Zero-shot object navigation tasks agents with locating target objects in unseen environments—a core capability of embodied intelligence. While recent vision-language navigation methods leverage Large Language Models (LLMs) for multimodal reasoning, they suffer from two key limitations: (1) semantic misalignment between language-grounded maps and real-world layouts, and (2) inefficiency due to LLMs’ lack of specialization for navigation-specific tasks. To address these challenges, we propose Chain-of-Search (CoS), a novel parameter-efficient framework that enables human-like decision-making via iterative semantic reasoning. First, CoS replaces traditional global maps with an optimal-benefit multi-map construction that continuously balances expected gain and cost throughout the navigation process. Second, we introduce a Parameter-Efficient Intent Aligner (PEIA), trained via a prompt-guided paradigm to align directional decisions with navigation intent. PEIA injects semantic cues into benefit-aware maps, enabling more rational and goal-consistent exploration. Finally, a Reflection-Guided Destination Verifier (RDV) confirms whether the target is reached via language-driven reasoning and corrects potential errors through self-reflection. CoS achieves state-of-the-art performance on HM3D (+2.8% SR) and MP3D (+1.2% SR) without relying on LLMs, demonstrating the effectiveness of lightweight, reasoning-centered navigation.

Vision-Language-Action models have recently emerged as a powerful paradigm for general-purpose robot learning, enabling agents to map visual observations and natural-language instructions into executable robotic actions. Though popular, they are primarily trained via supervised fine-tuning or training-time reinforcement learning, requiring explicit fine-tuning phases, human interventions, or controlled data collection. Consequently, existing methods remain unsuitable for challenging simulated- or physical-world deployments, where robots must respond autonomously and flexibly to evolving environments. To address this limitation, we introduce a Test-Time Reinforcement Learning for VLAs (TT-VLA), a framework that enables on-the-fly policy adaptation during inference. TT-VLA formulates a dense reward mechanism that leverages step-by-step task-progress signals to refine action policies during test time while preserving the SFT/RL-trained priors, making it an effective supplement to current VLA models. Empirical results show that our approach enhances overall adaptability, stability, and task success in dynamic, previously unseen scenarios under simulated and real-world settings. We believe TT-VLA offers a principled step toward self-improving, deployment-ready VLAs.

Xiqiao Xiong, Ouxiang Li, Zhuo Liu, Moxin Li, Wentao Shi, Fengbin Zhu, Qifan Wang, Fuli Feng. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Reasoning in Large Language Models (LLMs) has recently shown strong potential in enhancing generative recommendation through deep understanding of complex user preference. Existing approaches follow a reason-then-recommend paradigm, where LLMs perform step-by-step reasoning before item generation. However, this paradigm inevitably suffers from reasoning degradation (i.e., homogeneous or error-accumulated reasoning) due to the lack of intermediate verification, thus undermining the recommendation. To bridge this gap, we propose a novel reason-verify-recommend paradigm, which interleaves reasoning with verification to provide reliable feedback, guiding the reasoning process toward more faithful user preference understanding. To enable effective verification, we establish two key principles for verifier design: 1) reliability ensures accurate evaluation of reasoning correctness and informative guidance generation; and 2) multi-dimensionality emphasizes comprehensive verification across multi-dimensional user preferences. Accordingly, we propose an effective implementation called VRec. It employs a mixture of verifiers to ensure multi-dimensionality, while leveraging a proxy prediction objective to pursue reliability. Experiments on four real-world datasets demonstrate that VRec substantially enhances recommendation effectiveness and scalability without compromising efficiency.

Aligning Large Language Models (LLMs) with recommendation tasks represents an emerging paradigm in recommendation domain, exhibiting promising performance overall. However, these aligned recommendation LLMs often struggle with complex scenarios due to limitations in the current alignment task formulation, which optimizes LLMstodirectly generate user feedback without deliberation. To develop more reliable recommendation LLMs, we introduce a new Deliberative Recommendation task, which enforces explicit reasoning about user preferences as an additional alignment objective. To address this task, we propose a Reasoning-powered Recommender framework designed to enhance reasoning capabilities by leveraging verbalized user feedback in a step-wise manner. Specifically, this framework employs collaborative step-wise experts alongside specifically crafted expert-wise training strategies. Extensive experiments conducted on three real-world datasets demonstrate the rationality of the deliberative task formulation and the effectiveness of the proposed framework in improving both pre diction accuracy and reasoning quality. Our implementation is publicly available on GitHub: https://github.com/Peter-Fy/Reason4Rec.

Retrieval-Augmented Generation (RAG) can alleviate hallucinations of Large Language Models (LLMs) by referencing external documents. However, the misinformation in external documents may mislead LLMs' generation. To address this issue, we explore the task of "credibility-aware RAG", in which LLMs automatically adjust the influence of retrieved documents based on their credibility scores to counteract misinformation. To this end, we introduce a plug-and-play method named Credibility-aware Attention Modification (CrAM). CrAM identifies influential attention heads in LLMs and adjusts their attention weights based on the credibility of the documents, thereby reducing the impact of low-credibility documents. Experiments on Natual Questions and TriviaQA using Llama2-13B, Llama3-8B, and Qwen1.5-7B show that CrAM improves the RAG performance of LLMs against misinformation pollution by over 20%, even surpassing supervised fine-tuning methods.

Large Language Models (LLMs) are known to be vulnerable to backdoor attacks, where triggers embedded in poisoned samples can maliciously alter LLMs’ behaviors. In this paper, we move beyond attacking LLMs and instead examine backdoor attacks through the novel lens of natural language explanations. Specifically, we leverage LLMs’ generative capabilities to produce human-readable explanations for their decisions, enabling direct comparisons between explanations for clean and poisoned samples. Our results show that backdoored models produce coherent explanations for clean inputs but diverse and logically flawed explanations for poisoned data, a pattern consistent across classification and generation tasks for different backdoor attacks. Further analysis reveals key insights into the explanation generation process. At the token level, explanation tokens associated with poisoned samples only appear in the final few transformer layers. At the sentence level, attention dynamics indicate that poisoned inputs shift attention away from the original input context during explanation generation. These findings enhance our understanding of backdoor mechanisms in LLMs and present a promising framework for detecting vulnerabilities through explainability.

Peiqi Wang, ShengYun Peng, Xuewen Zhang, Hanchao Yu, Yibo Yang, Lifu Huang, Fujun Liu, Qifan Wang. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Large Multimodal Models (LMMs) have demonstrated impressive performance across numerous academic benchmarks.However, fine-tuning still remains essential to achieve satisfactory performance on downstream tasks, while the task-specific tuning samples are usually not readily available or expensive and time-consuming to obtain.To address this, we propose an error-driven data-efficient tuning framework that aims to efficiently adapt generic LMMs to newly emerging tasks without requiring extensive task-specific training samples.In our approach, a generic LMM, acting as a student model, is first evaluated on a small validation set of the target task, and then a more powerful model, acting as a teacher model, identifies the erroneous steps within the student model's reasoning steps and analyzes its capability gaps from fully addressing the target task.Based on these gaps, targeted training samples are further retrieved from existing taskagnostic datasets to tune the student model and tailor it to the target task.We perform extensive experiments across three different training data scales and seven tasks, demonstrating that our training paradigm significantly and efficiently improves LMM's performance on downstream tasks, achieving an average performance boost of 7.01% 1 .

Federated averaging (FedAvg) is the most fundamental algorithm in Federated learning (FL). Previous theoretical results assert that FedAvg convergence and generalization degenerate under heterogeneous clients. However, recent empirical results show that FedAvg can perform well in many real-world heterogeneous tasks. These results reveal an inconsistency between FL theory and practice that is not fully explained. In this paper, we show that common heterogeneity measures contribute to this inconsistency based on rigorous convergence analysis. Furthermore, we introduce a new measure \textit{client consensus dynamics} and prove that \textit{FedAvg can effectively handle client heterogeneity when an appropriate aggregation strategy is used}. Building on this theoretical insight, we present a simple and effective FedAvg variant termed FedAWARE. Extensive experiments on three datasets and two modern neural network architectures demonstrate that FedAWARE ensures faster convergence and better generalization in heterogeneous client settings. Moreover, our results show that FedAWARE can significantly enhance the generalization performance of advanced FL algorithms when used as a plug-in module.

Multiple-choice question (MCQ) benchmarks such as MMLU and GPQA are widely used to assess the capabilities of large language models (LLMs). While accuracy remains the standard evaluation metric, recent work has introduced uncertainty quantification (UQ) methods, such as entropy, conformal prediction, and verbalized confidence, as complementary measures of model reliability and calibration. However, we find that these UQ methods, when applied to MCQ tasks, are unexpectedly fragile. Specifically, we show that fine-tuning a model on just 1,000 examples to adjust the probability of the first generated token, under the common prompting setup where the model is instructed to output only a single answer choice, can systematically distort a broad range of UQ methods across models, prompts, and domains, all while leaving answer accuracy unchanged. We validate this phenomenon through extensive experiments on five instruction-tuned LLMs, tested under standard prompting, zero-shot chain-of-thought reasoning, and a biomedical question answering setting. In all cases, models retain similar accuracy but exhibit significantly degraded calibration. These results suggest that current UQ practices for MCQs are ''one-token deep'', driven more by first-token decoding behavior than by any deeper representation of uncertainty, and are easily manipulated through minimal interventions. Our findings call for more robust and interpretable approaches to uncertainty estimation, particularly in structured formats like MCQs, where confidence signals are often reduced to token-level heuristics.

Dense video captioning is crucial for enhancing video understanding in daily applications and presents a significant challenge in multimodal analysis. Existing methods often overlook video-to-dynamic-space mapping at varying scales, resulting in captions that lack specificity and remain overly general, failing to capture real-world physical detail. To address this limitation, we propose a multi-granularity Spatio-Temporal Reasoning (STaR) approach, which integrates: (i) efficient global feature integration to model long-term temporal dependencies, (ii) spatial attention mechanisms with position encoding to capture absolute spatial information, and (iii) cross-modal feature fusion to align and unify global, local, and spatial representations. Moreover, we enhance the framework using a Large Language Model (LLM) to improve the richness and naturalness of the generated descriptions. Comparative experiments have been conducted to evaluate the effectiveness of the proposed method on SoccerNet dataset. Experimental results demonstrate that our model effectively enhances localization accuracy and generates captions with superior temporal and spatial detail fidelity. The code is available at https://github.com/bread-555/STaR.

Mohammad Beigi, Ying Shen, Parshin Shojaee, Qifan Wang, Zichao Wang, Chandan K. Reddy, Ming Jin, Lifu Huang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Runjia Zeng, Guangyan Sun, Qifan Wang, Tong Geng, Sohail Dianat, Xiaotian Han, Raghuveer Rao, Xueling Zhang, Cheng Han, Lifu Huang, Dongfang Liu. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

How well can Multimodal Large Language Models (MLLMs) understand composite images? Composite images (CIs) are synthetic visuals created by merging multiple visual elements, such as charts, posters, or screenshots, rather than being captured directly by a camera. While CIs are prevalent in real-world applications, recent MLLM developments have primarily focused on interpreting natural images (NIs). Our research reveals that current MLLMs face significant challenges in accurately understanding CIs, often struggling to extract information or perform complex reasoning based on these images. We find that existing training data for CIs are mostly formatted for question-answer tasks (e.g., in datasets like ChartQA and ScienceQA), while high-quality image-caption datasets, critical for robust vision-language alignment, are only available for NIs. To bridge this gap, we introduce Composite Captions (CompCap), a flexible framework that leverages Large Language Models (LLMs) and automation tools to synthesize CIs with accurate and detailed captions. Using CompCap, we curate CompCap-118K, a dataset containing 118K image-caption pairs across six CI types. We validate the effectiveness of CompCap-118K by supervised fine-tuning MLLMs of three sizes: xGen-MM-inst.-4B and LLaVA-NeXT-Vicuna-7B/13B. Empirical results show that CompCap-118K significantly enhances MLLMs' understanding of CIs, yielding average gains of 1.7%, 2.0%, and 2.9% across eleven benchmarks, respectively.

Recently in robotics, Vision-Language-Action (VLA) models have emerged as a transformative approach, enabling robots to execute complex tasks by integrating visual and linguistic inputs within an end-to-end learning framework. Despite their significant capabilities, VLA models introduce new attack surfaces. This paper systematically evaluates their robustness. Recognizing the unique demands of robotic execution, our attack objectives target the inherent spatial and functional characteristics of robotic systems. In particular, we introduce two untargeted attack objectives that leverage spatial foundations to destabilize robotic actions, and a targeted attack objective that manipulates the robotic trajectory. Additionally, we design an adversarial patch generation approach that places a small, colorful patch within the camera's view, effectively executing the attack in both digital and physical environments. Our evaluation reveals a marked degradation in task success rates, with up to a 100\% reduction across a suite of simulated robotic tasks, highlighting critical security gaps in current VLA architectures. By unveiling these vulnerabilities and proposing actionable evaluation metrics, we advance both the understanding and enhancement of safety for VLA-based robotic systems, underscoring the necessity for continuously developing robust defense strategies prior to physical-world deployments.

