable per-step computation. An accuracy-speed tradeoff parameter $c$, set as $c = 8$, leads to a fast shrink rate $\sqrt{2}/4$ and small iteration complexity $\log_{2\sqrt{2}}n$. Analysis shows that w.h.p., the greedy algorithm on the pruned set of size $O(\log^2 n)$ can achieve a guarantee similar to that of processing the original dataset. In news and video summarization tasks, SS is able to substantially reduce both computational costs and memory usage, while maintaining (or even slightly exceeding) the quality of the original (and much more costly) greedy algorithm.

Knowledge graphs (KGs) have been widely used to represent relationships among entities, while KGs cannot capture new relationships between entities emerging along time. Since news often provides the latest information regarding the new entities and relationships, there is an opportunity to connect emerging relationships from news timely. However, it is a challenging task due to the source heterogeneity of structured KGs and unstructured news texts. In order to address the issue, we propose a tensor-based framework to capture the complex interactions among multiple types of relations, entities and text descriptions. We further develop an efficient Text-Aware MUlti-RElational learning method (TAMURE) that can learn the embedding representations of entities and relation types from both KGs and news, by jointly factorizing the interaction parameters. Furthermore, the complexity of TAMURE is linear in the number of parameters, which makes it suitable to large-scale KGs and news texts. Extensive experiments via TensorFlow demonstrate the effectiveness of the proposed TAMURE model compared with nine state-of-the-art methods on real-world datasets.

Network embedding leverages the node proximity manifested to learn a low-dimensional node vector representation for each node in the network. The learned embeddings could advance various learning tasks such as node classification, network clustering, and link prediction. Most, if not all, of the existing works, are overwhelmingly performed in the context of plain and static networks. Nonetheless, in reality, network structure often evolves over time with addition/deletion of links and nodes. Also, a vast majority of real-world networks are associated with a rich set of node attributes, and their attribute values are also naturally changing, with the emerging of new content patterns and the fading of old content patterns. These changing characteristics motivate us to seek an effective embedding representation to capture network and attribute evolving patterns, which is of fundamental importance for learning in a dynamic environment. To our best knowledge, we are the first to tackle this problem with the following two challenges: (1) the inherently correlated network and node attributes could be noisy and incomplete, it necessitates a robust consensus representation to capture their individual properties and correlations; (2) the embedding learning needs to be performed in an online fashion to adapt to the changes accordingly. In this paper, we tackle this problem by proposing a novel dynamic attributed network embedding framework - DANE. In particular, DANE first provides an offline method for a consensus embedding and then leverages matrix perturbation theory to maintain the freshness of the end embedding results in an online manner. We perform extensive experiments on both synthetic and real attributed networks to corroborate the effectiveness and efficiency of the proposed framework.

Users' addiction to online social networks is discovered to be highly correlated with their social connections in the networks. Dense social connections can effectively help online social networks retain their active users and improve the social network services. Therefore, it is of great importance to make a good prediction of the social links among users. Meanwhile, to enjoy more social network services, users nowadays are usually involved in multiple online social networks simultaneously. Formally, the social networks which share a number of common users are defined as the "aligned networks". With the information transferred from multiple aligned social networks, we can gain a more comprehensive knowledge about the social preferences of users in the pre-specified target network, which will benefit the social link prediction task greatly. However, when transferring the knowledge from other aligned source networks to the target network, there usually exists a shift in information distribution between different networks, namely domain difference. In this paper, we study the social link prediction problem of the target network, which is aligned with multiple social networks concurrently. To accommodate the domain difference issue, we project the features extracted for links from different aligned networks into a shared lower-dimensional feature space. Moreover, users in social networks usually tend to form communities and would only connect to a small number of users. Thus, the target network structure has both the low-rank and sparse properties. We propose a novel optimization framework, SLAMPRED, to combine both these two properties aforementioned of the target network and the information of multiple aligned networks with nice domain adaptations. Since the objective function is a linear combination of convex and concave functions involving nondifferentiable regularizers, we propose a novel optimization method to iteratively solve it. Extensive experiments have been done on real-world aligned social networks, and the experimental results demonstrate the effectiveness of the proposed model.

Skipjack tuna (Katsuwonus pelamis) is the major target species in Taiwan pelagic purse-seine in the western and central Pacific Ocean with approximate 90% of catch rate in all harvested species. This study attempts to identify the habitat suitability index (HSI) for skipjack tuna in the western and central Pacific Ocean. Fishery data was processed for comparing with satellite data included sea surface temperature (SST), chlorophyll-a concentration (Chl-a), sea surface height (SSH), and sea surface salinity (SSS). In addition, SST and Chl-a satellite images were applied to oceanic frontal band detection that front gradient magnitudes were used for the HSI estimation. Preliminary results in this study revealed that HSI of skipjack tuna greater than 0.6 was confirmed, where SST and Chl-a front gradient ranged between 0.065 to 0.216 (°C/10km) and 0.003 to 0.005 (ratio/10km), respectively and SSH changed from -0.107 to 0.016 (m), SSS changed from 34.5 to 35.19(psu).

Derelict fishing gear produced by oyster farming activities is being dispersed along the southwestern coast of Taiwan. The derelict gear fragments littering at the coastal areas and is responsible for ghost fishing being a type of marine pollution. To address this problem, the paper presents an experimental implementation of a method to localize radio-frequency identification tags in oyster farms for monitoring using drone technologies. The advantage of the proposed system is to provide a useful tool in analyzing identification of oyster rafts farming with less man power and time. The findings provide an initial feasibility study on combining radio-frequency identification system with drone technology to improve oyster farms management. In addition, this paper depicts the proposed system, shows the testing methodology and analyses some achieved performances in an experimental scenario.

We introduce the convex factorization machine (CFM), which is a convex variant of the widely used Factorization Machines (FMs). Specifically, we employ a linear+quadratic model and regularize the linear term with the ℓ2-regularizer and the quadratic term with the trace norm regularizer. Then, we formulate the CFM optimization as a semidefinite programming problem and propose an efficient optimization procedure with Hazan's algorithm. A key advantage of CFM over existing FMs is that it can find a globally optimal solution, while FMs may get a poor locally optimal solution since the objective function of FMs is non-convex. In addition, the proposed algorithm is simple yet effective and can be implemented easily. Finally, CFM is a general factorization method and can also be used for other factorization problems, including multi-view matrix factorization and tensor completion problems, in various domains including toxicogenomics and bioinformatics. Through synthetic and traditionally used movielens datasets, we first show that the proposed CFM achieves results competitive to FMs. We then show in a toxicogenomics prediction task that CFM predicts the toxic outcomes of a collection of drugs better than a state-of-the-art tensor factorization method.

Network embedding is to learn low-dimensional vector representations for nodes of a given social network, facilitating many tasks in social network analysis such as link prediction. The vast majority of existing embedding algorithms are designed for unsigned social networks or social networks with only positive links. However, networks in social media could have both positive and negative links, and little work exists for signed social networks. From recent findings of signed network analysis, it is evident that negative links have distinct properties and added value besides positive links, which brings about both challenges and opportunities for signed network embedding. In this paper, we propose a deep learning framework SiNE for signed network embedding. The framework optimizes an objective function guided by social theories that provide a fundamental understanding of signed social networks. Experimental results on two real-world datasets of social media demonstrate the effectiveness of the proposed framework SiNE.

Similarity join, which can find similar objects (e.g., products, names, addresses) across different sources, is powerful in dealing with variety in big data, especially web data. Threshold-driven similarity join, which has been extensively studied in the past, assumes that a user is able to specify a similarity threshold, and then focuses on how to efficiently return the object pairs whose similarities pass the threshold. We argue that the assumption about a well set similarity threshold may not be valid for two reasons. The optimal thresholds for different similarity join tasks may vary a lot. Moreover, the end-to-end time spent on similarity join is likely to be dominated by a back-and-forth threshold-tuning process.

Query auto completion (QAC) is a prominent feature in modern search engines. High quality QAC substantially improves search experiences by helping users in typing less while submitting the queries. Many studies have been proposed to improve quality and relevance of the QAC methods from different perspectives, including leveraging contexts in long term and short term query histories, investigating the temporal information for time-sensitive QAC, and analyzing user behaviors. Although these studies have shown the context, temporal, and user behavior data carry valuable information, most existing QAC approaches do not fully exploit or even completely ignore these information. We propose a novel Hawkes process based QAC algorithm, comprehensively taking into account the context, temporal, and position of the clicked recommended query completions (a type of user behavior data), for reliable query completion prediction. Our understanding of ranking query completions is consistent with the mathematical rationale of Hawke process; such a coincidence in turn validates our motivation of using Hawkes process for QAC. We also develop an efficient inference algorithm to compute the optimal solutions of the proposed QAC algorithm. The proposed method is evaluated on two real-world benchmark data in comparison with state-of-art methods, and the obtained experiments clearly demonstrate their effectiveness.

