Large language models (LLMs) are posttrained through reinforcement learning (RL) to evolve into Reasoning Language Models (RLMs), where the hallmark of this advanced reasoning is "aha" moments when they start to perform strategies, such as self-reflection and deep thinking, within chain of thoughts (CoTs).Motivated by this, this paper proposes a novel reinforced strategy injection mechanism (rSIM), that enables any LLM to become an RLM by employing a small planner to guide the LLM's CoT through the adaptive injection of reasoning strategies.To achieve this, the planner (leader agent) is jointly trained with an LLM (follower agent) using multiagent RL (MARL), based on a leader-follower framework and straightforward rule-based rewards.Experimental results show that rSIM enables Qwen2.5-0.5B to become an RLM and significantly outperform Qwen2.5-14Bacross mathematical, coding, and financial reasoning tasks.Moreover, the planner is generalizable: it only needs to be trained once and can be applied as a plug-in to substantially improve the reasoning capabilities of existing LLMs.In addition, the planner supports continual learning across various tasks, allowing its planning abilities to gradually improve and generalize to a wider range of problems.

Test-Time Scaling (TTS) methods for enhancing Large Language Model (LLM) reasoning often incur substantial inference costs, due to reliance on long chain-of-thought (CoT) generation, self-consistency sampling methods, or searching under Process Reward Models (PRMs).This paper introduces Guided by Gut (GG), an efficient self-guided TTS framework that enables LLMs to perform step-by-step reasoning at a low cost, without any reward models or verifiers.GG performs a lightweight tree search guided solely by intrinsic confidence signals of the LLM at each reasoning step and improves the reliability of such internal confidence signals by reinforcement learning.Empirical evaluations on challenging mathematical reasoning benchmarks demonstrate that GG enables smaller models (e.g., 1.5B-7B parameters) to achieve accuracy matching or surpassing significantly larger models (e.g., 32B-70B parameters), while reducing GPU memory usage by up to 10×.Compared to TTS with PRMs, GG achieves comparable accuracy with 8× faster inference speeds and 4-5× lower memory usage.Additionally, GG reduces KV cache memory usage by approximately 50% compared to Best-of-N sampling, facilitating more efficient and practical deployment of TTS techniques.

In this work, we propose an all-in-one video restoration framework that grounds degradation-aware semantic context of video frames in natural language via foundation models, offering interpretable and flexible guidance. Unlike prior art, our method assumes no degradation knowledge in train or test time and learns an approximation to the grounded knowledge such that the foundation model can be safely disentangled during inference adding no extra cost. Further, we call for standardization of benchmarks in all-in-one video restoration, and propose two benchmarks in multi-degradation setting, three-task (3D) and four-task (4D), and two time-varying composite degradation benchmarks; one of the latter being our proposed dataset with varying snow intensity, simulating how weather degradations affect videos naturally. We compare our method with prior works and report state-of-the-art performance on all benchmarks.

This article proposes an integrated fault-tolerant control architecture for unidirectional constrained control of elastic cable-driven parallel robots (CDPRs). In CDPRs, the cable tension can only be positive, requiring unidirectional control effort. In the proposed structure, the positive tension distribution (PTD) of the elastic CDPR is ensured using a prevention function. Compared to existing methods for elastic CDPRs, this new technique has significantly lower computation time, which is essential for real-time applications. Moreover, two adaptive observers are utilized to tackle undesirable events, particularly partial failures. The stability analysis of the closed-loop system is provided using Lyapunov’s second method. The simulation study shows that the proposed method improves the tracking accuracy and is significantly faster than conventional methods. The effectiveness of the proposed method is evaluated in the presence of model uncertainties and various faulty conditions.

Recent research explores the potential of Diffusion Models (DMs) for consistent object editing, which aims to modify object position, size, and composition, etc., while preserving the consistency of objects and background without changing their texture and attributes. Current inference-time methods often rely on DDIM inversion, which inherently compromises efficiency and the achievable consistency of edited images. Recent methods also utilize energy guidance which iteratively updates the predicted noise and can drive the latents away from the original image, resulting in distortions. In this paper, we propose PixelMan, an inversion-free and training-free method for achieving consistent object editing via Pixel Manipulation and generation, where we directly create a duplicate copy of the source object at target location in the pixel space, and introduce an efficient sampling approach to iteratively harmonize the manipulated object into the target location and inpaint its original location, while ensuring image consistency by anchoring the edited image to be generated to the pixel-manipulated image as well as by introducing various consistency-preserving optimization techniques during inference. Experimental evaluations based on benchmark datasets as well as extensive visual comparisons show that in as few as 16 inference steps, PixelMan outperforms a range of state-of-the-art training-based and training-free methods (usually requiring 50 steps) on multiple consistent object editing tasks.