Multimodal instruction tuning has proven to be an effective strategy for achieving zero-shot generalization by fine-tuning pre-trained Large Multimodal Models (LMMs) with instruction-following data. However, as the scale of LMMs continues to grow, fully fine-tuning these models has become highly parameter-intensive. Although Parameter-Efficient Fine-Tuning (PEFT) methods have been introduced to reduce the number of tunable parameters, a significant performance gap remains compared to full fine-tuning. Furthermore, existing PEFT approaches are often highly parameterized, making them difficult to interpret and control. In light of this, we introduce Multimodal Representation Tuning (MRT), a novel approach that focuses on directly editing semantically rich multimodal representations to achieve strong performance and provide intuitive control over LMMs. Empirical results show that our method surpasses current state-of-the-art baselines with significant performance gains (e.g., 1580.40 MME score) while requiring substantially fewer tunable parameters (e.g., 0.03% parameters). Additionally, we conduct experiments on editing instrumental tokens within multimodal representations, demonstrating that direct manipulation of these representations enables simple yet effective control over network behavior.

Deep neural networks have achieved remarkable accomplishments in practice. The success of these networks hinges on effective initialization methods, which are vital for ensuring stable and rapid convergence during training. Recently, initialization methods that maintain identity transition within layers have shown good efficiency in network training. These techniques (e.g., Fixup) set specific weights to zero to achieve identity control. However, settings of remaining weight (e.g., Fixup uses random values to initialize non-zero weights) will affect the inductive bias that is achieved only by a zero weight, which may be harmful to training. Addressing this concern, we introduce fully identical initialization (IDInit), a novel method that preserves identity in both the main and sub-stem layers of residual networks. IDInit employs a padded identity-like matrix to overcome rank constraints in non-square weight matrices. Furthermore, we show the convergence problem of an identity matrix can be solved by stochastic gradient descent. Additionally, we enhance the universality of IDInit by processing higher-order weights and addressing dead neuron problems. IDInit is a straightforward yet effective initialization method, with improved convergence, stability, and performance across various settings, including large-scale datasets and deep models.

Alignment techniques enable Large Language Models (LLMs) to generate outputs that align with human preferences and play a crucial role in their effectiveness. However, their impact often diminishes when applied to Small Language Models (SLMs), likely due to the limited capacity of these models. Instead of directly applying existing alignment techniques to SLMs, we propose to utilize a well-aligned teacher LLM to guide the alignment process for these models, thereby facilitating the transfer of the teacher's knowledge of human preferences to the student model. To achieve this, we first explore a straightforward approach, Dual-Constrained Knowledge Distillation (DCKD), that employs knowledge distillation with two KL-divergence constraints from the aligned teacher to the unaligned student. To further enhance the student's ability to distinguish between preferred and dispreferred responses, we then propose Advantage-Guided Distillation for Preference Alignment (ADPA), which leverages an advantage function from the aligned teacher to deliver more nuanced, distribution-level reward signals for the student's alignment. Our experimental results show that these two approaches appreciably improve the alignment of SLMs and narrow the performance gap with larger counterparts. Among them, ADPA demonstrates superior performance and achieves even greater effectiveness when integrated with DCKD. Our code is available at https://github.com/SLIT-AI/ADPA .

Achieving human-level intelligence requires refining cognitive distinctions between \textit{System 1} and \textit{System 2} thinking. While contemporary AI, driven by large language models, demonstrates human-like traits, it falls short of genuine cognition. Transitioning from structured benchmarks to real-world scenarios presents challenges for visual agents, often leading to inaccurate and overly confident responses. To address the challenge, we introduce \textbf{\textsc{FaST}}, which incorporates the \textbf{Fa}st and \textbf{S}low \textbf{T}hinking mechanism into visual agents. \textsc{FaST} employs a switch adapter to dynamically select between \textit{System 1/2} modes, tailoring the problem-solving approach to different task complexity. It tackles uncertain and unseen objects by adjusting model confidence and integrating new contextual data. With this novel design, we advocate a \textit{flexible system}, \textit{hierarchical reasoning} capabilities, and a \textit{transparent decision-making} pipeline, all of which contribute to its ability to emulate human-like cognitive processes in visual intelligence. Empirical results demonstrate that \textsc{FaST} outperforms various well-known baselines, achieving 80.8\% accuracy over $VQA^{v2}$ for visual question answering and 48.7\% $GIoU$ score over ReasonSeg for reasoning segmentation, demonstrate \textsc{FaST}'s superior performance. Extensive testing validates the efficacy and robustness of \textsc{FaST}'s core components, showcasing its potential to advance the development of cognitive visual agents in AI systems.

Recent advancements in Vision-Language Models (VLMs) have led to the emergence of Vision-Language Generalists (VLGs) capable of understanding and generating both text and images. However, seamlessly generating an arbitrary sequence of text and images remains a challenging task for the current VLGs. One primary limitation lies in applying a unified architecture and the same set of parameters to simultaneously model discrete text tokens and continuous image features. Recent works attempt to tackle this fundamental problem by introducing modality-aware expert models. However, they employ identical architectures to process both text and images, disregarding the intrinsic inductive biases in these two modalities. In this work, we introduce Modality-Specialized Synergizers (MoSS), a novel design that efficiently optimizes existing unified architectures of VLGs with modality-specialized adaptation layers, i.e., a Convolutional LoRA for modeling the local priors of image patches and a Linear LoRA for processing sequential text. This design enables more effective modeling of modality-specific features while maintaining the strong cross-modal integration gained from pretraining. In addition, to improve the instruction-following capability on interleaved text-and-image generation, we introduce LeafInstruct, the first open-sourced interleaved instruction tuning dataset comprising 184,982 high-quality instances on more than 10 diverse domains. Extensive experiments show that VLGs integrated with MoSS achieve state-of-the-art performance, significantly surpassing baseline VLGs in complex interleaved generation tasks. Furthermore, our method exhibits strong generalizability on different VLGs.

The task of item-to-item (I2I) retrieval is to identify a set of relevant and highly engaging items based on a given trigger item. It is a crucial component in modern recommendation systems, where users' previously engaged items serve as trigger items to retrieve relevant content for future engagement. However, existing I2I retrieval models in industry are primarily built on co-engagement data and optimized using the recall measure, which overly emphasizes co-engagement patterns while failing to capture semantic relevance. This often leads to overfitting short-term co-engagement trends at the expense of long-term benefits such as discovering novel interests and promoting content diversity. To address this challenge, we propose MTMH, a Multi-Task and Multi-Head I2I retrieval model that achieves both high recall and semantic relevance. Our model consists of two key components: 1) a multi-task learning loss for formally optimizing the trade-off between recall and semantic relevance, and 2) a multi-head I2I retrieval architecture for retrieving both highly co-engaged and semantically relevant items. We evaluate MTMH using proprietary data from a commercial platform serving billions of users and demonstrate that it can improve recall by up to 14.4% and semantic relevance by up to 56.6% compared with prior state-of-the-art models. We also conduct live experiments to verify that MTMH can enhance both short-term consumption metrics and long-term user-experience-related metrics. Our work provides a principled approach for jointly optimizing I2I recall and semantic relevance, which has significant implications for improving the overall performance of recommendation systems.

Human beings perceive the real world through a spectrum of sensory modalities, encompassing auditory, visual, and linguistic faculties. This work proposes an industry-level omni-modal large language model (LLM) pipeline that integrates auditory, visual, and linguistic modalities to overcome challenges such as limited tri-modal datasets, high computational costs, and complex feature alignments. Our pipeline consists of three main components: First, a modular, end-to-end framework enabling flexible configuration of various encoder-LLM-decoder architectures. Second, a lightweight training strategy that pre-trains audio-language alignment on the state-of-the-art vision-language model Qwen2.5-VL, thus avoiding the costly pre-training of vision-specific modalities. Third, an audio synthesis pipeline that generates high-quality audio-text data from diverse real-world scenarios, supporting applications such as Automatic Speech Recognition and Speech-to-Speech chat. To this end, we introduce an industry-level omni-modal LLM, NEXUS-O. Extensive experiments validate the efficacy of our pipeline, yielding the following key findings: (1) In the visual understanding task, NEXUSO exhibits superior performance compared with its backbone model - Qwen2.5-VL-7B, validating the efficiency of our training strategy. (2) Within the English Spoken Question-Answering task, the model achieves better accuracy than the same-period competitor (i.e, MiniCPM-o2.6-7B) in the LLaMA Q. benchmark. (3) In our realworld ASR testset, NEXUS-O achieves outstanding performance, indicating its robustness in real scenarios. (4) In the Speech-to-Text Translation task, our model outperforms Qwen2-Audio-Instruct-7B. (5) In the Text-to-Speech task, based on pretrained vocoder (e.g., Fishspeech1.4 or CosyVoice2.0), NEXUS-O is comparable to its backbone vocoder on Seed-TTS benchmark. (6) An in-depth analysis of tri-modal alignment reveals that incorporating the audio modality enhances representational alignment between vision and language.

Sanwoo Lee, Jiahao Liu, Qifan Wang, Jingang Wang, Xunliang Cai, Yunfang Wu. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Prompt-based learning has emerged as a parameter-efficient finetuning (PEFT) approach to facilitate Large Language Model (LLM) adaptation to downstream tasks by conditioning generation with task-aware guidance. Despite its successes, current prompt-based learning methods heavily rely on laborious grid searching for optimal prompt length and typically require considerable number of prompts, introducing additional computational burden. Worse yet, our pioneer findings indicate that the task-aware prompt design is inherently limited by its absence of instance-aware information, leading to a subtle attention interplay with the input sequence. In contrast, simply incorporating instance-aware information as a part of the guidance can enhance the prompt-tuned model performance without additional fine-tuning. Moreover, we find an interesting phenomenon, namely "attention anchor", that incorporating instance-aware tokens at the earliest position of the sequence can successfully preserve strong attention to critical structural information and exhibit more active attention interaction with all input tokens. In light of our observation, we introduce Capsule Prompt-Tuning (CaPT), an efficient and effective solution that leverages off-the-shelf, informative instance semantics into prompt-based learning. Our approach innovatively integrates both instance-aware and task-aware information in a nearly parameter-free manner (i.e., one single capsule prompt). Empirical results demonstrate that our method can exhibit superior performance across various language tasks (e.g., 84.03\% average accuracy on T5-Large), serving as an "attention anchor," while enjoying high parameter efficiency (e.g., 0.003\% of model parameters on Llama3.2-1B).

We introduce Autoregressive Retrieval Augmentation (AR-RAG), a novel paradigm that enhances image generation by autoregressively incorporating k-nearest neighbor retrievals at the patch level. Unlike prior methods that perform a single, static retrieval before generation and condition the entire generation on fixed reference images, AR-RAG performs context-aware retrievals at each generation step, using prior-generated patches as queries to retrieve and incorporate the most relevant patch-level visual references, enabling the model to respond to evolving generation needs while avoiding limitations (e.g., over-copying, stylistic bias, etc.) prevalent in existing methods. To realize AR-RAG, we propose two parallel frameworks: (1) Distribution-Augmentation in Decoding (DAiD), a training-free plug-and-use decoding strategy that directly merges the distribution of model-predicted patches with the distribution of retrieved patches, and (2) Feature-Augmentation in Decoding (FAiD), a parameter-efficient fine-tuning method that progressively smooths the features of retrieved patches via multi-scale convolution operations and leverages them to augment the image generation process. We validate the effectiveness of AR-RAG on widely adopted benchmarks, including Midjourney-30K, GenEval and DPG-Bench, demonstrating significant performance gains over state-of-the-art image generation models.

Training large language models (LLMs) from scratch can yield models with unique functionalities and strengths, but it is costly and often leads to redundant capabilities. A more cost-effective alternative is to fuse existing pre-trained LLMs with different architectures into a more powerful model. However, a key challenge in existing model fusion is their dependence on manually predefined vocabulary alignment, which may not generalize well across diverse contexts, leading to performance degradation in several evaluation. To solve this, we draw inspiration from distribution learning and propose the probabilistic token alignment method as a general and soft mapping for alignment, named as PTA-LLM. Our approach innovatively reformulates token alignment into a classic mathematical problem: optimal transport, seamlessly leveraging distribution-aware learning to facilitate more coherent model fusion. Apart from its inherent generality, PTA-LLM exhibits interpretability from a distributional perspective, offering insights into the essence of the token alignment. Empirical results demonstrate that probabilistic token alignment enhances the target model's performance across multiple capabilities.