Inferring the links among entities in networks is an important research problem for various disciplines. Depending on the specific application settings, the links to be inferred are usually subject to different cardinality constraints, like one-to-one, one-to-many and many-to-many. However, most existing research works on link prediction problems fail to consider such a kind of constraint. In this paper, we propose to study the link prediction problem with general cardinality constraints, which is formally defined as the CLP (Cardinality Constrained Link Prediction) problem. By minimizing the projection loss of links from feature vectors to labels, the CLP problem is formulated as an optimization problem involving multiple variables, where the cardinality constraints are modeled as mathematical constraints on node degrees. The objective function is shown to be not jointly convex and the optimal solution subject to the cardinality constraints can be very time-consuming to achieve. To solve the optimization problem, an iterative variable updating based link prediction framework ITERCLIPS (Iterative Constrained Link Prediction & Selection) is introduced in this paper, which involves the steps on link updating and selection alternatively. To overcome the high time cost problem, a greedy link selection step is introduced in this paper, which picks links greedily while preserving the link cardinality constraints simultaneously. Meanwhile, to ensure the effectiveness of ITERCLIPS on large-scale networks, a distributed implementation of ITERCLIPS is further presented as a scalable solution to the CLP problem. Extensive experiments have been done on three real-world network datasets with different types of cardinality constraints, and the experimental results achieved by ITERCLIPS on all these datasets can demonstrate the effectiveness and advantages of ITERCLIPS in solving the CLP problem.

The increasing popularity of real-world recommender systems produces data continuously and rapidly, and it becomes more realistic to study recommender systems under streaming scenarios. Data streams present distinct properties such as temporally ordered, continuous and high-velocity, which poses tremendous challenges to traditional recommender systems. In this paper, we investigate the problem of recommendation with stream inputs. In particular, we provide a principled framework termed sRec, which provides explicit continuous-time random process models of the creation of users and topics, and of the evolution of their interests. A variational Bayesian approach called recursive meanfield approximation is proposed, which permits computationally efficient instantaneous on-line inference. Experimental results on several real-world datasets demonstrate the advantages of our sRec over other state-of-the-arts.

Contextual data plays an important role in modeling search engine users' behaviors on both query auto-completion (QAC) log and normal query (click) log. User's recent search history on each log has been widely studied individually as the context to benefit the modeling of users' behaviors on that log. However, there is no existing work that explores or incorporates both logs together for contextual data. As QAC and click logs actually record users' sequential behaviors while interacting with a search engine, the available context of a user's current behavior based on the same type of log can be strengthened from the user's recent search history shown on the other type of log. Our paper proposes to model users' behaviors on both QAC and click logs simultaneously by utilizing both logs as the contextual data of each other. The key idea is to capture the correlation between users' behavior patterns on both logs. We model such correlation through a novel probabilistic model based on the Latent Dirichlet allocation (LDA) model. The learned users' behavior patterns on both logs are utilized to address not only the application of query auto-completion on QAC logs, but also the click prediction and relevance ranking of web documents on click logs. Experiments on real-world logs demonstrate the effectiveness of the proposed model on both applications.

Studies suggest that self-harm users found it easier to discuss self-harm-related thoughts and behaviors using social media than in the physical world. Given the enormous and increasing volume of social media data, on-line self-harm content is likely to be buried rapidly by other normal content. To enable voices of self-harm users to be heard, it is important to distinguish self-harm content from other types of content. In this paper, we aim to understand self-harm content and provide automatic approaches to its detection. We first perform a comprehensive analysis on self-harm social media using different input cues. Our analysis, the first of its kind in large scale, reveals a number of important findings. Then we propose frameworks that incorporate the findings to discover self-harm content under both supervised and unsupervised settings. Our experimental results on a large social media dataset from Flickr demonstrate the effectiveness of the proposed frameworks and the importance of our findings in discovering self-harm content.

Location information of Web pages plays an important role in location-sensitive tasks such as Web search ranking for location-sensitive queries. However, such information is usually ambiguous, incomplete, or even missing, which raises the problem of location prediction for Web pages. Meanwhile, Web pages are massive and often noisy, which pose challenges to the majority of existing algorithms for location prediction. In this paper, we propose a novel and scalable location prediction framework for Web pages based on the query-URL click graph. In particular, we introduce a concept of term location vectors to capture location distributions for all terms and develop an automatic approach to learn the importance of each term location vector for location prediction. Empirical results on a large URL set demonstrate that the proposed framework significantly improves the location prediction accuracy comparing with various representative baselines. We further provide a principled way to incorporate the proposed framework into the search ranking task and experimental results on a commercial search engine show that the proposed method remarkably boosts the ranking performance for location-sensitive queries.

Query auto-completion (QAC) is widely used by modern search engines to assist users by predicting their intended queries. Most QAC approaches rely on deterministic batch learning algorithms trained from past query log data. However, query popularities keep changing all the time and QAC operates in a real-time scenario where users interact with the search engine continually. So, ideally, QAC must be timely and adaptive enough to reflect time-sensitive changes in an online fashion. Second, due to the vertical position bias, a query suggestion with a higher rank tends to attract more clicks regardless of user's original intention. Hence, in the long run, it is important to place some lower ranked yet potentially more relevant queries to higher positions to collect more valuable user feedbacks. In order to tackle these issues, we propose to formulate QAC as a ranked Multi-Armed Bandits (MAB) problem which enjoys theoretical soundness. To utilize prior knowledge from query logs, we propose to use Bayesian inference and Thompson Sampling to solve this MAB problem. Extensive experiments on large scale datasets show that our QAC algorithm has the capacity to adaptively learn temporal trends, and outperforms existing QAC algorithms in ranking qualities.

In recent years, mobile devices have become the most popular interface for users to retrieve and access information: recent reports show that users spend significantly more time and issue more search queries on mobile devices than on desktops in the United States. 1 The accelerated growth of mobile usage brings unique opportunities to the information retrieval and data mining research communities. Mobile devices capture rich contextual and personal signals that can be leveraged to accurately predict users’ intent for serving more relevant content and can even proactively provide novel zero-query recommendations. Apple Siri, Google Now, and Microsoft Cortana are recent examples of such emerging systems. Furthermore, mobile devices constantly generate a huge amount of sensor footprints (e.g., GPS, motion sensors) and user activity data (e.g., used apps) that are often missing from their desktop counterparts. These new sources of implicit and explicit user feedback are valuable for discovering actionable knowledge, and designing better systems that serve each individual the right content at the right time and location. In addition, by aggregating mobile interactions across individuals, one can infer interesting conclusions beyond search and recommendation. Generating real-time traffic estimates is one example of such applications. This special issue focuses on research problems of search, mining, and their applications in mobile devices. Topics of interest in this special issue include but are not limited to mobile data mining and management, mobile search, personalization and recommendation, mobile user interfaces and human-computer interaction, and new applications in the mobile environment. The aim of this special issue is to bring together top experts across multiple disciplines, including information retrieval, data mining, mobile computing, and cyberphysical systems, such that academic and industrial researchers can exchange ideas and share the latest developments on the state of the art and practice of mobile search and mobile data mining.

The pervasive presence of social media greatly enriches online users' social activities, resulting in abundant social relations. Social relations provide an independent source for recommendation, bringing about new opportunities for recommender systems. Exploiting social relations to improve recommendation performance attracts a great amount of attention in recent years. Most existing social recommender systems treat social relations homogeneously and make use of direct connections (or strong dependency connections). However, connections in online social networks are intrinsically heterogeneous and are a composite of various relations. While connected users in online social networks form groups, and users in a group share similar interests, weak dependency connections are established among these users when they are not directly connected. In this paper, we investigate how to exploit the heterogeneity of social relations and weak dependency connections for recommendation. In particular, we employ social dimensions to simultaneously capture heterogeneity of social relations and weak dependency connections, and provide principled ways to model social dimensions, and propose a recommendation framework SoDimRec which incorporates heterogeneity of social relations and weak dependency connections based on social dimensions. Experimental results on real-world data sets demonstrate the effectiveness of the proposed framework. We conduct further experiments to understand the important role of social dimensions in the proposed framework.

Real-world knowledge is growing rapidly nowadays. New entities arise with time, resulting in large volumes of relations that do not exist in current knowledge graphs (KGs). These relations containing at least one new entity are called emerging relations. They often appear in news, and hence the latest information about new entities and relations can be learned from news timely. In this paper, we focus on the problem of discovering emerging relations from news. However, there are several challenges for this task: (1) at the beginning, there is little information for emerging relations, causing problems for traditional sentence-based models; (2) no negative relations exist in KGs, creating difficulties in utilizing only positive cases for emerging relation detection from news; and (3) new relations emerge rapidly, making it necessary to keep KGs up to date with the latest emerging relations. In order to address these issues, we start from a global graph perspective and propose a novel Heterogeneous graph Embedding framework for Emerging Relation detection (HEER) that learns a classifier from positive and unlabeled instances by utilizing information from both news and KGs. Furthermore, we implement HEER in an incremental manner to timely update KGs with the latest detected emerging relations. Extensive experiments on real-world news datasets demonstrate the effectiveness of the proposed HEER model.

It is widely known that there exists a semantic gap between web documents and user queries and bridging this gap is crucial to advance information retrieval systems. The task of query rewriting, aiming to alter a given query to a rewrite query that can close the gap and improve information retrieval performance, has attracted increasing attention in recent years. However, the majority of existing query rewriters are not designed to boost search performance and consequently their rewrite queries could be sub-optimal. In this paper, we propose a learning to rewrite framework that consists of a candidate generating phase and a candidate ranking phase. The candidate generating phase provides us the flexibility to reuse most of existing query rewriters; while the candidate ranking phase allows us to explicitly optimize search relevance. Experimental results on a commercial search engine demonstrate the effectiveness of the proposed framework. Further experiments are conducted to understand the important components of the proposed framework.

Crowdsourcing services make it possible to collect huge amount of annotations from less trained crowd workers in an inexpensive and efficient manner. However, unlike making binary or pairwise judgements, labeling complex structures such as ranked lists by crowd workers is subject to large variance and low efficiency, mainly due to the huge labeling space and the annotators' non-expert nature. Yet ranked lists offer the most informative knowledge for training and testing in various data mining and information retrieval tasks such as learning to rank. In this paper, we propose a novel generative model called "Thurstonian Pairwise Preference" (TPP) to infer the true ranked list out of a collection of crowdsourced pairwise annotations. The key challenges that TPP addresses are to resolve the inevitable incompleteness and inconsistency of judgements, as well as to model variable query difficulty and different labeling quality resulting from workers' domain expertise and truthfulness. Experimental results on both synthetic and real-world datasets demonstrate that TPP can effectively bind pairwise preferences of the crowd into rankings and substantially outperforms previously published methods.

Data of many problems in real-world systems such as link prediction and one-class recommendation share common characteristics. First, data are in the form of positive unlabeled (PU) measurements (e.g. Twitter "following", Facebook "like", etc.) that do not provide negative information, which can be naturally represented as networks. Second, in the era of big data, such data are generated temporally-ordered, continuously and rapidly, which determines its streaming nature. These common characteristics allow us to unify many problems into a novel framework -- PU learning in streaming networks. In this paper, a principled probabilistic approach SPU is proposed to leverage the characteristics of the streaming PU inputs. In particular, SPU captures temporal dynamics and provides real-time adaptations and predictions by identifying the potential negative signals concealed in unlabeled data. Our empirical results on various real-world datasets demonstrate the effectiveness of the proposed framework over other state-of-the-art methods in both link prediction and recommendation.

Search engines play a crucial role in our daily lives. Relevance is the core problem of a commercial search engine. It has attracted thousands of researchers from both academia and industry and has been studied for decades. Relevance in a modern search engine has gone far beyond text matching, and now involves tremendous challenges. The semantic gap between queries and URLs is the main barrier for improving base relevance. Clicks help provide hints to improve relevance, but unfortunately for most tail queries, the click information is too sparse, noisy, or missing entirely. For comprehensive relevance, the recency and location sensitivity of results is also critical. In this paper, we give an overview of the solutions for relevance in the Yahoo search engine. We introduce three key techniques for base relevance -- ranking functions, semantic matching features and query rewriting. We also describe solutions for recency sensitive relevance and location sensitive relevance. This work builds upon 20 years of existing efforts on Yahoo search, summarizes the most recent advances and provides a series of practical relevance solutions. The performance reported is based on Yahoo's commercial search engine, where tens of billions of urls are indexed and served by the ranking system.

Click-through logs over query-document pairs provide rich and valuable information for multiple tasks in information retrieval. This paper proposes a vector propagation algorithm on the click graph to learn vector representations for both queries and documents in the same semantic space. The proposed approach incorporates both click and content information, and the produced vector representations can directly improve ranking performance for queries and documents that have been observed in the click log. For new queries and documents that are not in the click log, we propose a two-step framework to generate the vector representation, which significantly improves the coverage of our vectors while maintaining the high quality. Experiments on Web-scale search logs from a major commercial search engine demonstrate the effectiveness and scalability of the proposed method. Evaluation results show that NDCG scores are significantly improved against multiple baselines by using the proposed method both as a ranking model and as a feature in a learning-to-rank framework.

User interfaces for web image search engine results differ significantly from interfaces for traditional (text) web search results, supporting a richer interaction. In particular, users can see an enlarged image preview by hovering over a result image, and an `image preview' page allows users to browse further enlarged versions of the results, and to click-through to the referral page where the image is embedded. No existing work investigates the utility of these interactions as implicit relevance feedback for improving search ranking, beyond using clicks on images displayed in the search results page. In this paper we propose a number of implicit relevance feedback features based on these additional interactions: hover-through rate, 'converted-hover' rate, referral page click through, and a number of dwell time features. Also, since images are never self-contained, but always embedded in a referral page, we posit that clicks on other images that are embedded on the same referral webpage as a given image can carry useful relevance information about that image. We also posit that query-independent versions of implicit feedback features, while not expected to capture topical relevance, will carry feedback about the quality or attractiveness of images, an important dimension of relevance for web image search. In an extensive set of ranking experiments in a learning to rank framework, using a large annotated corpus, the proposed features give statistically significant gains of over 2% compared to a state of the art baseline that uses standard click features.

No abstract available.

Modern search engines aggregate results from different verticals: webpages, news, images, video, shopping, knowledge cards, local maps, etc. Unlike "ten blue links", these search results are heterogeneous in nature and not even arranged in a list on the page. This revolution directly challenges the conventional "ranked list" formulation in ad hoc search. Therefore, finding proper presentation for a gallery of heterogeneous results is critical for modern search engines.

Search tasks in users' query sequences are dynamic and interconnected. The formulation of search tasks can be influenced by multiple latent factors such as user characteristics, product features and search interactions, which makes search task identification a challenging problem. In this paper, we propose an unsupervised approach to identify search tasks via topic membership along with topic transition probabilities, thus it becomes possible to interpret how user's search intent emerges and evolves over time. Moreover, a novel hidden semi-Markov model is introduced to model topic transitions by considering not only the semantic information of queries but also the latent search factors originated from user search behaviors. A variational inference algorithm is developed to identify remarkable search behavior patterns, typical topic transition tracks, and the topic membership of each query from query logs. The learned topic transition tracks and the inferred topic memberships enable us to identify both small search tasks, where a user searches the same topic, and big search tasks, where a user searches a series of related topics. We extensively evaluate the proposed approach and compare with several state-of-the-art search task identification methods on both synthetic and real-world query log data, and experimental results illustrate the effectiveness of our proposed model.

Detection of abusive language in user generated online content has become an issue of increasing importance in recent years. Most current commercial methods make use of blacklists and regular expressions, however these measures fall short when contending with more subtle, less ham-fisted examples of hate speech. In this work, we develop a machine learning based method to detect hate speech on online user comments from two domains which outperforms a state-of-the-art deep learning approach. We also develop a corpus of user comments annotated for abusive language, the first of its kind. Finally, we use our detection tool to analyze abusive language over time and in different settings to further enhance our knowledge of this behavior.

We study the new mobile query auto-completion (QAC) problem to exploit mobile devices' exclusive signals, such as those related to mobile applications (apps). We propose AppAware, a novel QAC model using installed app and recently opened app signals to suggest queries for matching input prefixes on mobile devices. To overcome the challenge of noisy and voluminous signals, AppAware optimizes composite objectives with a lighter processing cost at a linear rate of convergence. We conduct experiments on a large commercial data set of mobile queries and apps. Installed app and recently opened app signals consistently and significantly boost the accuracy of various baseline QAC models on mobile devices.

In social media analysis, one critical task is detecting a burst of topics or buzz , which is reflected by extremely frequent mentions of certain keywords in a short-time interval. Detecting buzz not only provides useful insights into the information propagation mechanism, but also plays an essential role in preventing malicious rumors. However, buzz modeling is a challenging task because a buzz time-series often exhibits sudden spikes and heavy tails, wherein most existing time-series models fail. In this article, we propose novel buzz modeling approaches that capture the rise and fade temporal patterns via Product Lifecycle (PLC) model, a classical concept in economics. More specifically, we propose to model multiple peaks in buzz time-series with PLC mixture or PLC group mixture and develop a probabilistic graphical model (K-Mixture of Product Lifecycle ( K-MPLC ) to automatically discover inherent lifecycle patterns within a collection of buzzes. Furthermore, we effectively utilize the model parameters of PLC mixture or PLC group mixture for burst prediction. Our experimental results show that our proposed methods significantly outperform existing leading approaches on buzz clustering and buzz-type prediction.