Leveraging Large Language Models (LLMs) for generative recommendation has attracted significant research interest, where item tokenization is a critical step. It involves assigning item identifiers for LLMs to encode user history and generate the next item. Existing approaches leverage either token-sequence identifiers, representing items as discrete token sequences, or single-token identifiers, using ID or semantic embeddings. Token-sequence identifiers face issues such as the local optima problem in beam search and low generation efficiency due to step-by-step generation. In contrast, single-token identifiers fail to capture rich semantics or encode Collaborative Filtering (CF) information, resulting in suboptimal performance.

Large language model-based agents are increasingly applied in the recommendation field due to their extensive knowledge and strong planning capabilities. While prior research has primarily focused on enhancing either the recommendation agent or the user agent individually, the collaborative interaction between the two has often been overlooked. Towards this research gap, we propose a novel framework that emphasizes the feedback loop process to facilitate the collaboration between the recommendation agent and the user agent. Specifically, the recommendation agent refines its understanding of user preferences by analyzing the feedback from the user agent on the item recommendation. Conversely, the user agent further identifies potential user interests based on the items and recommendation reasons provided by the recommendation agent. This iterative process enhances the ability of both agents to infer user behaviors, enabling more effective item recommendations and more accurate user simulations. Extensive experiments on three datasets demonstrate the effectiveness of the agentic feedback loop: the agentic feedback loop yields an average improvement of 11.52% over the single recommendation agent and 21.12% over the single user agent. Furthermore, the results show that the agentic feedback loop does not exacerbate popularity or position bias, which are typically amplified by the real-world feedback loop, highlighting its robustness. The source code is available at https://github.com/Lanyu0303/AFL.

Dense retrieval (DR) converts queries and documents into dense embeddings and measures the similarity between queries and documents in vector space. One of the major challenges in DR is the lack of domain-specific training data. While DR models can learn from large-scale public datasets like MS MARCO through transfer learning, evidence shows that not all DR models and domains can benefit from transfer learning. Recently, researchers have resorted to large language models (LLMs) to improve the zero-shot and few-shot DR models. However, the hard prompts or human-written prompts utilized in these works are suboptimal and the generated weak queries are often sensitive to the prompts. To tackle this, we propose soft prompt tuning for augmenting DR (SPTAR): for each task, we leverage soft prompt tuning to optimize a task-specific soft prompt on limited ground truth data and then prompt the LLMs to tag unlabeled documents with weak queries, yielding weak document–query pairs to train task-specific dense retrievers. We design a filter to select high-quality example document–query pairs in the prompt to further improve the quality of weak tagged queries. To the best of our knowledge, there is no prior work utilizing soft prompt tuning to augment DR models. Moreover, unlike much of the existing work, ours is based on popular open-source LLMs to ensure reproducible and deterministic results. Our experimental results demonstrate that SPTAR outperforms both unsupervised baselines and the recently proposed LLMs-based augmentation method for DR. • First to use LLMs with soft prompt tuning for augmenting dense retrieval tasks. • Novel soft prompt filter improves weak data quality. • Experiments show our method outperforms strong baselines. • Uses open-source LLMs for reproducible, robust results.

Leveraging Large Language Models as recommenders, referred to as LLMRec, is gaining traction and brings novel dynamics for modeling user preferences, particularly for cold-start users. However, existing LLMRec approaches primarily focus on text semantics and overlook the crucial aspect of incorporating collaborative information from user-item interactions, leading to potentially sub-optimal performance in warm-start scenarios. To ensure superior recommendations across both warm and cold scenarios, we introduceCoLLM, an innovative LLMRec approach that explicitly integrates collaborative information for recommendations. CoLLM treats collaborative information as a distinct modality, directly encoding it from well-established traditional collaborative models, and then tunes a mapping module to align this collaborative information with the LLM's input text token space for recommendations. By externally integrating traditional models, CoLLM ensures effective collaborative information modeling without modifying the LLM itself, providing the flexibility to adopt diverse collaborative information modeling mechanisms. Extensive experimentation validates that CoLLM adeptly integrates collaborative information into LLMs, resulting in enhanced recommendation performance.

Federated learning (FL) has surged in prominence due to its capability of collaborative model training without direct data sharing. However, the vast disparity in local data distributions among clients, often termed the nonindependent identically distributed (Non-IID) challenge, poses a significant hurdle to FL's generalization efficacy. The scenario becomes even more complex when not all clients participate in the training process, a common occurrence due to unstable network connections or limited computational capacities. This can greatly complicate the assessment of the trained models' generalization abilities. While a plethora of recent studies has centered on the generalization gap pertaining to unseen data from participating clients with diverse distributions, the distinction between the training distributions of participating clients and the testing distributions of nonparticipating ones has been largely overlooked. In response, our paper unveils an information-theoretic generalization framework for FL. Specifically, it quantifies generalization errors by evaluating the information entropy of local distributions and discerning discrepancies across these distributions. Inspired by our deduced generalization bounds, we introduce a weighted aggregation approach and a duo of client selection strategies. These innovations are designed to strengthen FL's ability to generalize and thus ensure that trained models perform better on nonparticipating clients by incorporating a more diverse range of client data distributions. Our extensive empirical evaluations reaffirm the potency of our proposed methods, aligning seamlessly with our theoretical construct.

Recommendation unlearning is an emerging task to serve users for erasing unusable data (e.g., some historical behaviors) from a well-trained recommender model. Existing methods process unlearning requests by fully or partially retraining the model after removing the unusable data. However, these methods are impractical due to the high computation cost of full retraining and the highly possible performance damage of partial training. In this light, a desired recommendation unlearning method should obtain a similar model as full retraining in a more efficient manner, i.e., achieving complete, efficient and harmless unlearning. In this work, we propose a new Influence Function-based Recommendation Unlearning (IFRU) framework, which efficiently updates the model without retraining by estimating the influence of the unusable data on the model via the influence function . In the light that recent recommender models use historical data for both the constructions of the optimization loss and the computational graph (e.g., neighborhood aggregation), IFRU jointly estimates the direct influence of unusable data on optimization loss and the spillover influence on the computational graph to pursue complete unlearning. Furthermore, we propose an importance-based pruning algorithm to reduce the cost of the influence function. IFRU is harmless and applicable to mainstream differentiable models. Extensive experiments demonstrate that IFRU achieves more than 250 times acceleration compared to retraining-based methods with recommendation performance comparable to full retraining. Codes are available at https://github.com/baiyimeng/IFRU.

Cross-lingual named entity recognition (CrossNER) faces challenges stemming from uneven performance due to the scarcity of multilingual corpora, especially for non-English data. While prior efforts mainly focus on data-driven transfer methods, a significant aspect that has not been fully explored is aligning both semantic and token-level representations across diverse languages. In this paper, we propose Multi-view Contrastive Learning for Cross-lingual Named Entity Recognition (MCL-NER). Specifically, we reframe the CrossNER task into a problem of recognizing relationships between pairs of tokens. This approach taps into the inherent contextual nuances of token-to-token connections within entities, allowing us to align representations across different languages. A multi-view contrastive learning framework is introduced to encompass semantic contrasts between source, codeswitched, and target sentences, as well as contrasts among token-to-token relations. By enforcing agreement within both semantic and relational spaces, we minimize the gap between source sentences and their counterparts of both codeswitched and target sentences. This alignment extends to the relationships between diverse tokens, enhancing the projection of entities across languages. We further augment CrossNER by combining self-training with labeled source data and unlabeled target data. Our experiments on the XTREME benchmark, spanning 40 languages, demonstrate the superiority of MCL-NER over prior data-driven and model-based approaches. It achieves a substantial increase of nearly +2.0 F1 scores across a broad spectrum and establishes itself as the new state-of-the-art performer.

Zihao Lin, Mohammad Beigi, Hongxuan Li, Yufan Zhou, Yuxiang Zhang, Qifan Wang, Wenpeng Yin, Lifu Huang. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Multimodal Large Language Models (MLLMs) have demonstrated remarkable proficiency in diverse tasks across different domains, with an increasing focus on improving their zeroshot generalization capabilities for unseen multimodal tasks.Multimodal instruction tuning has emerged as a successful strategy for achieving zero-shot generalization by fine-tuning pretrained models on diverse multimodal tasks through instructions.As MLLMs grow in complexity and size, the need for parameterefficient fine-tuning methods like Low-Rank Adaption (LoRA), which fine-tunes with a minimal set of parameters, becomes essential.However, applying LoRA in multimodal instruction tuning presents the challenge of task interference, which leads to performance degradation, especially when dealing with a broad array of multimodal tasks.To address this, this paper introduces a novel approach that integrates multimodal instruction tuning with Conditional Mixture-of-LoRA (MixLoRA).It innovates upon LoRA by dynamically constructing low-rank adaptation matrices tailored to the unique demands of each input instance, aiming to mitigate task interference.Experimental results on various multimodal evaluation datasets indicate that MixLoRA not only outperforms the conventional LoRA with the same or even higher ranks, demonstrating its efficacy and adaptability in diverse multimodal tasks 1 .

With the growing privacy concerns surrounding natural language understanding (NLU) applications, the need to train high-quality models while safeguarding data privacy has reached unprecedented importance. Federated learning (FL) offers a promising approach to collaborative model training by exchanging model gradients. However, many studies show that eavesdroppers in FL could develop sophisticated data reconstruction attack (DRA) to accurately reconstruct clients’ data from the shared gradients. Regrettably, current DRA methods in federated NLU have been mostly conducted on public datasets, lacking a comprehensive evaluation of real-world privacy datasets. To address this limitation, this paper presents a pioneering study that reexamines the performance of these DRA methods as well as corresponding defense methods. Specifically, we introduce a novel real-world privacy dataset called FedAttack which leads to a significant discovery: existing DRA methods usually fail to accurately recover the original text of real-world privacy data. In detail, the tokens within a recovery sentence are disordered and intertwined with tokens from other sentences in the same training batch. Moreover, our experiments demonstrate that the performance of DRA is also influenced by different languages and domains. By discovering these findings, our work lays a solid foundation for further research into the development of more practical DRA methods and corresponding defenses.

In this work, we introduce PRoMoTION, a unified proto-typical transformer-based framework engineered to model fundamental motion tasks. PRoMoTION offers a range of compelling attributes that set it apart from current task-specific paradigms. (1) We adopt a prototypical perspective, establishing a unified paradigm that harmonizes disparate motion learning approaches. This novel paradigm stream-lines the architectural design, enabling the simultaneous assimilation of diverse motion information. (2) We capitalize on a dual mechanism involving the feature denoiser and the prototypical learner to decipher the intricacies of motion. This approach effectively circumvents the pitfalls of ambiguity in pixel-wise feature matching, significantly bolstering the robustness of motion representation. (3)) We demon-strate a profound degree of transferability across distinct motion patterns. This inherent versatility reverberates robustly across a comprehensive spectrum of both 2D and 3D downstream tasks. Empirical results demonstrate that PRoMOTION outperforms various well-known specialized architectures, achieving 0.54 and 0.054$AbsRel$error on the Sintel and KITTI depth datasets, 1.04 and 2.01 average endpoint error on the clean and final pass of Sintel flow benchmark, and 4.30 F1-all error on the KITTI flow bench-mark. For its efficacy, we hope our work can catalyze a paradigm shift in universal models in computer vision.

The ability of large language models (LLMs) to process visual inputs has given rise to general-purpose vision systems, unifying various vision-language (VL) tasks by instruction tuning. However, due to the enormous diversity in input-output formats in the vision domain, existing general-purpose models fail to successfully integrate segmentation and multi-image inputs with coarse-level tasks into a single framework. In this work, we introduce VistaLLM, a powerful visual system that addresses coarse- and fine-grained VL tasks over single and multiple input images using a unified framework. VistaLLM utilizes an instruction-guided image tokenizer that filters global embeddings using task descriptions to extract compressed and refined features from numerous images. Moreover, VistaLLM employs a gradient-aware adaptive sampling technique to represent binary segmentation masks as sequences, significantly improving over previously used uniform sampling. To bolster the desired capability of VistaLLM, we curate CoinIt, a comprehensive coarse-to-fine instruction tuning dataset with 6.8M samples. We also address the lack of multi-image grounding datasets by introducing a novel task, AttCoSeg (Attribute-level Co-Segmentation), which boosts the model's reasoning and grounding capability over multiple input images. Extensive experiments on a wide range of V- and VL tasks demonstrate the effectiveness of VistaLLM by achieving consistent state-of-the-art performance over strong baselines across many downstream tasks. Our project page can be found at https://shramanpramanick.github.io/VistaLLM/.