Advertising effectiveness measurement is a fundamental problem in online advertising. Various causal inference methods have been employed to measure the causal effects of ad treatments. However, existing methods mainly focus on linear logistic regression for univariate and binary treatments and are not well suited for complex ad treatments of multi-dimensions, where each dimension could be discrete or continuous. In this paper we propose a novel two-stage causal inference framework for assessing the impact of complex ad treatments. In the first stage, we estimate the propensity parameter via a sparse additive model; in the second stage, a propensity-adjusted regression model is applied for measuring the treatment effect. Our approach is shown to provide an unbiased estimation of the ad effectiveness under regularity conditions. To demonstrate the efficacy of our approach, we apply it to a real online advertising campaign to evaluate the impact of three ad treatments: ad frequency, ad channel, and ad size. We show that the ad frequency usually has a treatment effect cap when ads are showing on mobile device. In addition, the strategies for choosing best ad size are completely different for mobile ads and online ads.

We address the collective matrix completion problem of jointly recovering a collection of matrices with shared structure from partial (and potentially noisy) observations. To ensure well–posedness of the problem, we impose a joint low rank structure, wherein each component matrix is low rank and the latent space of the low rank factors corresponding to each entity is shared across the entire collection. We first develop a rigorous algebra for representing and manipulating collective–matrix structure, and identify sufficient conditions for consistent estimation of collective matrices. We then propose a tractable convex estimator for solving the collective matrix completion problem, and provide the first non–trivial theoretical guarantees for consistency of collective matrix completion. We show that under reasonable assumptions stated in Sec. 3.1, with high probability, the proposed estimator exactly recovers the true matrices whenever sample complexity requirements dictated by Theorem 1 are met. The sample complexity requirement derived in the paper are optimum up to logarithmic factors, and significantly improve upon the requirements obtained by trivial extensions of standard matrix completion. Finally, we propose a scalable approximate algorithm to solve the proposed convex program, and corroborate our results through simulated and real life experiments.

Entities (e.g., person, movie or place) play an important role in real-world applications and learning entity types has attracted much attention in recent years. Most conventional automatic techniques use large corpora, such as news articles, to learn types of entities. However, such text corpora focus on general knowledge about entities in an objective way. Hence, it is difficult to satisfy those users with specific and personalized needs for an entity. Recent years have witnessed an explosive expansion in the mining of search query logs, which contain billions of entities. The word patterns and click-throughs in search logs are not found in text corpora, thus providing a complemental source for discovering entity types based on user behaviors. In this paper, we study the problem of learning entity types from search query logs and address the following challenges: (1) queries are short texts, and information related to entities is usually very sparse; (2) large amounts of irrelevant information exists in search logs, bringing noise in detecting entity types. In this paper, we first model query logs using a bipartite graph with entities and their auxiliary information, such as contextual words and clicked URLs. Then we propose a graph-based framework called ELP (Ensemble framework based on Lable Propagation) to simultaneously learn the types of both entities and auxiliary signals. In ELP, two separate strategies are designed to fix the problems of sparsity and noise in query logs. Extensive empirical studies are conducted on real search logs to evaluate the effectiveness of the proposed ELP framework.

The explosive popularity of microblogging services encourages more and more online users to share their opinions, and sentiment analysis on such opinion-rich resources has been proven to be an effective way to understand public opinions. On the one hand, the brevity and informality of microblogging data plus its wide variety and rapid evolution of language in microblogging pose new challenges to the vast majority of existing methods. On the other hand, microblogging texts contain various types of emotional signals strongly associated with their sentiment polarity, which brings about new opportunities for sentiment analysis. In this paper, we investigate propagation-based sentiment analysis for microblogging data. In particular, we provide a propagating process to incorporate various types of emotional signals in microblogging data into a coherent model, and propose a novel sentiment analysis framework PSA which learns from both labeled and unlabeled data by iteratively alternating a propagating process and a fitting process. We conduct experiments on real-world microblogging datasets, and the results demonstrate the effectiveness of the proposed framework. Further experiments are conducted to probe the working of the key components of the proposed framework.

Web search relevance is a billion dollar challenge, while there is a disadvantage of backwardness in web search competition. Vertical search result can be incorporated to enrich web search content, therefore vertical search relevance is critical to provide differentiated search results. Machine learning based ranking algorithms have shown their effectiveness for both web search and vertical search tasks. In this talk, the speaker will not only introduce state-of-the-art ranking algorithms for web search, but also cover the challenges to improve relevance of various vertical search engines: local search, shopping search, news search, etc.

While many studies have been conducted on query understanding, there is limited understanding on why users start searches and how to predict search intent. In this paper, we propose to study this important but less explored problem. Our key intuition is that searches are triggered by different pre-search contexts, but the triggering relations are often hidden. For example, a user may search "bitcoin" because of a news article or an email the user just read, but the system does not know which of the pre-search contexts (the news article or the email) is the triggering source. Following this intuition, we conduct an in-depth analysis of pre-search context on a large-scale user log, which not only verifies the hidden triggering relations in the real world but also identifies a set of important characteristics of pre-search context and their triggered queries. Since the hidden triggering relations make it challenging to directly use pre-search context for intent prediction, we develop a mixture generative model to learn without any supervision how queries are triggered by different types of pre-search context. Further, we discuss how to apply our model to improve query prediction and query auto-completion. Our experiments on a large-scale of real-world data show that our model could accurately predict user search intent with pre-search context and improve upon the state-of-the-art methods significantly.

Query auto-completion (QAC) plays an important role in assisting users typing less while submitting a query. The QAC engine generally offers a list of suggested queries that start with a user's input as a prefix, and the list of suggestions is changed to match the updated input after the user types each keystroke. Therefore rich user interactions can be observed along with each keystroke until a user clicks a suggestion or types the entire query manually. It becomes increasingly important to analyze and understand users' interactions with the QAC engine, to improve its performance. Existing works on QAC either ignored users' interaction data, or assumed that their interactions at each keystroke are independent from others. Our paper pays high attention to users' sequential interactions with a QAC engine in and across QAC sessions, rather than users' interactions at each keystroke of each QAC session separately. Analyzing the dependencies in users' sequential interactions improves our understanding of the following three questions: 1) how is a user's skipping/viewing move at the current keystroke influenced by that at the previous keystroke? 2) how to improve search engines' query suggestions at short keystrokes based on those at latter long keystrokes? and 3) facing a targeted query shown in the suggestion list, why does a user decide to continue typing rather than click the intended suggestion? We propose a probabilistic model that addresses those three questions in a unified way, and illustrate how the model determines users' final click decisions. By comparing with state-of-the-art methods, our proposed model does suggest queries that better satisfy users' intents.

Query auto-completion (QAC) facilitates user query composition by suggesting queries given query prefix inputs. In 2014, global users of Yahoo! Search saved more than 50% keystrokes when submitting English queries by selecting suggestions of QAC. Users' preference of queries can be inferred during user-QAC interactions, such as dwelling on suggestion lists for a long time without selecting query suggestions ranked at the top. However, the wealth of such implicit negative feedback has not been exploited for designing QAC models. Most existing QAC models rank suggested queries for given prefixes based on certain relevance scores.

As the online advertising industry has evolved into an age of diverse ad formats and delivery channels, users are exposed to complex ad treatments involving various ad characteristics. The diversity and generality of ad treatments call for accurate and causal measurement of ad effectiveness, i.e., how the ad treatment causes the changes in outcomes without the confounding effect by user characteristics. Various causal inference approaches have been proposed to measure the causal effect of ad treatments. However, most existing causal inference methods focus on univariate and binary treatment and are not well suited for complex ad treatments. Moreover, to be practical in the data-rich online environment, the measurement needs to be highly general and efficient, which is not addressed in conventional causal inference approaches. In this paper we propose a novel causal inference framework for assessing the impact of general advertising treatments. Our new framework enables analysis on uni- or multi-dimensional ad treatments, where each dimension (ad treatment factor) could be discrete or continuous. We prove that our approach is able to provide an unbiased estimation of the ad effectiveness by controlling the confounding effect of user characteristics. The framework is computationally efficient by employing a tree structure that specifies the relationship between user characteristics and the corresponding ad treatment. This tree-based framework is robust to model misspecification and highly flexible with minimal manual tuning. To demonstrate the efficacy of our approach, we apply it to two advertising campaigns. In the first campaign we evaluate the impact of different ad frequencies, and in the second one we consider the synthetic ad effectiveness across TV and online platforms. Our framework successfully provides the causal impact of ads with different frequencies in both campaigns. Moreover, it shows that the ad frequency usually has a treatment effect cap, which is usually over-estimated by naive estimation.

As the web information exponentially grows and the needs of users become more specific, traditional general web search engines are not able to perfectly satisfy the nowadays user requirement. Vertical search engines have emerged in various domains, which more focus on specific segments of online content, including local, shopping, medical information, travel search, etc. Vertical search engines start attracting more attention while relevance ranking in different vertical search engines is becoming the key technology. In addition, vertical search results are often slotted into general Web search results. Hence, designing effective ranking functions for vertical search has become practically important to improve users' experience in both web search and vertical search. The workshop bring together researchers from IR, ML, NLP, and other areas of computer and information science, who are working on or interested in this area. It provides a forum for the researchers to identify the issues and the challenges, to share their latest research results, to express a diverse range of opinions about this topic, and to discuss future directions.

The goal of cross-domain matching (CDM) is to find correspondences between two sets of objects in different domains in an unsupervised way. CDM has various interesting applications, including photo album summarization where photos are automatically aligned into a designed frame expressed in the Cartesian coordinate system, and temporal alignment which aligns sequences such as videos that are potentially expressed using different features. In this paper, we propose an information-theoretic CDM framework based on squared-loss mutual information (SMI). The proposed approach can directly handle non-linearly related objects/sequences with different dimensions, with the ability that hyper-parameters can be objectively optimized by cross-validation. We apply the proposed method to several real-world problems including image matching, unpaired voice conversion, photo album summarization, cross-feature video and cross-domain video-to-mocap alignment, and Kinect-based action recognition, and experimentally demonstrate that the proposed method is a promising alternative to state-of-the-art CDM methods.

Learning to rank arises in many data mining applications, ranging from web search engine, online advertising to recommendation system. In learning to rank, the performance of a ranking model is strongly affected by the number of labeled examples in the training set; on the other hand, obtaining labeled examples for training data is very expensive and time-consuming. This presents a great need for the active learning approaches to select most informative examples for ranking learning; however, in the literature there is still very limited work to address active learning for ranking. In this paper, we propose a general active learning framework, expected loss optimization (ELO), for ranking. The ELO framework is applicable to a wide range of ranking functions. Under this framework, we derive a novel algorithm, expected discounted cumulative gain (DCG) loss optimization (ELO-DCG), to select most informative examples. Then, we investigate both query and document level active learning for raking and propose a two-stage ELO-DCG algorithm which incorporate both query and document selection into active learning. Furthermore, we show that it is flexible for the algorithm to deal with the skewed grade distribution problem with the modification of the loss function. Extensive experiments on real-world web search data sets have demonstrated great potential and effectiveness of the proposed framework and algorithms.

Trust plays a crucial role in helping users collect reliable information in an online world, and has attracted more and more attention in research communities lately. As a conceptual counterpart of trust, distrust can be as important as trust. However, distrust is rarely studied in social media because distrust information is usually unavailable. The value of distrust has been widely recognized in social sciences and recent work shows that distrust can benefit various online applications in social media. In this work, we investigate whether we can obtain distrust information via learning when it is not directly available, and propose to study a novel problem - predicting distrust using pervasively available interaction data in an online world. In particular, we analyze interaction data, provide a principled way to mathematically incorporate interaction data in a novel framework dTrust to predict distrust information. Experimental results using real-world data show that distrust information is predictable with interaction data by the proposed framework dTrust. Further experiments are conducted to gain a deep understand on which factors contribute to the effectiveness of the proposed framework.

Question-and-answer (Q&A) websites, such as Yahoo! Answers, Stack Overflow and Quora, have become a popular and powerful platform for Web users to share knowledge on a wide range of subjects. This has led to a rapidly growing volume of information and the consequent challenge of readily identifying high quality objects (questions, answers and users) in Q&A sites. Exploring the interdependent relationships among different types of objects can help find high quality objects in Q&A sites more accurately. In this paper, we specifically focus on the ranking problem of co-ranking questions, answers and users in a Q&A website. By studying the tightly connected relationships between Q&A objects, we can gain useful insights toward solving the co-ranking problem. However, co-ranking multiple objects in Q&A sites is a challenging task: a) With the large volumes of data in Q&A sites, it is important to design a model that can scale well; b) The large-scale Q&A data makes extracting supervised information very expensive. In order to address these issues, we propose an unsupervised Network-based Co-Ranking framework (NCR) to rank multiple objects in Q&A sites. Empirical studies on real-world Yahoo! Answers datasets demonstrate the effectiveness and the efficiency of the proposed NCR method.

In social media analysis, one critical task is detecting burst of topics or buzz, which is reflected by extremely frequent mentions of certain key words in a short time interval. Detecting buzz not only provides useful insights into the information propagation mechanism, but also plays an essential role in preventing malicious rumors. However, buzz modeling is a challenging task because a buzz time-series usually exhibits sudden spikes and heavy tails, which fails most existing time-series models. To deal with buzz time-series sequences, we propose a novel time-series modeling approach which captures the rise and fade temporal patterns via Product Life Cycle (PLC) models, a classical concept in economics. More specifically, we propose a mixture of PLC models to capture the multiple peaks in buzz time-series and furthermore develop a probabilistic graphical model (K-MPLC) to automatically discover inherent life cycle patterns within a collection of buzzes. Our experiment results show that our proposed method significantly outperforms existing state-of-the-art approaches on buzzes clustering.

We consider a search task as a set of queries that serve the same user information need. Analyzing search tasks from user query streams plays an important role in building a set of modern tools to improve search engine performance. In this paper, we propose a probabilistic method for identifying and labeling search tasks based on the following intuitive observations: queries that are issued temporally close by users in many sequences of queries are likely to belong to the same search task, meanwhile, different users having the same information needs tend to submit topically coherent search queries. To capture the above intuitions, we directly model query temporal patterns using a special class of point processes called Hawkes processes, and combine topic models with Hawkes processes for simultaneously identifying and labeling search tasks. Essentially, Hawkes processes utilize their self-exciting properties to identify search tasks if influence exists among a sequence of queries for individual users, while the topic model exploits query co-occurrence across different users to discover the latent information needed for labeling search tasks. More importantly, there is mutual reinforcement between Hawkes processes and the topic model in the unified model that enhances the performance of both. We evaluate our method based on both synthetic data and real-world query log data. In addition, we also apply our model to query clustering and search task identification. By comparing with state-of-the-art methods, the results demonstrate that the improvement in our proposed approach is consistent and promising.

Query auto-completion (QAC) facilitates faster user query input by predicting users' intended queries. Most QAC algorithms take a learning-based approach to incorporate various signals for query relevance prediction. However, such models are trained on simulat- ed user inputs from query log data. The lack of real user interaction data in the QAC process prevents them from further improving the QAC performance. In this work, for the first time we collect a high-resolution QAC query log that records every keystroke in a QAC session. Based on this data, we discover two user behaviors, namely the horizontal skipping bias and vertical position bias which are crucial for rele- vance prediction in QAC. In order to better explain them, we pro- pose a novel two-dimensional click model for modeling the QAC process with emphasis on these behaviors. Extensive experiments on our QAC data set from both PC and mobile devices demonstrate that our proposed model can accurate- ly explain the users' behaviors in interacting with a QAC system, and the resulting relevance model significant improves the QAC performance over existing click models. Furthermore, the learned knowledge about the skipping behavior can be effectively incorpo- rated into existing learning-based models to further improve their performance.

Searchers' information needs are diverse and cover a broad range of topics; hence, it is important for search engines to accurately understand each individual user's search intents in order to provide optimal search results. Search log data, which records users' search behaviors when interacting with search engines, provides a valuable source of information about users' search intents. Therefore, properly characterizing the heterogeneity among the users' observed search behaviors is the key to accurately understanding their search intents and to further predicting their behaviors.

Emerging trends and products pose a challenge to modern search engines since they must adapt to the constantly changing needs and interests of users. For example, vertical search engines, such as Amazon, eBay, Walmart, Yelp and Yahoo! Local, provide business category hierarchies for people to navigate through millions of business listings. The category information also provides important ranking features that can be used to improve search experience. However, category hierarchies are often manually crafted by some human experts and they are far from complete. Manually constructed category hierarchies cannot handle the ever-changing and sometimes long-tail user information needs. In this paper, we study the problem of how to expand an existing category hierarchy for a search/navigation system to accommodate the information needs of users more comprehensively. We propose a general framework for this task, which has three steps: 1) detecting meaningful missing categories; 2) modeling the category hierarchy using a hierarchical Dirichlet model and predicting the optimal tree structure according to the model; 3) reorganizing the corpus using the complete category structure, i.e., associating each webpage with the relevant categories from the complete category hierarchy. Experimental results demonstrate that our proposed framework generates a high-quality category hierarchy and significantly boosts the retrieval performance.

Web portal services have become an important medium to deliver digital content (e.g. news, advertisements, etc.) to Web users in a timely fashion. To attract more users to various content modules on the Web portal, it is necessary to design a recommender system that can effectively achieve Web portal content optimization by automatically estimating content item attractiveness and relevance to user interests. The state-of-the-art online learning methodology adapts dedicated pointwise models to independently estimate the attractiveness score for each candidate content item. Although such pointwise models can be easily adapted for online recommendation, there still remain a few critical problems. First, this pointwise methodology fails to use invaluable user preferences between content items. Moreover, the performance of pointwise models decreases drastically when facing the problem of sparse learning samples. To address these problems, we propose exploring a new dynamic pairwise learning methodology for Web portal content optimization in which we exploit dynamic user preferences extracted based on users' actions on portal services to compute the attractiveness scores of content items. In this article, we introduce two specific pairwise learning algorithms, a straightforward graph-based algorithm and a formalized Bayesian modeling one. Experiments on large-scale data from a commercial Web portal demonstrate the significant improvement of pairwise methodologies over the baseline pointwise models. Further analysis illustrates that our new pairwise learning approaches can benefit personalized recommendation more than pointwise models, since the data sparsity is more critical for personalized content optimization.