Barry Yao, Sijia Wang, Yu Chen, Qifan Wang, Minqian Liu, Zhiyang Xu, Licheng Yu, Lifu Huang. Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Adapting Large Language Models (LLMs) for agent tasks is critical in developing language agents.Direct Preference Optimization (DPO) is a promising technique for this adaptation with the alleviation of compounding errors, offering a means to directly optimize Reinforcement Learning (RL) objectives.However, applying DPO to multi-turn tasks presents challenges due to the inability to cancel the partition function.Overcoming this obstacle involves making the partition function independent of the current state and addressing length disparities between preferred and dis-preferred trajectories.In this light, we replace the policy constraint with the state-action occupancy measure constraint in the RL objective and add length normalization to the Bradley-Terry model, yielding a novel loss function named DMPO for multi-turn agent tasks with theoretical explanations.Extensive experiments on three multi-turn agent task datasets confirm the effectiveness and superiority of the DMPO loss.

Taowen Wang, Yiyang Liu, James Chenhao Liang, Junhan Zhao, Yiming Cui, Yuning Mao, Shaoliang Nie, Jiahao Liu, Fuli Feng, Zenglin Xu, Cheng Han, Lifu Huang, Qifan Wang, Dongfang Liu. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

Masked Language Modeling (MLM) has been one of the most prominent approaches for pretraining bidirectional text encoders due to its simplicity and effectiveness. One notable concern about MLM is that the special $\texttt{[MASK]}$ symbol causes a discrepancy between pretraining data and downstream data as it is present only in pretraining but not in fine-tuning. In this work, we offer a new perspective on the consequence of such a discrepancy: We demonstrate empirically and theoretically that MLM pretraining allocates some model dimensions exclusively for representing $\texttt{[MASK]}$ tokens, resulting in a representation deficiency for real tokens and limiting the pretrained model's expressiveness when it is adapted to downstream data without $\texttt{[MASK]}$ tokens. Motivated by the identified issue, we propose MAE-LM, which pretrains the Masked Autoencoder architecture with MLM where $\texttt{[MASK]}$ tokens are excluded from the encoder. Empirically, we show that MAE-LM improves the utilization of model dimensions for real token representations, and MAE-LM consistently outperforms MLM-pretrained models on the GLUE and SQuAD benchmarks.

This paper introduces the Fair Fairness Benchmark (FFB), a benchmarking framework for in-processing group fairness methods. Ensuring fairness in machine learning is important for ethical compliance. However, there exist challenges in comparing and developing fairness methods due to inconsistencies in experimental settings, lack of accessible algorithmic implementations, and limited extensibility of current fairness packages and tools. To address these issues, we introduce an open-source standardized benchmark for evaluating in-processing group fairness methods and provide a comprehensive analysis of state-of-the-art methods to ensure different notions of group fairness. This work offers the following key contributions: the provision of flexible, extensible, minimalistic, and research-oriented open-source code; the establishment of unified fairness method benchmarking pipelines; and extensive benchmarking, which yields key insights from 45,079 experiments, 14,428 GPU hours. We believe that our work will significantly facilitate the growth and development of the fairness research community. The benchmark is available at https://github.com/ahxt/fair_fairness_benchmark.

We introduce the novel Diffusion Visual Programmer (DVP), a neuro-symbolic image translation framework. Our proposed DVP seamlessly embeds a condition-flexible diffusion model within the GPT architecture, orchestrating a coherent sequence of visual programs ($i.e.$, computer vision models) for various pro-symbolic steps, which span RoI identification, style transfer, and position manipulation, facilitating transparent and controllable image translation processes. Extensive experiments demonstrate DVP’s remarkable performance, surpassing concurrent arts. This success can be attributed to several key features of DVP: First, DVP achieves condition-flexible translation via instance normalization, enabling the model to eliminate sensitivity caused by the manual guidance and optimally focus on textual descriptions for high-quality content generation. Second, the frame work enhances in-context reasoning by deciphering intricate high-dimensional concepts in feature spaces into more accessible low-dimensional symbols ($e.g.$, [Prompt], [RoI object]), allowing for localized, context-free editing while maintaining overall coherence. Last but not least, DVP improves systemic controllability and explainability by offering explicit symbolic representations at each programming stage, empowering users to intuitively interpret and modify results. Our research marks a substantial step towards harmonizing artificial image translation processes with cognitive intelligence, promising broader applications.

As the scale of vision models continues to grow, the emergence of Visual Prompt Tuning (VPT) as a parameter-efficient transfer learning technique has gained attention due to its superior performance compared to traditional full-finetuning. However, the conditions favoring VPT (the "when") and the underlying rationale (the "why") remain unclear. In this paper, we conduct a comprehensive analysis across 19 distinct datasets and tasks. To understand the "when" aspect, we identify the scenarios where VPT proves favorable by two dimensions: task objectives and data distributions. We find that VPT is preferrable when there is 1) a substantial disparity between the original and the downstream task objectives ($e.g.$, transitioning from classification to counting), or 2) a notable similarity in data distributions between the two tasks ($e.g.$, both involve natural images). In exploring the "why" dimension, our results indicate VPT's success cannot be attributed solely to overfitting and optimization considerations. The unique way VPT preserves original features and adds parameters appears to be a pivotal factor. Our study provides insights into VPT's mechanisms, and offers guidance for its optimal utilization.

In this work, we introduce the Prototypical Transformer (ProtoFormer), a general and unified framework that approaches various motion tasks from a prototype perspective. ProtoFormer seamlessly integrates prototype learning with Transformer by thoughtfully considering motion dynamics, introducing two innovative designs. First, Cross-Attention Prototyping discovers prototypes based on signature motion patterns, providing transparency in understanding motion scenes. Second, Latent Synchronization guides feature representation learning via prototypes, effectively mitigating the problem of motion uncertainty. Empirical results demonstrate that our approach achieves competitive performance on popular motion tasks such as optical flow and scene depth. Furthermore, it exhibits generality across various downstream tasks, including object tracking and video stabilization.

states without platform intervention; 2. offline experiments employing our proposed intervention mechanisms on diverse datasets; and 3. results from a three-week online experiment conducted on Instagram Reels short-video recommendation platform.

Song Jiang, Zahra Shakeri, Aaron Chan, Maziar Sanjabi, Hamed Firooz, Yinglong Xia, Bugra Akyildiz, Yizhou Sun, Jinchao Li, Qifan Wang, Asli Celikyilmaz. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Suyu Ge, Chunting Zhou, Rui Hou, Madian Khabsa, Yi-Chia Wang, Qifan Wang, Jiawei Han, Yuning Mao. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Chi Han, Qifan Wang, Hao Peng, Wenhan Xiong, Yu Chen, Heng Ji, Sinong Wang. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

On User-Generated Content (UGC) platforms, recommendation algorithms significantly impact creators' motivation to produce content as they compete for algorithmically allocated user traffic. This phenomenon subtly shapes the volume and diversity of the content pool, which is crucial for the platform's sustainability. In this work, we demonstrate, both theoretically and empirically, that a purely relevance-driven policy with low exploration strength boosts short-term user satisfaction but undermines the long-term richness of the content pool. In contrast, a more aggressive exploration policy may slightly compromise user satisfaction but promote higher content creation volume. Our findings reveal a fundamental trade-off between immediate user satisfaction and overall content production on UGC platforms. Building on this finding, we propose an efficient optimization method to identify the optimal exploration strength, balancing user and creator engagement. Our model can serve as a pre-deployment audit tool for recommendation algorithms on UGC platforms, helping to align their immediate objectives with sustainable, long-term goals.

With the scale of vision Transformer-based models continuing to grow, finetuning these large-scale pretrained models for new tasks has become increasingly parameter-intensive. Visual prompt tuning is introduced as a parameter-efficient finetuning (PEFT) method to this trend. Despite its successes, a notable research challenge persists within almost all PEFT approaches: significant performance degradation is observed when there is a substantial disparity between the datasets applied in pretraining and finetuning phases. To address this challenge, we draw inspiration from human visual cognition, and propose the Visual Fourier Prompt Tuning (VFPT) method as a general and effective solution for adapting large-scale transformer-based models. Our approach innovatively incorporates the Fast Fourier Transform into prompt embeddings and harmoniously considers both spatial and frequency domain information. Apart from its inherent simplicity and intuitiveness, VFPT exhibits superior performance across all datasets, offering a general solution to dataset challenges, irrespective of data disparities. Empirical results demonstrate that our approach outperforms current state-of-the-art baselines on two benchmarks, with low parameter usage (e.g., 0.57% of model parameters on VTAB-1k) and notable performance enhancements (e.g., 73.20% of mean accuracy on VTAB-1k). Our code is avaliable at https://github.com/runtsang/VFPT.

Personalization has emerged as a critical research area in modern intelligent systems, focusing on mining users' behavioral history and adapting to their preferences for delivering tailored experiences. Despite the remarkable few-shot capabilities exhibited by black-box large language models (LLMs), the inherent opacity of their model parameters presents significant challenges in aligning the generated output with individual expectations. Existing solutions have primarily focused on prompt design to incorporate user-specific profiles and behaviors; however, such approaches often struggle to generalize effectively due to their inability to capture shared knowledge among all users. To address these challenges, we propose HYDRA, a model factorization framework that captures both user-specific behavior patterns from historical data and shared general knowledge among all users to deliver personalized generation. In order to capture user-specific behavior patterns, we first train a reranker to prioritize the most useful information from top-retrieved relevant historical records. By combining the prioritized history with the corresponding query, we train an adapter to align the output with individual user-specific preferences, eliminating the reliance on access to inherent model parameters of black-box LLMs. Both the reranker and the adapter can be decomposed into a base model with multiple user-specific heads, resembling a hydra. The base model maintains shared knowledge across users, while the multiple personal heads capture user-specific preferences. Experimental results demonstrate that \method outperforms existing state-of-the-art prompt-based methods by an average relative improvement of 9.01% across five diverse personalization tasks in the LaMP benchmark.

Planning for both immediate and long-term benefits becomes increasingly important in recommendation. Existing methods apply Reinforcement Learning (RL) to learn planning capacity by maximizing cumulative reward for long-term recommendation. However, the scarcity of recommendation data presents challenges such as instability and susceptibility to overfitting when training RL models from scratch, resulting in sub-optimal performance. In this light, we propose to leverage the remarkable planning capabilities over sparse data of Large Language Models (LLMs) for long-term recommendation. The key to achieving the target lies in formulating a guidance plan following principles of enhancing long-term engagement and grounding the plan to effective and executable actions in a personalized manner. To this end, we propose a Bi-level Learnable LLM Planner framework, which consists of a set of LLM instances and breaks down the learning process into macro-learning and micro-learning to learn macro-level guidance and micro-level personalized recommendation policies, respectively. Extensive experiments validate that the framework facilitates the planning ability of LLMs for long-term recommendation. Our code and data can be found at https://github.com/jizhi-zhang/BiLLP.

Monocular Depth Estimation (MDE) plays a vital role in applications such as autonomous driving. However, various attacks target MDE models, with physical attacks posing significant threats to system security. Traditional adversarial training methods, which require ground-truth labels, are not directly applicable to MDE models that lack ground-truth depth. Some self-supervised model hardening techniques (e.g., contrastive learning) overlook the domain knowledge of MDE, resulting in suboptimal performance. In this work, we introduce a novel self-supervised adversarial training approach for MDE models, leveraging view synthesis without the need for ground-truth depth. We enhance adversarial robustness against real-world attacks by incorporating$L_{0}$-norm-bounded perturbation during training. We evaluate our method against supervised learning-based and contrastive learning-based approaches specifically designed for MDE. Our experiments with two representative MDE networks demonstrate improved robustness against various adversarial attacks, with minimal impact on benign performance.

Optical flow is an indispensable building block for various important computer vision tasks, including motion estimation, object tracking, and disparity measurement. To date, the dominant methods are CNN-based, leaving plenty of room for improvement. In this work, we propose TransFlow, a transformer architecture for optical flow estimation. Compared to dominant CNN-based methods, TransFlow demonstrates three advantages. First, it provides more accurate correlation and trustworthy matching in flow estimation by utilizing spatial self-attention and cross-attention mechanisms between adjacent frames to effectively capture global dependencies; Second, it recovers more compromised information (e.g., occlusion and motion blur) in flow estimation through long-range temporal association in dynamic scenes; Third, it introduces a concise self-learning paradigm, eliminating the need for complex and laborious multi-stage pre-training procedures. The versatility and superiority of TransFlow extend seamlessly to 3D scene motion, yielding competitive outcomes in 3D scene flow estimation. Our approach attains state-of-the-art results on benchmark datasets such as Sintel and KITTI-15, while also exhibiting exceptional performance on downstream tasks, including video object detection using the ImageNet VID dataset, video frame interpolation using the GoPro dataset, and video stabilization using the DeepStab dataset. We believe that the effectiveness of TransFlow positions it as a flexible baseline for both optical flow and scene flow estimation, offering promising avenues for future research and development.

Most previous sequential labeling models are task-specific, while recent years have witnessed the rise of generative models due to the advantage of unifying all named entity recognition (NER) tasks into the encoder-decoder framework. Although achieving promising performance, our pilot studies demonstrate that existing generative models are ineffective at detecting entity boundaries and estimating entity types. In this paper, we propose a multi-task Transformer, which incorporates an entity boundary detection task into the named entity recognition task. More concretely, we achieve entity boundary detection by classifying the relations between tokens within the sentence. To improve the accuracy of entity-type mapping during decoding, we adopt an external knowledge base to calculate the prior entity-type distributions and then incorporate the information into the model via the self- and cross-attention mechanisms. We perform experiments on extensive NER benchmarks, including flat, nested, and discontinuous NER datasets involving long entities. It substantially increases nearly$+0.3 \sim +1.5\;{F_1}$scores across a broad spectrum or performs closely to the best generative NER model. Experimental results show that our approach improves the performance of the generative NER model considerably.

Predicting personality traits based on online posts has emerged as an important task in many fields such as social network analysis. One of the challenges of this task is assembling information from various posts into an overall profile for each user. While many previous solutions simply concatenate the posts into a long text and then encode the text by sequential or hierarchical models, they introduce unwarranted orders for the posts, which may mislead the models. In this paper, we propose a dynamic deep graph convolutional network (D-DGCN) to overcome the above limitation. Specifically, we design a learn-to-connect approach that adopts a dynamic multi-hop structure instead of a deterministic structure, and combine it with the DGCN module to automatically learn the connections between posts. The modules of post encoder, learn-to-connect, and DGCN are jointly trained in an end-to-end manner. Experimental results on the Kaggle and Pandora datasets show the superior performance of D-DGCN to state-of-the-art baselines. Our code is available at https://github.com/djz233/D-DGCN.

Brihi Joshi, Ziyi Liu, Sahana Ramnath, Aaron Chan, Zhewei Tong, Shaoliang Nie, Qifan Wang, Yejin Choi, Xiang Ren. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Chinese Spelling Correction (CSC) aims to detect and correct erroneous characters in Chinese texts.Although efforts have been made to introduce phonetic information (Hanyu Pinyin) in this task, they typically merge phonetic representations with character representations, which tends to weaken the representation effect of normal texts.In this work, we propose to disentangle the two types of features to allow for direct interaction between textual and phonetic information.To learn useful phonetic representations, we introduce a pinyin-to-character objective to ask the model to predict the correct characters based solely on phonetic information, where a separation mask is imposed to disable attention from phonetic input to text.To avoid overfitting the phonetics, we further design a self-distillation module to ensure that semantic information plays a major role in the prediction.Extensive experiments on three CSC benchmarks demonstrate the superiority of our method in using phonetic information 1 .

Jiduan Liu, Jiahao Liu, Qifan Wang, Jingang Wang, Wei Wu, Yunsen Xian, Dongyan Zhao, Kai Chen, Rui Yan. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Qifan Wang, Jingang Wang, Xiaojun Quan, Fuli Feng, Zenglin Xu, Shaoliang Nie, Sinong Wang, Madian Khabsa, Hamed Firooz, Dongfang Liu. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Knowledge distillation has attracted a great deal of interest recently to compress pre-trained language models.However, existing knowledge distillation methods suffer from two limitations.First, the student model simply imitates the teacher's behavior while ignoring the underlying reasoning.Second, these methods usually focus on the transfer of sophisticated model-specific knowledge but overlook dataspecific knowledge.In this paper, we present a novel attribution-driven knowledge distillation approach, which explores the token-level rationale behind the teacher model based on Integrated Gradients (IG) and transfers attribution knowledge to the student model.To enhance the knowledge transfer of model reasoning and generalization, we further explore multi-view attribution distillation on all potential decisions of the teacher.Comprehensive experiments are conducted with BERT on the GLUE benchmark.The experimental results demonstrate the superior performance of our approach to several state-of-the-art methods.

Optical flow is an indispensable building block for various important computer vision tasks, including motion estimation, object tracking, and disparity measurement. In this work, we propose TransFlow, a pure transformer architecture for optical flow estimation. Compared to dominant CNN-based methods, TransFlow demonstrates three advantages. First, it provides more accurate correlation and trustworthy matching in flow estimation by utilizing spatial self-attention and crossattention mechanisms between adjacent frames to effectively capture global dependencies; Second, it recovers more compromised information (e.g., occlusion and motion blur) in flow estimation through long-range temporal association in dynamic scenes; Third, it enables a concise self-learning paradigm and effectively eliminate the complex and laborious multi-stage pre-training procedures. We achieve the state-of-the-art results on the Sintel, KITTI-15, as well as several downstream tasks, including video object detection, interpolation and stabilization. For its efficacy, we hope TransFlow could serve as a flexible baseline for optical flow estimation.

Recently, self-supervised learning (SSL) was shown to be vulnerable to patch-based data poisoning backdoor attacks. It was shown that an adversary can poison a small part of the unlabeled data so that when a victim trains an SSL model on it, the final model will have a back-door that the adversary can exploit. This work aims to defend self-supervised learning against such attacks. We use a three-step defense pipeline, where we first train a model on the poisoned data. In the second step, our proposed defense algorithm (PatchSearch) uses the trained model to search the training data for poisoned samples and removes them from the training set. In the third step, a final model is trained on the cleaned-up training set. Our results show that PatchSearch is an effective defense. As an example, it improves a model's accuracy on images containing the trigger from 38.2% to 63.7% which is very close to the clean model's accuracy, 64.6%. More-over, we show that PatchSearch outperforms baselines and state-of-the-art defense approaches including those using additional clean, trusted data. Our code is available at https://github.com/UCDvision/PatchSearch

Chain-of-Thought (CoT) prompting enables large language models to solve complex reasoning problems by generating intermediate steps.However, confined by its inherent singlepass and sequential generation process, CoT heavily relies on the initial decisions, causing errors in early steps to accumulate and impact the final answers.In contrast, humans adopt recursive thinking when tackling complex reasoning problems, i.e., iteratively breaking the original problem into approachable subproblems and aggregating their answers to resolve the original one.Inspired by the human cognitive process, we propose SOCRATIC QUESTIONING, a divide-and-conquer style algorithm that mimics the recursive thinking process.Specifically, SOCRATIC QUESTIONING leverages large language models to raise and answer sub-questions until collecting enough information to tackle the original question.Unlike CoT, SOCRATIC QUESTIONING explicitly navigates the thinking space, stimulates effective recursive thinking, and is more robust towards errors in the thinking process.Extensive experiments on several complex reasoning tasks, including MMLU, MATH, LogiQA, and visual question-answering demonstrate significant performance improvements over the stateof-the-art prompting methods, such as CoT, and Tree-of-Thought.The qualitative analysis clearly shows that the intermediate reasoning steps elicited by SOCRATIC QUESTIONING are similar to humans' recursively thinking process of complex reasoning problems 12 .

Efficient knowledge retrieval plays a pivotal role in ensuring the success of end-to-end taskoriented dialogue systems by facilitating the selection of relevant information necessary to fulfill user requests.However, current approaches generally integrate knowledge retrieval and response generation, which poses scalability challenges when dealing with extensive knowledge bases.Taking inspiration from open-domain question answering, we propose a retrievergenerator architecture that harnesses a retriever to retrieve pertinent knowledge and a generator to generate system responses.Due to the lack of retriever training labels, we propose relying on feedback from the generator as pseudo-labels to train the retriever.To achieve this, we introduce a dual-feedback mechanism that generates both positive and negative feedback based on the output of the generator.Our method demonstrates superior performance in task-oriented dialogue tasks, as evidenced by experimental results on three benchmark datasets.

Contrastively trained vision-language models have achieved remarkable progress in vision and language representation learning.However, recent research has highlighted severe limitations of these models in their ability to perform compositional reasoning over objects, attributes, and relations.Scene graphs have emerged as an effective way to understand images compositionally.These are graphstructured semantic representations of images that contain objects, their attributes, and relations with other objects in a scene.In this work, we consider the scene graph parsed from text as a proxy for the image scene graph and propose a graph decomposition and augmentation framework along with a coarse-to-fine contrastive learning objective between images and text that aligns sentences of various complexities to the same image.We also introduce novel negative mining techniques in the scene graph space for improving attribute binding and relation understanding.Through extensive experiments, we demonstrate the effectiveness of our approach that significantly improves attribute binding, relation understanding, systematic generalization, and productivity on multiple recently proposed benchmarks (For example, improvements up to 18% for systematic generalization, 16.5% for relation understanding over a strong baseline), while achieving similar or better performance than CLIP on various general multimodal tasks.

Qifan Wang, Yuning Mao, Jingang Wang, Hanchao Yu, Shaoliang Nie, Sinong Wang, Fuli Feng, Lifu Huang, Xiaojun Quan, Zenglin Xu, Dongfang Liu. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.

Nan Wang, Qifan Wang, Yi-Chia Wang, Maziar Sanjabi, Jingzhou Liu, Hamed Firooz, Hongning Wang, Shaoliang Nie. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.

Transformer-based models have achieved great success on sentence pair modeling tasks, such as answer selection and natural language inference (NLI).These models generally perform cross-attention over input pairs, leading to prohibitive computational costs.Recent studies propose dual-encoder and late interaction architectures for faster computation.However, the balance between the expressive of crossattention and computation speedup still needs better coordinated.To this end, this paper introduces a novel paradigm MixEncoder for efficient sentence pair modeling.MixEncoder involves a lightweight cross-attention mechanism.It avoids the repeated encoding of the same query for different candidates, thus allowing modeling the query-candidate interaction in parallel.Extensive experiments conducted on four tasks demonstrate that our Mix-Encoder can speed up sentence pairing by over 113x while achieving comparable performance as the more expensive cross-attention models.The source code is available at https: //github.com/ysngki/MixEncoder.

Named entity recognition (NER) is an important research problem in natural language processing. There are three types of NER tasks, including flat, nested and discontinuous entity recognition. Most previous sequential labeling models are task-specific, while recent years have witnessed the rising of generative models due to the advantage of unifying all NER tasks into the seq2seq model framework. Although achieving promising performance, our pilot studies demonstrate that existing generative models are ineffective at detecting entity boundaries and estimating entity types. This paper proposes a multi-task Transformer, which incorporates an entity boundary detection task into the named entity recognition task. More concretely, we achieve entity boundary detection by classifying the relations between tokens within the sentence. To improve the accuracy of entity-type mapping during decoding, we adopt an external knowledge base to calculate the prior entity-type distributions and then incorporate the information into the model via the self and cross-attention mechanisms. We perform experiments on an extensive set of NER benchmarks, including two flat, three nested, and three discontinuous NER datasets. Experimental results show that our approach considerably improves the generative NER model’s performance.

As the size of transformer-based, models continues to grow, fine-tuning these large-scale pretrained vision models for new tasks has become increasingly parameter-intensive. Parameter-efficient learning has been developed to reduce the number of tunable parameters during fine-tuning. Although these methods show promising results, there is still a significant performance gap compared to full fine-tuning. To address this challenge, we propose an Effective and Efficient Visual Prompt Tuning (E2VPT) approach for large-scale transformer-based model adaptation. Specifically, we introduce a set of learnable key-value prompts and visual prompts into self-attention and input layers, respectively, to improve the effectiveness of model fine-tuning. Moreover, we design a prompt pruning procedure to systematically prune low importance prompts while preserving model performance, which largely enhances the model’s efficiency. Empirical results demonstrate that our approach outperforms several state-of-the-art baselines on two benchmarks, with considerably low parameter usage (e.g., 0.32% of model parameters on VTAB-1k). Our code is available at https://github.com/ChengHan111/E2VPT.

Fine-tuning large vision-language models is a challenging task. Prompt tuning approaches have been introduced to learn fixed textual or visual prompts while freezing the pre-trained model in downstream tasks. Despite the effectiveness of prompt tuning, what do those learnable prompts learn remains unexplained. In this work, we explore whether prompts in the fine-tuning can learn knowledge-aware prompts from the pre-training, by designing two different sets of prompts in pre-training and fine-tuning phases respectively. Specifically, we present a Video-Language Prompt tuning (VL-Prompt) approach for video captioning, which first efficiently pre-train a video-language model to extract key information (e.g., actions and objects) with flexibly generated Knowledge-Aware Prompt (KAP). Then, we design a Video-Language Prompt (VLP) to transfer the knowledge from the knowledge-aware prompts and fine-tune the model to generate full captions. Experimental results show the superior performance of our approach over several state-of-the-art baselines. We further demonstrate that the video-language prompts are well learned from the knowledge-aware prompts.