Evaluation metric is an essential and integral part of a ranking system. In the past, several evaluation metrics have been proposed in information retrieval and web search, among them Discounted Cumulative Gain (DCG) has emerged as one that is widely adopted for evaluating the performance of ranking functions used in web search. However, the two sets of parameters, the gain values and discount factors, used in DCG are usually determined in a rather ad-hoc way, and their impacts have not been carefully analyzed. In this paper, we first show that DCG is generally not coherent, i.e., comparing the performance of ranking functions using DCG very much depends on the particular gain values and discount factors used. We then propose a novel methodology that can learn the gain values and discount factors from user preferences over rankings, modeled as a special case of learning linear utility functions. We also discuss how to extend our methods to handle tied preference pairs and how to explore active learning to reduce preference labeling. Numerical simulations illustrate the effectiveness of our proposed methods. Moreover, experiments are also conducted over a side-by-side comparison data set from a commercial search engine to validate the proposed methods on real-world data.

Today's popular web search engines expand the search process beyond crawled web pages to specialized corpora ("verticals") like images, videos, news, local, sports, finance, shopping etc., each with its own specialized search engine. Search federation deals with problems of the selection of search engines to query and merging of their results into a single result set. Despite a few recent advances, the problem is still very challenging. First, due to the heterogeneous nature of different verticals, how the system merges the vertical results with the web documents to serve the user's information need is still an open problem. Moreover, the scale of the search engine and the increasing number of vertical properties requires a solution which is efficient and scaleable. In this paper, we propose a unified framework for the search federation problem. We model the search federation as a contextual bandit problem. The system uses reward as a proxy for user satisfaction. Given a query, our system predicts the expected reward for each vertical, then organizes the search result page (SERP) in a way which maximizes the total reward. Instead of relying on human judges, our system leverages implicit user feedback to learn the model. The method is efficient to implement and can be applied to verticals of different nature. We have successfully deployed the system to three different markets, and it handles multiple verticals in each market. The system is now serving hundreds of millions of queries live each day, and has improved user metrics considerably.

Twitter has become one of the most popular platforms for users to share information in real time. However, as an individual tweet is short and lacks sufficient contextual information, users cannot effectively understand or consume information on Twitter, which can either make users less engaged or even detached from using Twitter. In order to provide informative context to a Twitter user, we propose the task of Twitter context summarization, which generates a succinct summary from a large but noisy Twitter context tree. Traditional summarization techniques only consider text information, which is insufficient for Twitter context summarization task, since text information on Twitter is very sparse. Given that there are rich user interactions in Twitter, we thus study how to improve summarization methods by leveraging such signals. In particular, we study how user influence models, which project user interaction information onto a Twitter context tree, can help Twitter context summarization within a supervised learning framework. To evaluate our methods, we construct a data set by asking human editors to manually select the most informative tweets as a summary. Our experimental results based on this editorial data set show that Twitter context summarization is a promising research topic and pairwise user influence signals can significantly improve the task performance.

The World Wide Web is highly dynamic and is constantly evolving to cover the latest information about the physical and social updates in the world. At the same time, the changes in web contents are entangled with new information needs and time-sensitive user interactions with information sources. To address these temporal information needs effectively, it is essential for the search engines to model web dynamics and understand the changes in user behavior over time that are caused by them.

Click models aim at extracting intrinsic relevance of documents to queries from biased user clicks. One basic modeling assumption made in existing work is to treat such intrinsic relevance as an atomic query-document-specific parameter, which is solely estimated from historical clicks without using any content information about a document or relationship among the clicked/skipped documents under the same query. Due to this overly simplified assumption, existing click models can neither fully explore the information about a document's relevance quality nor make predictions of relevance for any unseen documents.

In Web search and vertical search, recency ranking refers to retrieving and ranking documents by both relevance and freshness. As impoverished in-links and click information is the the biggest challenge for recency ranking, we advocate the use of Twitter data to address the challenge in this article. We propose a method to utilize Twitter TinyURL to detect fresh and high-quality documents, and leverage Twitter data to generate novel and effective features for ranking. The empirical experiments demonstrate that the proposed approach effectively improves a commercial search engine for both Web search ranking and tweet vertical ranking.

Web portal services have become an important medium to deliver digital content and service, such as news, advertisements, and so on, to Web users in a timely fashion. To attract more users to various content modules on the Web portal, it is necessary to design a recommender system that can effectively achieve online content optimization by automatically estimating content items' attractiveness and relevance to users' interests. User interaction plays a vital role in building effective content optimization, as both implicit user feedbacks and explicit user ratings on the recommended items form the basis for designing and learning recommendation models. However, user actions on real-world Web portal services are likely to represent many implicit signals about users' interests and content attractiveness, which need more accurate interpretation to be fully leveraged in the recommendation models. To address this challenge, we investigate a couple of critical aspects of the online learning framework for personalized content optimization on Web portal services, and, in this paper, we propose deeper user action interpretation to enhance those critical aspects. In particular, we first propose an approach to leverage historical user activity to build behavior-driven user segmentation; then, we introduce an approach for interpreting users' actions from the factors of both user engagement and position bias to achieve unbiased estimation of content attractiveness. Our experiments on the large-scale data from a commercial Web recommender system demonstrate that recommendation models with our user action interpretation can reach significant improvement in terms of online content optimization over the baseline method. The effectiveness of our user action interpretation is also proved by the online test results on real user traffic.

We consider the problem of identifying primary categories of a business listing among the categories provided by the owner of the business. The category information submitted by business owners cannot be trusted with absolute certainty since they may purposefully add some secondary or irrelevant categories to increase recall in local search results, which makes category search very challenging for local search engines. Thus, identifying primary categories of a business is a crucial problem in local search. This problem can be cast as a multi-label classification problem with a large number of categories. However, the large scale of the problem makes it infeasible to use conventional supervised-learning-based text categorization approaches.

With the rapid growth of E-Commerce on the Internet, online product search service has emerged as a popular and effective paradigm for customers to find desired products and select transactions. Most product search engines today are based on adaptations of relevance models devised for information retrieval. However, there is still a big gap between the mechanism of finding products that customers really desire to purchase and that of retrieving products of high relevance to customers' query. In this paper, we address this problem by proposing a new ranking framework for enhancing product search based on dynamic best-selling prediction in E-Commerce. Specifically, we first develop an effective algorithm to predict the dynamic best-selling, i.e. the volume of sales, for each product item based on its transaction history. By incorporating such best-selling prediction with relevance, we propose a new ranking model for product search, in which we rank higher the product items that are not only relevant to the customer's need but with higher probability to be purchased by the customer. Results of a large scale evaluation, conducted over the dataset from a commercial product search engine, demonstrate that our new ranking method is more effective for locating those product items that customers really desire to buy at higher rank positions without hurting the search relevance.

We study the problem of labeling the edges of a social network graph (e.g., acquaintance connections in Facebook) as either positive (i.e., trust, true friendship) or negative (i.e., distrust, possible frenemy) relations. Such signed relations provide much stronger signal in tying the behavior of online users than the unipolar Homophily effect, yet are largely unavailable as most social graphs only contain unsigned edges.

Many vertical search tasks such as local search focus on specific domains. The meaning of relevance in these verticals is domain-specific and usually consists of multiple well-defined aspects (e.g., text matching and distance in local search). Thus the overall relevance between a query and a document is a tradeoff between multiple relevance aspects. Such a tradeoff can vary for different types of queries or in different contexts. In this paper, we explore these vertical-specific aspects in the learning to rank setting. We propose a novel formulation in which the relevance between a query and a document is assessed with respect to each aspect, forming the multi-aspect relevance. In order to compute a ranking function, we study two types of learning-based approaches to estimate the tradeoff between these relevance aspects: a label aggregation method and a model aggregation method. Since there are only a few aspects, a minimal amount of training data is needed to learn the tradeoff. We conduct both offline and online test experiments on a local search engine and the experimental results show that our proposed multi-aspect relevance formulation is very promising. The two types of aggregation methods perform more effectively than a set of baseline methods including a conventional learning to rank method.

Learning to rank arises in many information retrieval applications, ranging from Web search engine, online advertising to recommendation systems. Traditional ranking mainly focuses on one type of data source, and effective modeling relies on a sufficiently large number of labeled examples, which require expensive and time-consuming labeling process. However, in many real-world applications, ranking over multiple related heterogeneous domains becomes a common situation, where in some domains we may have a relatively large amount of training data while in some other domains we can only collect very little. Theretofore, how to leverage labeled information from related heterogeneous domain to improve ranking in a target domain has become a problem of great interests. In this paper, we propose a novel probabilistic model, pairwise cross-domain factor model, to address this problem. The proposed model learns latent factors(features) for multi-domain data in partially-overlapped heterogeneous feature spaces. It is capable of learning homogeneous feature correlation, heterogeneous feature correlation, and pairwise preference correlation for cross-domain knowledge transfer. We also derive two PCDF variations to address two important special cases. Under the PCDF model, we derive a stochastic gradient based algorithm, which facilitates distributed optimization and is flexible to adopt different loss functions and regularization functions to accommodate different data distributions. The extensive experiments on real world data sets demonstrate the effectiveness of the proposed model and algorithm.