This paper presents ClusterFormer, a universal vision model that is based on the Clustering paradigm with TransFormer. It comprises two novel designs: 1) recurrent cross-attention clustering, which reformulates the cross-attention mechanism in Transformer and enables recursive updates of cluster centers to facilitate strong representation learning; and 2) feature dispatching, which uses the updated cluster centers to redistribute image features through similarity-based metrics, resulting in a transparent pipeline. This elegant design streamlines an explainable and transferable workflow, capable of tackling heterogeneous vision tasks (i.e., image classification, object detection, and image segmentation) with varying levels of clustering granularity (i.e., image-, box-, and pixel-level). Empirical results demonstrate that ClusterFormer outperforms various well-known specialized architectures, achieving 83.41% top-1 acc. over ImageNet-1K for image classification, 54.2% and 47.0% mAP over MS COCO for object detection and instance segmentation, 52.4% mIoU over ADE20K for semantic segmentation, and 55.8% PQ over COCO Panoptic for panoptic segmentation. This work aims to initiate a paradigm shift in universal visual understanding and to benefit the broader field.

The task of point cloud segmentation, comprising semantic, instance, and panoptic segmentation, has been mainly tackled by designing task-specific network architectures, which often lack the flexibility to generalize across tasks, thus resulting in a fragmented research landscape. In this paper, we introduce ProtoSEG, a prototype-based model that unifies semantic, instance, and panoptic segmentation tasks. Our approach treats these three homogeneous tasks as a classification problem with different levels of granularity. By leveraging a Transformer architecture, we extract point embeddings to optimize prototype-class distances and dynamically learn class prototypes to accommodate the end tasks. Our prototypical design enjoys simplicity and transparency, powerful representational learning, and ad-hoc explainability. Empirical results demonstrate that ProtoSEG outperforms concurrent well-known specialized architectures on 3D point cloud benchmarks, achieving 72.3%, 76.4% and 74.2% mIoU for semantic segmentation on S3DIS, ScanNet V2 and SemanticKITTI, 66.8% mCov and 51.2% mAP for instance segmentation on S3DIS and ScanNet V2, 62.4% PQ for panoptic segmentation on SemanticKITTI, validating the strength of our concept and the effectiveness of our algorithm. The code and models are available at https://github.com/zyqin19/PROTOSEG.

The past decade has witnessed the flourishing of a new profession as media content creators, who rely on revenue streams from online content recommendation platforms. The reward mechanism employed by these platforms creates a competitive environment among creators which affects their production choices and, consequently, content distribution and system welfare. It is thus crucial to design the platform's reward mechanism in order to steer the creators' competition towards a desirable welfare outcome in the long run. This work makes two major contributions in this regard: first, we uncover a fundamental limit about a class of widely adopted mechanisms, coined \emph{Merit-based Monotone Mechanisms}, by showing that they inevitably lead to a constant fraction loss of the optimal welfare. To circumvent this limitation, we introduce \emph{Backward Rewarding Mechanisms} (BRMs) and show that the competition game resultant from BRMs possesses a potential game structure. BRMs thus naturally induce strategic creators' collective behaviors towards optimizing the potential function, which can be designed to match any given welfare metric. In addition, the class of BRM can be parameterized so that it allows the platform to directly optimize welfare within the feasible mechanism space even when the welfare metric is not explicitly defined.

Sequential recommender models typically generate predictions in a single step during testing, without considering additional prediction correction to enhance performance as humans would. To improve the accuracy of these models, some researchers have attempted to simulate human analogical reasoning to correct predictions for testing data by drawing analogies with the prediction errors of similar training data. However, there are inherent gaps between testing and training data, which can make this approach unreliable. To address this issue, we propose an Abductive Prediction Correction (APC) framework for sequential recommendation. Our approach simulates abductive reasoning to correct predictions. Specifically, we design an abductive reasoning task that infers the most probable historical interactions from the future interactions predicted by a recommender, and minimizes the discrepancy between the inferred and true historical interactions to adjust the predictions. We perform the abductive inference and adjustment using a reversed sequential model in the forward and backward propagation manner of neural networks. Our APC framework is applicable to various differentiable sequential recommender models. We implement it on three backbone models and demonstrate its effectiveness. We release the code at https://github.com/zyang1580/APC.

Recommender systems not only serve users but also affect user preferences through personalized recommendations. Recent researches investigate the effects of the entire recommender system on user preferences, i.e., system-level effects, and find that recommendations may lead to problems such as echo chambers and filter bubbles. To properly alleviate the problems, it is necessary to estimate the effects of recommending a specific item on user preferences, i.e., item-level effects. For example, by understanding whether recommending an item aggravates echo chambers, we can better decide whether to recommend it or not.

Taxonomies, which organize knowledge hierarchically, support various practical web applications such as product navigation in online shopping and user profile tagging on social platforms. Given the continued and rapid emergence of new entities, maintaining a comprehensive taxonomy in a timely manner through human annotation is prohibitively expensive. Therefore, expanding a taxonomy automatically with new entities is essential. Most existing methods for expanding taxonomies encode entities into vector embeddings (i.e., single points). However, we argue that vectors are insufficient to model the “is-a” hierarchy in taxonomy (asymmetrical relation), because two points can only represent pairwise similarity (symmetrical relation). To this end, we propose to project taxonomy entities into boxes (i.e., hyperrectangles). Two boxes can be "contained", "disjoint" and "intersecting", thus naturally representing an asymmetrical taxonomic hierarchy. Upon box embeddings, we propose a novel model BoxTaxo for taxonomy expansion. The core of BoxTaxo is to learn boxes for entities to capture their child-parent hierarchies. To achieve this, BoxTaxo optimizes the box embeddings from a joint view of geometry and probability. BoxTaxo also offers an easy and natural way for inference: examine whether the box of a given new entity is fully enclosed inside the box of a candidate parent from the existing taxonomy. Extensive experiments on two benchmarks demonstrate the effectiveness of BoxTaxo compared to vector based models.

Historical interactions are the default choice for recommender model training, which typically exhibit high sparsity, i.e., most user-item pairs are unobserved missing data. A standard choice is treating the missing data as negative training samples and estimating interaction likelihood between user-item pairs along with the observed interactions. In this way, some potential interactions are inevitably mislabeled during training, which will hurt the model fidelity, hindering the model to recall the mislabeled items, especially the long-tail ones. In this work, we investigate the mislabeling issue from a new perspective ofaleatoric uncertainty, which describes the inherent randomness of missing data. The randomness pushes us to go beyond merely the interaction likelihood and embrace aleatoric uncertainty modeling. Towards this end, we propose a newAleatoric Uncertainty-aware Recommendation(AUR) framework that consists of a new uncertainty estimator along with a normal recommender model. According to the theory of aleatoric uncertainty, we derive a new recommendation objective to learn the estimator. As the chance of mislabeling reflects the potential of a pair, AUR makes recommendations according to the uncertainty, which is demonstrated to improve the recommendation performance of less popular items without sacrificing the overall performance. We instantiate AUR on three representative recommender models: Matrix Factorization (MF), LightGCN, and VAE from mainstream model architectures. Extensive results on four real-world datasets validate the effectiveness of AUR w.r.t. better recommendation results, especially on long-tail items.

Instance segmentation in videos, which aims to segment and track multiple objects in video frames, has garnered a flurry of research attention in recent years. In this paper, we present a novel weakly supervised framework with Spatio-Temporal Collaboration for instance Segmentation in videos, namely STC-Seg. Concretely, STC-Seg demonstrates four contributions. First, we leverage the complementary representations from unsupervised depth estimation and optical flow to produce effective pseudo-labels for training deep networks and predicting high-quality instance masks. Second, to enhance the mask generation, we devise a puzzle loss, which enables end-to-end training using box-level annotations. Third, our tracking module jointly utilizes bounding-box diagonal points with spatio-temporal discrepancy to model movements, which largely improves the robustness to different object appearances. Finally, our framework is flexible and enables image-level instance segmentation methods to operate the video-level task. We conduct an extensive set of experiments on the KITTI MOTS and YT-VIS datasets. Experimental results demonstrate that our method achieves strong performance and even outperforms fully supervised TrackR-CNN and MaskTrack R-CNN. We believe that STC-Seg can be a valuable addition to the community, as it reflects the tip of an iceberg about the innovative opportunities in the weakly supervised paradigm for instance segmentation in videos.

Recent studies on Graph Convolutional Networks (GCNs) reveal that the initial node representations (i.e., the node representations before the first-time graph convolution) largely affect the final model performance. However, when learning the initial representation for a node, most existing work linearly combines the embeddings of node features, without considering the interactions among the features (or feature embeddings). We argue that when the node features are categorical, e.g., in many real-world applications like user profiling and recommender system, feature interactions usually carry important signals for predictive analytics. Ignoring them will result in suboptimal initial node representation and thus weaken the effectiveness of the follow-up graph convolution. In this paper, we propose a new GCN model named CatGCN, which is tailored for graph learning on categorical node features. Specifically, we integrate two ways of explicit interaction modeling into the learning of initial node representation, i.e., local interaction modeling on each pair of node features and global interaction modeling on an artificial feature graph. We then refine the enhanced initial node representations with the neighborhood aggregation-based graph convolution. We train CatGCN in an end-to-end fashion and demonstrate it on the task of node classification. Extensive experiments on three tasks of user profiling (the prediction of user age, city, and purchase level) from Tencent and Alibaba datasets validate the effectiveness of CatGCN, especially the positive effect of performing feature interaction modeling before graph convolution.

One of the important factors affecting micro-video recommender systems is to model the multi-modal user preference on the micro-video. Despite the remarkable performance of prior arts, they are still limited by fusing the user preference derived from different modalities in a unified manner, ignoring the users tend to place different emphasis on different modalities. Furthermore, modality-missing is ubiquity and unavoidable in the micro-video recommendation, some modalities information of micro-videos are lacked in many cases, which negatively affects the multi-modal fusion operations. To overcome these disadvantages, we propose a novel framework for the micro-video recommendation, dubbed Dual Graph Neural Network (DualGNN), upon the user-microvideo bipartite and user co-occurrence graphs, which leverages the correlation between users to collaboratively mine the particular fusion pattern for each user. Specifically, we first introduce a single-modal representation learning module, which performs graph operations on the user-microvideo graph in each modality to capture single-modal user preferences on different modalities. And then, we devise a multi-modal representation learning module to explicitly model the user’s attentions over different modalities and inductively learn the multi-modal user preference. Finally, we propose a prediction module to rank the potential micro-videos for users. Extensive experiments on two public datasets demonstrate the significant superiority of our DualGNN over state-of-the-arts methods.

Pre-trained Language Models (PLMs) have achieved remarkable performance gains across numerous downstream tasks in natural language understanding. Various Chinese PLMs have been successively proposed for learning better Chinese language representation. However, most current models use Chinese characters as inputs and are not able to encode semantic information contained in Chinese words. While recent pre-trained models incorporate both words and characters simultaneously, they usually suffer from deficient semantic interactions and fail to capture the semantic relation between words and characters. To address the above issues, we propose a simple yet effective PLM CLOWER, which adopts the Contrastive Learning Over Word and charactER representations. In particular, CLOWER implicitly encodes the coarse-grained information (i.e., words) into the fine-grained representations (i.e., characters) through contrastive learning on multi-grained information. CLOWER is of great value in realistic scenarios since it can be easily incorporated into any existing fine-grained based PLMs without modifying the production pipelines. Extensive experiments conducted on a range of downstream tasks demonstrate the superior performance of CLOWER over several state-of-the-art baselines.

Task-oriented dialog (TOD) systems are often required to interact with an external knowledge base (KB) to retrieve necessary entity (e.g., restaurants) information to support their response generation. Most current end-to-end TOD systems either retrieve the KB information explicitly or embed it into model parameters for implicit access. While the first approach demands scanning the KB at each turn of response generation, which is inefficient when the KB scales up, the second approach shows higher flexibility and efficiency. In either approach, the response shall contain attributes of the same entity, however the systems may generate a response with conflicting entities. To address this, we propose to generate the entity autoregressively before leveraging it to guide the response generation in an end-to-end system. To ensure entity consistency, we impose a trie constraint on the decoding of an entity. We also introduce a logit concatenation strategy to facilitate gradient backpropagation for end-to-end training. Experiments on MultiWOZ 2.1 single and CAMREST show that our system can generate more high-quality and entity-consistent responses in an end-to-end manner.

Text matching is a fundamental research problem in natural language understanding.Interaction-based approaches treat the text pair as a single sequence and encode it through cross encoders, while representation-based models encode the text pair independently with siamese or dual encoders.Interactionbased models require dense computations and thus are impractical in real-world applications.Representation-based models have become the mainstream paradigm for efficient text matching.However, these models suffer from severe performance degradation due to the lack of interactions between the pair of texts.To remedy this, we propose a Virtual InteRacTion mechanism (VIRT) for improving representation-based text matching while maintaining its efficiency.In particular, we introduce an interactive knowledge distillation module that is only applied during training.It enables deep interaction between texts by effectively transferring knowledge from the interaction-based model.A light interaction strategy is designed to fully leverage the learned interactive knowledge.Experimental results on six text matching benchmarks demonstrate the superior performance of our method over several state-of-the-art representationbased models.We further show that VIRT can be integrated into existing methods as plugins to lift their performances.

Prompt tuning learns soft prompts to condition the frozen Pre-trained Language Models (PLMs) for performing downstream tasks in a parameter-efficient manner.While prompt tuning has gradually reached the performance level of fine-tuning as the model scale increases, there is still a large performance gap between prompt tuning and fine-tuning for models of moderate and small scales (typically less than 11B parameters).In this paper, we empirically show that the trained prompt tokens can have a negative impact on a downstream task and thus degrade its performance.To bridge the gap, we propose a novel PROMPT tuning model with an eXtremely small scale (XPROMPT) under the regime of lottery tickets hypothesis.Specifically, XPROMPT eliminates the negative prompt tokens at different granularity levels through a hierarchical structured pruning, yielding a more parameter-efficient prompt yet with a competitive performance.Comprehensive experiments are carried out on the SuperGLUE tasks, and the results indicate that XPROMPT is able to close the performance gap at smaller model scales. 1 Recently, Prompt-Tuning (Lester et al., 2021; Liu et al., 2021b) has been proposed to address this issue by prepending a soft prompt to the input and only updating the parameters of prompt tokens during tuning.Prompt-Tuning provides a parameter-efficient alternative to fine-tuning, since the scale of the soft prompt is tens of thousand smaller.It is also conceptually simpler and more flexible than other parameter-efficient tuning methods (such as Adapters), that require intrusive modifications to transformer layers (Houlsby et al., 2019;Guo et al., 2021).Using fewer tunable parameters, prompt tuning achieves competitive performance to fine-tuning with the increase of the model scale.However, there is still a large performance gap between prompt tuning and fine-tuning for models of smaller scales (as shown in Figure 1).This paper aims to fill the gap, from the perspective of the lottery tickets hypothesis (LTH) (Frankle and Carbin, 2019).We are motivated by an observation that, on a specific task, not all prompt tokens contribute equally to the task performance, while certain prompt tokens may even bring a negative

Automatic question generation (AQG) is the task of generating a question from a given passage and an answer.Most existing AQG methods aim at encoding the passage and the answer to generate the question.However, limited work has focused on modeling the correlation between the target answer and the generated question.Moreover, unseen or rare word generation has not been studied in previous works.In this paper, we propose a novel approach which incorporates question generation with its dual problem, question answering, into a unified primal-dual framework.Specifically, the question generation component consists of an encoder that jointly encodes the answer with the passage, and a decoder that produces the question.The question answering component then re-asks the generated question on the passage to ensure that the target answer is obtained.We further introduce a knowledge distillation module to improve the model generalization ability.We conduct an extensive set of experiments on SQuAD and HotpotQA benchmarks.Experimental results demonstrate the superior performance of the proposed approach over several state-of-the-art methods.

With the necessity of privacy protection, it becomes increasingly vital to train deep neural models in a federated learning manner for natural language processing (NLP) tasks.However, recent studies show eavesdroppers (i.e., dishonest servers) can still reconstruct the private input in federated learning (FL).Such a data reconstruction attack relies on the mappings between vocabulary and associated word embedding in NLP tasks, which are unfortunately less studied in current FL methods.In this paper, we propose a fedrated model decomposition method that protects the privacy of vocabularies, shorted as FEDEVOCAB.In FEDEVOCAB, each participant keeps the local embedding layer in the local device and detaches the local embedding parameters from federated aggregation.However, it is challenging to train an accurate NLP model when the private mappings are unknown and vary across participants in a cross-device FL setting.To address this problem, we further propose an adaptive updating technique to improve the performance of local models.Experimental results show that FEDEVOCAB maintains competitive performance and provides better privacy-preserving capacity compared to status quo methods. * Co-corresponding authorHi, recently I feel very thirsty and urinate more.And I get hungry easily and eat more than usual.But being 165cm tall, my weight even dropped to 40kg.It has continued for half a month.Hi! Does your family have a history of diabetes?If so, there is a high probability of diabetes.

Video captioning is a challenging task as it needs to accurately transform visual understanding into natural language description. To date, state-of-the-art methods inadequately model global-local representation across video frames for caption generation, leaving plenty of room for improvement. In this work, we approach the video captioning task from a new perspective and propose a GL-RG framework for video captioning, namely a Global-Local Representation Granularity. Our GL-RG demonstrates three advantages over the prior efforts: 1) we explicitly exploit extensive visual representations from different video ranges to improve linguistic expression; 2) we devise a novel global-local encoder to produce rich semantic vocabulary to obtain a descriptive granularity of video contents across frames; 3) we develop an incremental training strategy which organizes model learning in an incremental fashion to incur an optimal captioning behavior. Experimental results on the challenging MSR-VTT and MSVD datasets show that our DL-RG outperforms recent state-of-the-art methods by a significant margin. Code is available at https://github.com/ylqi/GL-RG.

Fine-tuning large pre-trained language models on downstream tasks is apt to suffer from overfitting when limited training data is available. While dropout proves to be an effective antidote by randomly dropping a proportion of units, existing research has not examined its effect on the self-attention mechanism. In this paper, we investigate this problem through self-attention attribution and find that dropping attention positions with low attribution scores can accelerate training and increase the risk of overfitting. Motivated by this observation, we propose Attribution-Driven Dropout (AD-DROP), which randomly discards some high-attribution positions to encourage the model to make predictions by relying more on low-attribution positions to reduce overfitting. We also develop a cross-tuning strategy to alternate fine-tuning and AD-DROP to avoid dropping high-attribution positions excessively. Extensive experiments on various benchmarks show that AD-DROP yields consistent improvements over baselines. Analysis further confirms that AD-DROP serves as a strategic regularizer to prevent overfitting during fine-tuning.

Attribute value extraction refers to the task of identifying values of an attribute of interest from product information. Product attribute values are essential in many e-commerce scenarios, such as customer service robots, product ranking, retrieval and recommendations. While in the real world, the attribute values of a product are usually incomplete and vary over time, which greatly hinders the practical applications. In this paper, we introduce MAVE, a new dataset to better facilitate research on product attribute value extraction. MAVE is composed of a curated set of 2.2 million products from Amazon pages, with 3 million attribute-value annotations across 1257 unique categories. MAVE has four main and unique advantages: First, MAVE is the largest product attribute value extraction dataset by the number of attribute-value examples. Second, MAVE includes multi-source representations from the product, which captures the full product information with high attribute coverage. Third, MAVE represents a more diverse set of attributes and values relative to what previous datasets cover. Lastly, MAVE provides a very challenging zero-shot test set, as we empirically illustrate in the experiments. We further propose a novel approach that effectively extracts the attribute value from the multi-source product information. We conduct extensive experiments with several baselines and show that MAVE is an effective dataset for attribute value extraction task. It is also a very challenging task on zero-shot attribute extraction. Data is available at \urlhttps://github.com/google-research-datasets/MAVE .

Structure information extraction refers to the task of extracting structured text fields from web pages, such as extracting a product offer from a shopping page including product title, description, brand and price. It is an important research topic which has been widely studied in document understanding and web search. Recent natural language models with sequence modeling have demonstrated state-of-the-art performance on web information extraction. However, effectively serializing tokens from unstructured web pages is challenging in practice due to a variety of web layout patterns. Limited work has focused on modeling the web layout for extracting the text fields. In this paper, we introduce WebFormer, a Web-page transFormer model for structure information extraction from web documents. First, we design HTML tokens for each DOM node in the HTML by embedding representations from their neighboring tokens through graph attention. Second, we construct rich attention patterns between HTML tokens and text tokens, which leverages the web layout for effective attention weight computation. We conduct an extensive set of experiments on SWDE and Common Crawl benchmarks. Experimental results demonstrate the superior performance of the proposed approach over several state-of-the-art methods.

Video captioning is a challenging task as it needs to accurately transform visual understanding into natural language description. To date, state-of-the-art methods inadequately model global-local vision representation for sentence generation, leaving plenty of room for improvement. In this work, we approach the video captioning task from a new perspective and propose a GLR framework, namely a global-local representation granularity. Our GLR demonstrates three advantages over the prior efforts. First, we propose a simple solution, which exploits extensive vision representations from different video ranges to improve linguistic expression. Second, we devise a novel global-local encoder, which encodes different video representations including long-range, short-range and local-keyframe, to produce rich semantic vocabulary for obtaining a descriptive granularity of video contents across frames. Finally, we introduce the progressive training strategy which can effectively organize feature learning to incur optimal captioning behavior. Evaluated on the MSR-VTT and MSVD dataset, we outperform recent state-of-the-art methods including a well-tuned SA-LSTM baseline by a significant margin, with shorter training schedules. Because of its simplicity and efficacy, we hope that our GLR could serve as a strong baseline for many video understanding tasks besides video captioning. Code will be available.

Graph Convolutional Network (GCN) is an emerging technique for information retrieval (IR) applications. While GCN assumes the homophily property of a graph, real-world graphs are never perfect: the local structure of a node may contain discrepancy, e.g., the labels of a node's neighbors could vary. This pushes us to consider the discrepancy of local structure in GCN modeling. Existing work approaches this issue by introducing an additional module such as graph attention, which is expected to learn the contribution of each neighbor. However, such module may not work reliably as expected, especially when there lacks supervision signal, e.g., when the labeled data is small. Moreover, existing methods focus on modeling the nodes in the training data, and never consider the local structure discrepancy of testing nodes.

Joshua Ainslie, Santiago Ontanon, Chris Alberti, Vaclav Cvicek, Zachary Fisher, Philip Pham, Anirudh Ravula, Sumit Sanghai, Qifan Wang, Li Yang. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2020.

Attribute value extraction refers to the task of identifying values of an attribute of interest from product information. It is an important research topic which has been widely studied in e-Commerce and relation learning. There are two main limitations in existing attribute value extraction methods: scalability and generalizability. Most existing methods treat each attribute independently and build separate models for each of them, which are not suitable for large scale attribute systems in real-world applications. Moreover, very limited research has focused on generalizing extraction to new attributes.

Transformers-based models, such as BERT, have been one of the most successful deep learning models for NLP. Unfortunately, one of their core limitations is the quadratic dependency (mainly in terms of memory) on the sequence length due to their full attention mechanism. To remedy this, we propose, BigBird, a sparse attention mechanism that reduces this quadratic dependency to linear. We show that BigBird is a universal approximator of sequence functions and is Turing complete, thereby preserving these properties of the quadratic, full attention model. Along the way, our theoretical analysis reveals some of the benefits of having $O(1)$ global tokens (such as CLS), that attend to the entire sequence as part of the sparse attention mechanism. The proposed sparse attention can handle sequences of length up to 8x of what was previously possible using similar hardware. As a consequence of the capability to handle longer context, BigBird drastically improves performance on various NLP tasks such as question answering and summarization. We also propose novel applications to genomics data.

In this paper, we consider the problem of constructing a comprehensive database of events taking place around the world. Events include small hyper-local events like farmer's markets, neighborhood garage sales, as well as larger concerts and festivals. Designing a high-precision and high-recall event extractor from unstructured pages across the whole web is a challenging problem. We cannot resort overly to domain-specific strategies since it needs to work on all web pages, including on new domains; we need to account for variations in page layouts and structure across websites. Further, we need to deal with low-quality pages on the web with limited structure. We have built an ML-powered extraction system to solve this problem, using schema.org annotations as training data. Our extraction system operates in two phases. In the first phase, we generate raw event information from individual web pages. To do this, an \em event page classifier predicts if a web page contains any event information; this is then followed by a \em single/multiple classifier that decides if the page contains a single event or multiple events; the first phase concludes by applying \em event extractors that extract the key fields of a public event (the title, the date/time information, and the location information). In the second phase, we further improve the extraction quality via three novel algorithms, \em repeated patterns, \em event consolidation and \em wrapper induction, which are designed to use the raw event extractions as input and generate events whose quality is significantly higher. We evaluate our extraction models on two large scale publicly available web corpus, Common Crawl and ClueWeb12. Experimental analysis shows that our methodology achieves over 95% extraction precision and recall on both datasets.