In contrast to traditional Web search, where topical relevance is often the main selection criterion, news search is characterized by the increased importance of freshness. However, the estimation of relevance and freshness, and especially the relative importance of these two aspects, are highly specific to the query and the time when the query was issued. In this work, we propose a unified framework for modeling the topical relevance and freshness, as well as their relative importance, based on click logs. We use click statistics and content analysis techniques to define a set of temporal features, which predict the right mix of freshness and relevance for a given query. Experimental results on both historical click data and editorial judgments demonstrate the effectiveness of the proposed approach.

Traditional machine-learned ranking systems for Web search are often trained to capture stationary relevance of documents to queries, which have limited ability to track nonstationary user intention in a timely manner. In recency search, for instance, the relevance of documents to a query on breaking news often changes significantly over time, requiring effective adaptation to user intention. In this article, we focus on recency search and study a number of algorithms to improve ranking results by leveraging user click feedback. Our contributions are threefold. First, we use commercial search engine sessions collected in a random exploration bucket for reliable offline evaluation of these algorithms, which provides an unbiased comparison across algorithms without online bucket tests. Second, we propose an online learning approach that reranks and improves the search results for recency queries near real-time based on user clicks. This approach is very general and can be combined with sophisticated click models. Third, our empirical comparison of a dozen algorithms on real-world search data suggests importance of a few algorithmic choices in these applications, including generalization across different query-document pairs, specialization to popular queries, and near real-time adaptation of user clicks for reranking.

Definition acquisition is a necessary step in building an artificial cognitive assistant that helps military personnel to gain fast and precise understanding of the various terms and procedures defined in applicable legal documents. We approach the task of identifying definitional sentences from operations law documents by formalizing this task as a sentence-classification task and solving it by using machine-learning methods. This paper reports on a series of empirical experiments in that we evaluate and compare the performance of learning algorithms in terms of label-prediction accuracy. Using supervised techniques results in an F1 score of 95.93% and a 96.72% recall rate. However, for real-world applications, it would be too costly and unrealistic to ask personnel involved in military operations to label substantial amounts of data in order to build a new classifier for different types or genres of text data. Therefore, we propose and implement a semisupervised (SS) solution that trades off prediction accuracy to label efficiency. Our SS approach achieves a 90.47% F1 score and 93.44% recall rate by using only eight sentences labeled by a human expert.

With rapid growth of commercial search engines, detecting multilingual and multi-regional intent underlying search queries becomes a critical challenge to serve international users with diverse language and region requirements. We introduce a query intent probabilistic model, whose input is the number of clicks on documents from different regions and in different language, while the output of this model is a smoothed probabilistic distribution of multilingual and multi-regional query intent. Based on an editorial test to evaluate the accuracy of the intent classifier, our probabilistic model could improve the accuracy of multilingual intent detection for 15%, and improve multi-regional intent detection for 18%. To improve web search quality, we propose a set of new ranking features to combine multilingual and multi-regional query intent with document language/region attributes, and apply different approaches in integrating intent information to directly affect ranking. The experiments show that the novel features could provide 2.31% NDCG@1 improvement and 1.81% NDCG@5 improvement.

Local search service (e.g. Yelp, Yahoo! Local) has emerged as a popular and effective paradigm for a wide range of information needs for local businesses; it now provides a viable and even more effective alternative to general purpose web search for queries on local businesses. However, due to the diversity of information needs behind local search, it is necessary to use different information retrieval strategies for different query types in local search. In this paper, we explore a taxonomy of local search driven by users' information needs, which categorizes local search queries into three types: business category, chain business, and non-chain business. To decide which search strategy to use for each category in this taxonomy without placing the burden on the web users, it is indispensable to build an automatic local query classifier. However, since local search queries yield few online features and it is expensive to obtain editorial labels, it is insufficient to use only a supervised learning approach. In this paper, we address these problems by developing a semi-supervised approach for mining information needs from a vast amount of unlabeled data from local query logs to boost local query classification. Results of a large scale evaluation over queries from a commercial local search site illustrate that the proposed semi-supervised method allow us to accurately classify a substantially larger proportion of local queries than the supervised learning approach.

User interaction plays a vital role in recommender systems. Previous studies on algorithmic recommender systems have mainly focused on modeling techniques and feature development. Traditionally, implicit user feedback or explicit user ratings on the recommended items form the basis for designing and training of recommendation algorithms. But user interactions in real-world Web applications (e.g., a portal website with different recommendation modules in the interface) are unlikely to be as ideal as those assumed by previously proposed models. To address this problem, we build an online learning framework for personalized recommendation. We argue that appropriate user action interpretation is critical for a recommender system. The main contribution in this paper is an approach of interpreting users' actions for the online learning to achieve better item relevance estimation. Our experiments on the large-scale data from a commercial Web recommender system demonstrate significant improvement in terms of a precision metric over the baseline model that does not incorporate user action interpretation. The efficacy of this new algorithm is also proved by the online test results on real user traffic.

Mobile search is still in infancy compared with general purpose web search. With limited training data and weak relevance features, the ranking performance in mobile search is far from satisfactory. To address this problem, we propose to leverage the knowledge of Web search to enhance the ranking of mobile search. In this paper, we first develop an equivalent page conversion between web search and mobile search, then we design a few novel ranking features, generated from the click-through data in web search, for estimating the relevance of mobile search. Large scale evaluations demonstrate that the knowledge from web search is quite effective for boosting the relevance of ranking on mobile search.

Web search ranking functions are typically learned to rank search results based on features of individual documents, i.e., pointwise features. Hence, the rich relationships among documents, which contain multiple types of useful information, are either totally ignored or just explored very limitedly. In this paper, we propose to explore multiple pairwise relationships between documents in a learning setting to rerank search results. In particular, we use a set of pairwise features to capture various kinds of pairwise relationships and design two machine learned re-ranking methods to effectively combine these features with a base ranking function: a pairwise comparison method and a pairwise function decomposition method. Furthermore, we propose several schemes to estimate the potential gains of our re-ranking methods on each query and selectively apply them to queries with high confidence. Our experiments on a large scale commercial search engine editorial data set show that considering multiple pairwise relationships is quite beneficial and our proposed methods can achieve significant gain over methods which only consider pointwise features or a single type of pairwise relationship.

In this paper, we present a system for clustering the search results of a news search engine. The news search interface includes the relevant news articles to a given query organized in terms of related news stories. Here each cluster corresponds to a news story and the news articles are clustered into stories. We present a system that clusters the search results of a news search system in a fast and scalable manner. The clustering system is organized into three components including offline clustering, incremental clustering and realtime clustering. We propose novel techniques for clustering the search results in realtime. The experimental results with large collections of news documents reveal that our system is both scalable and also achieves good accuracy in clustering the news search results.

With the explosive growth of online news readership, recommending interesting news articles to users has become extremely important. While existing Web services such as Yahoo! and Digg attract users' initial clicks by leveraging various kinds of signals, how to engage such users algorithmically after their initial visit is largely under-explored. In this paper, we study the problem of post-click news recommendation. Given that a user has perused a current news article, our idea is to automatically identify "related" news articles which the user would like to read afterwards. Specifically, we propose to characterize relatedness between news articles across four aspects: relevance, novelty, connection clarity, and transition smoothness. Motivated by this understanding, we define a set of features to capture each of these aspects and put forward a learning approach to model relatedness. In order to quantitatively evaluate our proposed measures and learn a unified relatedness function, we construct a large test collection based on a four-month commercial news corpus with editorial judgments. The experimental results show that the proposed heuristics can indeed capture relatedness, and that the learned unified relatedness function works quite effectively.

Recency ranking refers to the ranking of web results by accounting for both relevance and freshness. This is particularly important for "recency sensitive" queries such as breaking news queries. In this study, we propose a set of novel click features to improve machine learned recency ranking. Rather than computing simple aggregate click through rates, we derive these features using the temporal click through data and query reformulation chains. One of the features that we use is click buzz that captures the spiking interest of a url for a query. We also propose time weighted click through rates which treat recent observations as being exponentially more important. The promotion of fresh content is typically determined by the query intent which can change dynamically over time. Quite often users query reformulations convey clues about the query's intent. Hence we enrich our click features by following query reformulations which typically benefit the first query in the chain of reformulations. Our experiments show these novel features can improve the NDCG5 of a major online search engine's ranking for "recency sensitive" queries by up to 1.57%. This is one of the very few studies that exploits temporal click through data and query reformulations for recency ranking.