Label propagation is a popular semi-supervised learning technique that transfers information from labeled examples to unlabeled examples through a graph. Most label propagation methods construct a graph based on example-to-example similarity, assuming that the resulting graph connects examples that share similar labels. Unfortunately, example-level similarity is sometimes badly defined. For instance, two images may contain two different objects, but have similar overall appearance due to large similar background. In this case, computing similarities based on whole-image would fail propagating information to the right labels. This paper proposes a novel Instance-Level Label Propagation (ILLP) approach that integrates label propagation with multi-instance learning. Each example is treated as containing multiple instances, as in the case of an image consisting of multiple regions. We first construct a graph based on instance-level similarity and then simultaneously identify the instances carrying the labels and propagate the labels across instances in the graph. Optimization is based on an iterative Expectation Maximization (EM) algorithm. Experimental results on two benchmark datasets demonstrate the effectiveness of the proposed approach over several state-of-the-art methods.

Query expansion (QE) is a well known technique to improve retrieval effectiveness, which expands original queries with extra terms that are predicted to be relevant. A recent trend in the literature is Supervised Query Expansion (SQE), where supervised learning is introduced to better select expansion terms. However, an important but neglected issue for SQE is its efficiency, as applying SQE in retrieval can be much more time-consuming than applying Unsupervised Query Expansion (UQE) algorithms. In this paper, we point out that the cost of SQE mainly comes from term feature extraction, and propose a Two-stage Feature Selection framework (TFS) to address this problem. The first stage is adaptive expansion decision, which determines if a query is suitable for SQE or not. For unsuitable queries, SQE is skipped and no term features are extracted at all, which reduces the most time cost. For those suitable queries, the second stage is cost constrained feature selection, which chooses a subset of effective yet inexpensive features for supervised learning. Extensive experiments on four corpora (including three academic and one industry corpus) show that our TFS framework can substantially reduce the time cost for SQE, while maintaining its effectiveness.

Wiberg matrix factorization breaks a matrix Y into low-rank factors U and V by solving for V in closed form given U, linearizing V(U) about U, and iteratively minimizing ||Y - UV(U)||2 with respect to U only. This approach factors the matrix while effectively removing V from the minimization. Recently Eriksson and van den Hengel extended this approach to L1 , minimizing ||Y - UV(U)||1 . We generalize their approach beyond factorization to minimize ||Y - f(U, V)||1 for more general functions f(U, V) that are nonlinear in each of two sets of variables. We demonstrate the idea with a practical Wiberg algorithm for L1 bundle adjustment. One Wiberg minimization can be nested inside another, effectively removing two of three sets of variables from a minimization. We demonstrate this idea with a nested Wiberg algorithm for L1 projective bundle adjustment, solving for camera matrices, points, and projective depths. Wiberg minimization also generalizes to handle nonlinear constraints, and we demonstrate this idea with Constrained Wiberg Minimization for Multiple Instance Learning (CWM-MIL), which removes one set of variables from the constrained optimization. Our experiments emphasize isolating the effect of Wiberg by comparing against the algorithm it modifies, successive linear programming.

As one of the most important state-of-the-art classification techniques, Support Vector Machine (SVM) has been widely adopted in many real-world applications, such as object detection, face recognition, text categorization, etc., due to its competitive practical performance and elegant theoretical interpretation. However, it treats all samples independently, and ignores the fact that, in many real situations especially when data are in high dimensional space, samples typically lie on low dimensional manifolds of the feature space and thus a sample can be related to its neighbors by being represented as a linear combination of other samples on the same manifold. This linear representation, which is usually sparse, reflects the structure of underlying manifolds. It has been extensively explored in the recent literature and proven to be critical for the performance of classification. To benefit from both the underlying low dimensional manifold structure and the large margin classifier, this paper proposes a novel method called Sparsity Preserving Support Vector Machine(SP-SVM), which explicitly considers the sparse representation of samples while maximizing the margin between different classes. Consequently, SP-SVM inherits both the discriminative power of support vector machine and the merits of sparsity. A set of experiments on real-world benchmark data sets show that SP-SVM achieves significantly higher precision on recognition task than various competitive baselines including the traditional SVM, the sparse representation based method and the classical nearest neighbor classifier.

Hashing techniques have been widely applied for large scale similarity search problems due to the computational and memory efficiency.However, most existing hashing methods assume data examples are independently and identically distributed.But there often exists various additional dependency/structure information between data examplesin many real world applications. Ignoring this structure information may limit theperformance of existing hashing algorithms.This paper explores the research problemof learning to Hash on Structured Data (HSD) and formulates anovel framework that considers additional structure information.In particular, the hashing function is learned in a unified learning framework by simultaneously ensuring the structural consistency and preserving the similarities between data examples.An iterative gradient descent algorithm is designed as the optimization procedure. Furthermore, we improve the effectiveness of hashing function through orthogonal transformation by minimizing the quantization error.Experimentalresults on two datasets clearly demonstrate the advantages ofthe proposed method over several state-of-the-art hashing methods.

Currently cyber infrastructures are facing increasingly stealthy attacks that implant malicious payloads under the cover of benign programs. Existing attack detection approaches based on statistical learning methods may generate misleading decision boundaries when processing noisy data with such a mixture of benign and malicious behaviors. On the other hand, attack detection based on formal program analysis may lack completeness or adaptivity when modelling attack behaviors. In light of these limitations, we have developed LEAPS, an attack detection system based on supervised statistical learning to classify benign and malicious system events. Furthermore, we leverage control flow graphs inferred from the system event logs to enable automatic pruning of the training data, which leads to a more accurate classification model when applied to the testing data. Our extensive evaluation shows that, compared with pure statistical learning models, LEAPS achieves consistently higher accuracy when detecting real-world camouflaged attacks with benign program cover-up.

This paper studies Cumulative Citation Recommendation (CCR) for Knowledge Base Acceleration (KBA). The CCR task aims to detect potential citations of a set of target entities with priorities from a volume of temporally-ordered stream corpus. Previous approaches for CCR that build an individual relevance model for each entity fail to handle unseen entities without annotation. A baseline solution is to build a global entity-unspecific model for all entities regardless of the relationship information among entities, which cannot guarantee to achieve satisfactory result for each entity. In this paper, we propose a novel entity class-dependent discriminative mixture model by introducing a latent entity class layer to model the correlations between entities and latent entity classes. The model can better adjust to different types of entities and achieve better performance when dealing with a broad range of entities. An extensive set of experiments has been conducted on TREC-KBA-2013 dataset, and the experimental results demonstrate that the proposed model can achieve the state-of-the-art performance.

Sentiment analysis of such opinionated online texts as reviews and comments has received increasingly close attention, yet most of the work is intended to deal with the detection of authors’ emotion. In contrast, this article presents our study of the social emotion detection problem, the objective of which is to identify the evoked emotions of readers by online documents such as news articles. A novel Latent Discriminative Model (LDM) is proposed for this task. LDM works by introducing intermediate hidden variables to model the latent structure of input text corpora. To achieve this, it defines a joint distribution over emotions and latent variables, conditioned on the observed text documents. Moreover, we assume that social emotions are not independent but correlated with one another, and the dependency of them is capable of providing additional guidance to LDM in the training process. The inclusion of this emotional dependency into LDM gives rise to a new Emotional Dependency-based LDM (eLDM). We evaluate the proposed models through a series of empirical evaluations on two real-world corpora of news articles. Experimental results verify the effectiveness of LDM and eLDM in social emotion detection.

Multiple Instance Learning (MIL) is a popular learning technique in various vision tasks including image classification. However, most existing MIL methods do not consider the problem of insufficient examples in the given target category. In this case, it is difficult for traditional MIL methods to build an accurate classifier due to the lack of training examples. Motivated by the empirical success of transfer learning, this paper proposes a novel approach of Adaptive Knowledge Transfer for Multiple Instance Learning (AKT-MIL) in image classification. The new method transfers cross-category knowledge from source categories under multiple instance setting for boosting the learning process. A unified learning framework with a data-dependent mixture model is designed to adaptively combine the transferred knowledge from sources with a weak classifier built in the target domain. Based on this framework, an iterative coordinate descent method with Constraint Concave-Convex Programming (CCCP) is proposed as the optimization procedure. An extensive set of experimental results demonstrate that the proposed AKT-MIL approach substantially outperforms several state-of-the-art algorithms on two benchmark datasets, especially in the scenario when very few training examples are available in the target domain.

Similarity search, or finding approximate nearest neighbors, is an important technique in various large scale information retrieval applications such as document retrieval. Many recent research demonstrate that hashing methods can achieve promising results for large scale similarity search due to its computational and memory efficiency. However, most existing hashing methods ignore the hidden semantic structure of documents but only use the keyword features (e.g., tf-idf) in hashing codes learning. This paper proposes a novel sparse semantic hashing (SpSH) approach that explores the hidden semantic representation of documents in learning their corresponding hashing codes. In particular, a unified framework is designed for ensuring the hidden semantic structure among the documents by a sparse coding model, while at the same time preserving the document similarity via graph Laplacian. An iterative coordinate descent procedure is then proposed for solving the optimization problem. Extensive experiments on two large scale datasets demonstrate the superior performance of the proposed research over several state-of-the-art hashing methods.

Nonnegative Matrix Factorization (NMF) is a widely used technique in many applications such as face recognition, motion segmentation, etc. It approximates the nonnegative data in an original high dimensional space with a linear representation in a low dimensional space by using the product of two nonnegative matrices. In many applications data are often partially corrupted with large additive noise. When the positions of noise are known, some existing variants of N-MF can be applied by treating these corrupted entries as missing values. However, the positions are often unknown in many real world applications, which prevents the usage of traditional NMF or other existing variants of NMF. This paper proposes a Robust Nonnegative Matrix Factorization (RobustNMF) algorithm that explicitly models the partial corruption as large additive noise without requiring the information of positions of noise. In particular, the proposed method jointly approximates the clean data matrix with the product of two nonnegative matrices and estimates the positions and values of outliers/noise. An efficient iterative optimization algorithm with a solid theoretical justification has been proposed to learn the desired matrix factorization. Experimental results demonstrate the advantages of the proposed algorithm.

Similarity search is an important problem in many large scale applications such as image and text retrieval. Hashing method has become popular for similarity search due to its fast search speed and low storage cost. Recent research has shown that hashing quality can be dramatically improved by incorporating supervised information, e.g. semantic tags/labels, into hashing function learning. However, most existing supervised hashing methods can be regarded as passive methods, which assume that the labeled data are provided in advance. But in many real world applications, such supervised information may not be available.

Recommender systems usually need to compare a large number of items before users' most preferred ones can be found This process can be very costly if recommendations are frequently made on large scale datasets. In this paper, a novel hashing algorithm, named Preference Preserving Hashing (PPH), is proposed to speed up recommendation. Hashing has been widely utilized in large scale similarity search (e.g. similar image search), and the search speed with binary hashing code is significantly faster than that with real-valued features. However, one challenge of applying hashing to recommendation is that, recommendation concerns users' preferences over items rather than their similarities. To address this challenge, PPH contains two novel components that work with the popular matrix factorization (MF) algorithm. In MF, users' preferences over items are calculated as the inner product between the learned real-valued user/item features. The first component of PPH constrains the learning process, so that users' preferences can be well approximated by user-item similarities. The second component, which is a novel quantization algorithm,generates the binary hashing code from the learned real-valued user/item features. Finally, recommendation can be achieved efficiently via fast hashing code search. Experiments on three real world datasets show that the recommendation speed of the proposed PPH algorithm can be hundreds of times faster than original MF with real-valued features, and the recommendation accuracy is significantly better than previous work of hashing for recommendation.

In many online news services, users often write comments towards news in subjective emotions such as sadness, happiness or anger. Knowing such emotions can help understand the preferences and perspectives of individual users, and therefore may facilitate online publishers to provide more relevant services to users. Although building emotion classifiers is a practical task, it highly depends on sufficient training data that is not easy to be collected directly and the manually labeling work of comments can be quite labor intensive. Also, online news has different domains, which makes the problem even harder as different word distributions of the domains require different classifiers with corresponding distinct training data.

Tags have been popularly utilized in many applications with image and text data for better managing, organizing and searching for useful information. Tag completion provides missing tag information for a set of existing images or text documents while tag prediction recommends tag information for any new image or text document. Valuable prior research has focused on improving the accuracy of tag completion and prediction, but limited research has been conducted for the efficiency issue in tag completion and prediction, which is a critical problem in many large scale real world applications.

Similarity search, or finding approximate nearest neighbors, is an important technique for many applications. Many recent research demonstrate that hashing methods can achieve promising results for large scale similarity search due to its computational and memory efficiency.

It is an important research problem to design efficient and effective solutions for large scale similarity search. One popular strategy is to represent data examples as compact binary codes through semantic hashing, which has produced promising results with fast search speed and low storage cost. Many existing semantic hashing methods generate binary codes for documents by modeling document relationships based on similarity in a keyword feature space. Two major limitations in existing methods are: (1) Tag information is often associated with documents in many real world applications, but has not been fully exploited yet; (2) The similarity in keyword feature space does not fully reflect semantic relationships that go beyond keyword matching.