Temporal information can be leveraged and incorporated to improve web search ranking. In this work, we propose a method to improve the ranking of search results by identifying the fundamental properties of temporal behavior of low-quality hosts and spam-prone queries in search logs and modeling those properties as quantifiable features. In particular, we introduce the concepts of host churn, a measure of changes in host visibility for user queries, and query volatility, a measure of semantic instability of query results, and propose the methods for construction of temporal profiles from search query logs that can be used for estimation of a set of features based on the introduced concepts. The utility of the proposed concepts has been experimentally demonstrated for two language-independent search tasks: the regression-based ranking of search results and a novel classification problem of detecting spam-prone queries introduced in this work.

There have been great needs to develop effective methods for combining multiple rankings from heterogeneous domains into one single rank list arising from many recent web search applications, such as integrating web search results from multiple engines, facets, or verticals. We define this problem as Learning to blend rankings from multiple domains. We propose a class of learning-to-blend methods that learn a monotonically increasing transformation for each ranking so that the rank order in each domain is preserved and the transformed values are comparable across multiple rankings. The transformation learning can be tackled by solving a quadratic programming problem. The novel machine learning method for blending multiple ranking lists is evaluated with queries sampled from a commercial search engine and a promising improvement of Discounted Cumulative Gain has been observed.

Learning to rank arises in many information retrieval applications, ranging from Web search engine, online advertising to recommendation system. In learning to rank, the performance of a ranking function heavily depends on the number of labeled examples in the training set; on the other hand, obtaining labeled examples for training data is very expensive and time-consuming. This presents a great need for making use of available auxiliary data, i.e., the within-domain unlabeled data and the out-of-domain labeled data. In this paper, we propose a general framework for ranking with auxiliary data, which is applicable to various ranking applications. Under this framework, we derive a generic ranking algorithm to effectively make use of both the within-domain unlabeled data and the out-of-domain labeled data. The proposed algorithm iteratively learns ranking functions for target domain and source domains and enforces their consensus on the unlabeled data in the target domain.

Traditional machine-learned ranking algorithms for web search are trained in batch mode, which assume static relevance of documents for a given query. Although such a batch-learning framework has been tremendously successful in commercial search engines, in scenarios where relevance of documents to a query changes over time, such as ranking recent documents for a breaking news query, the batch-learned ranking functions do have limitations. Users' real-time click feedback becomes a better and timely proxy for the varying relevance of documents rather than the editorial judgments provided by human editors. In this paper, we propose an online learning algorithm that can quickly learn the best re-ranking of the top portion of the original ranked list based on real-time users' click feedback. In order to devise our algorithm and evaluate it accurately, we collected exploration bucket data that removes positional biases on clicks on the documents for recency-classified queries. Our initial experimental result shows that our scheme is more capable of quickly adjusting the ranking to track the varying relevance of documents reflected in the click feedback, compared to batch-trained ranking functions.

Learning to rank arises in many information retrieval applications, ranging from Web search engine, online advertising to recommendation system. In learning to rank, the performance of a ranking model is strongly affected by the number of labeled examples in the training set; on the other hand, obtaining labeled examples for training data is very expensive and time-consuming. This presents a great need for the active learning approaches to select most informative examples for ranking learning; however, in the literature there is still very limited work to address active learning for ranking. In this paper, we propose a general active learning framework, Expected Loss Optimization (ELO), for ranking. The ELO framework is applicable to a wide range of ranking functions. Under this framework, we derive a novel algorithm, Expected DCG Loss Optimization (ELO-DCG), to select most informative examples. Furthermore, we investigate both query and document level active learning for raking and propose a two-stage ELO-DCG algorithm which incorporate both query and document selection into active learning. Extensive experiments on real-world Web search data sets have demonstrated great potential and effective-ness of the proposed framework and algorithms.

In web search, recency ranking refers to ranking documents by relevance which takes freshness into account. In this paper, we propose a retrieval system which automatically detects and responds to recency sensitive queries. The system detects recency sensitive queries using a high precision classifier. The system responds to recency sensitive queries by using a machine learned ranking model trained for such queries. We use multiple recency features to provide temporal evidence which effectively represents document recency. Furthermore, we propose several training methodologies important for training recency sensitive rankers. Finally, we develop new evaluation metrics for recency sensitive queries. Our experiments demonstrate the efficacy of the proposed approaches.

Ranking a set of retrieved documents according to their relevance to a given query has become a popular problem at the intersection of web search, machine learning, and information retrieval. Recent work on ranking focused on a number of different paradigms, namely, pointwise, pairwise, and list-wise approaches. Each of those paradigms focuses on a different aspect of the dataset while largely ignoring others. The current paper shows how a combination of them can lead to improved ranking performance and, moreover, how it can be implemented in log-linear time.

Realtime web search refers to the retrieval of very fresh content which is in high demand. An effective portal web search engine must support a variety of search needs, including realtime web search. However, supporting realtime web search introduces two challenges not encountered in non-realtime web search: quickly crawling relevant content and ranking documents with impoverished link and click information. In this paper, we advocate the use of realtime micro-blogging data for addressing both of these problems. We propose a method to use the micro-blogging data stream to detect fresh URLs. We also use micro-blogging data to compute novel and effective features for ranking fresh URLs. We demonstrate these methods improve effective of the portal web search engine for realtime web search.

Both the quality and quantity of training data have significant impact on the performance of ranking functions in the context of learning to rank for web search. Due to resource constraints, training data for smaller search engine markets are scarce and we need to leverage existing training data from large markets to enhance the learning of ranking function for smaller markets. In this paper, we present a boosting framework for learning to rank in the multi-task learning context for this purpose. In particular, we propose to learn non-parametric common structures adaptively from multiple tasks in a stage-wise way. An algorithm is developed to iteratively discover super-features that are effective for all the tasks. The estimation of the functions for each task is then learned as a linear combination of those super-features. We evaluate the performance of this multi-task learning method for web search ranking using data from a search engine. Our results demonstrate that multi-task learning methods bring significant relevance improvements over existing baseline methods.

In many Web search engines, a ranking function is selected for deployment mainly by comparing the relevance measurements over candidates. Due to the dynamical nature of the Web, the ranking features and the query and URL distribution on which the ranking functions are built, may change dramatically over time. The actual relevance of the function may degrade, and thus the previous function selection conclusions become invalid. In this work we suggest to select Web ranking functions according to both their relevance and robustness to the changes that may lead to relevance degradation over time. We argue that the ranking robustness can be effectively measured by taking into account the ranking score distribution across search results. We then propose two alternatives to the NDCG metric that both incorporate ranking robustness into ranking function evaluation and selection. A machine learning approach is developed to learn the parameters that control the metric sensitivity to score turbulence, from human-judged preference data.

Discounted cumulative gain (DCG) is widely used for evaluating ranking functions. It is therefore natural to learn a ranking function that directly optimizes DCG. However, DCG is non-smooth, rendering gradient-based optimization algorithms inapplicable. To remedy this, smoothed versions of DCG have been proposed but with only partial success. In this paper, we first present analysis that shows it is ineffective using the gradient of the smoothed DCG to drive the optimization algorithm. We then propose a novel approach, SHF-SDCG, for smoothing DCG by using smoothed hinge functions (SHF). It has the advantage of seamlessly transition from driving the optimization mimicking pairwise learning when the ranking function does not fit the data well, to driving the optimization using DCG when the ranking function becomes more accurate. SHF-SDCG is then extended to REG-SHF-SDCG, an algorithm which gradually transits from pointwise and pairwise to listwise learning. Finally experimental results are provided to validate the effectiveness of SHF-SDCG and REG-SHF-SDCG.

To overcome the training data insufficiency problem for dedicated model in topical ranking, this paper proposes to utilize click-through data to improve learning. The efficacy of click-through data is explored under the framework of preference learning. The empirical experiment on a commercial search engine shows that, the model trained with the dedicated labeled data combined with skip-next preferences could beat the baseline model and the generic model in NDCG5 for 4.9% and 2.4% respectively.

Proximity of query terms in a document is an important criterion in IR. However, no investigation has been made to determine the most useful term sequences for which proximity should be considered. In this study, we test the effectiveness of using proximity of partial term sequences (n-grams) for Web search. We observe that the proximity of sequences of 3 to 5 terms is most effective for long queries, while shorter or longer sequences appear less useful. This suggests that combinations of 3 to 5 terms can best capture the intention in user queries. In addition, we also experiment with weighing the importance of query sub-sequences using query log frequencies. Our preliminary tests show promising empirical results.

People identification is an essential task for video content analysis in a surveillance system. To construct a good classifier requires a large amount of training data, which may not be obtained in some scenario. In this paper, we propose an approach to augment insufficient training data by labeling identical video images that have removed people's identities by masking faces. We show user study results that human subjects can perform reasonably well in labeling pairwise constraints from face obscured images. We also present a new discriminative learning algorithm WPKLR to handle uncertainties in pairwise constraints. The effectiveness of the proposed approach is demonstrated using video captured in a nursing home environment. The experiments show that the WPKLR approach can obtain a high accuracy of people identification using limited labeled data and noisy pairwise constraints, and meanwhile minimize the risk of exposing people's identities