Diffusion Models (DM) have democratized AI image generation through an iterative denoising process. Quantization is a major technique to alleviate the inference cost and reduce the size of DM denoiser networks. However, as denoisers evolve from variants of convolutional U-Nets toward newer Transformer architectures, it is of growing importance to understand the quantization sensitivity of different weight layers, operations and architecture types to performance. In this work, we address this challenge with Qua2SeDiMo, a mixed-precision Post-Training Quantization framework that generates explainable insights on the cost-effectiveness of various model weight quantization methods for different denoiser operation types and block structures. We leverage these insights to make high-quality mixed-precision quantization decisions for a myriad of diffusion models ranging from foundational U-Nets to state-of-the-art Transformers. As a result, Qua2SeDiMo can construct 3.4-bit, 3.9-bit, 3.65-bit and 3.7-bit weight quantization on PixArt-α, PixArt-Σ, Hunyuan-DiT and SDXL, respectively. We further pair our weight-quantization configurations with 6-bit activation quantization and outperform existing approaches in terms of quantitative metrics and generative image quality.

Diffusion models have demonstrated outstanding performance in generative tasks, making them ideal candidates for image editing. Recent studies highlight their ability to apply desired edits effectively by following textual instructions, yet with two key challenges remaining. First, these models struggle to apply multiple edits simultaneously, resulting in computational inefficiencies due to their reliance on sequential processing. Second, relying on textual prompts to determine the editing region can lead to unintended alterations to the image. We introduce FunEditor, an efficient diffusion model designed to learn atomic editing functions and perform complex edits by aggregating simpler functions. This approach enables complex editing tasks, such as object movement, by aggregating multiple functions and applying them simultaneously to specific areas. Our experiments demonstrate that FunEditor significantly outperforms recent inference-time optimization methods and fine-tuned models, either quantitatively across various metrics or through visual comparisons or both, on complex tasks like object movement and object pasting. In the meantime, with only 4 steps of inference, FunEditor achieves 5--24 times inference speedups over existing popular methods.

Federated learning is a decentralized framework that enables client participation in collaborative learning without centralized data collection. However, the framework is susceptible to suboptimal model convergence induced by heterogeneity among the client datasets. These discrepancies, including label imbalance, dissimilarity in data distributions, and uneven data volumes between clients, may cause disagreements among local client updates, affecting the ability of the global model to converge effectively during aggregation. We suggest that one potential solution to this problem lies in weighting the model aggregation by client importance and client-to-client relationships. Based on this idea, we propose HyperGenFL (HG-FL), a hypernetwork that generates aggregation weights from learnable client embeddings without requiring any training or benchmarking data. HG-FL utilizes the attention mechanism to capture inter-client relationships based on learnable client-specific embeddings in order to generate model aggregation weights dynamically during federated learning. By guiding the aggregation process with these learnable relationships between local models, HG-FL reduces update conflicts and improves global model performance. We assess HG-FL under various data-heterogeneous environments based on different benchmark datasets including Fashion-MNIST, CIFAR10, CIFAR100 and Tiny-ImageNet. Experimental results demonstrate that HG-FL can achieve superior performance over a range of existing baseline methods under challenging cases with various heterogeneous environments, large models and a large number of clients.

Aligning Large Language Models (LLMs) with nuanced user instructions is critical for their effective deployment in real-world applications. While prior methods focus on enhancing data diversity and complexity, they often overlook models' sensitivity to fine-grained variations in semantically similar instructions. To address this, we introduce DeMoRecon, a data augmentation framework that decomposes complex instructions into sub-components, modifies individual elements, and reconstructs them into instruction variants. This method preserves contextual integrity while injecting targeted variability essential for fine-grained instruction-following. Based on DeMoRecon, we construct the FGIV dataset, comprising over 1,700 seed instructions and thousands of nuanced variants designed for both supervised fine-tuning and preference-based alignment. Experimental results show that LLMs trained with FGIV achieve up to +10.2% improvement on our fine-grained FGIV-Eval benchmark and up to +8.8% on existing benchmarks such as FollowBench and InfoBench. These findings highlight the value of FGIV in advancing instruction sensitivity and robustness in LLMs.

The fine-tuning of Large Language Models (LLMs) specialized in code generation has seen notable advancements through the use of open-domain coding queries. Despite the successes, existing methodologies like Evol-Instruct encounter performance limitations, impeding further enhancements in code generation tasks. This paper examines the constraints of existing prompt evolution techniques and introduces a novel approach, Instruction Fusion (IF). IF innovatively combines two distinct prompts through a hybridization process, thereby enhancing the evolution of training prompts for code LLMs. Our experimental results reveal that the proposed novel method effectively addresses the shortcomings of prior methods, significantly improving the performance of Code LLMs across five code generation benchmarks, namely HumanEval, HumanEval+, MBPP, MBPP+ and MultiPL-E, which underscore the effectiveness of Instruction Fusion in advancing the capabilities of LLMs in code generation.

The detection of methane emissions has drawn significant attention from researchers, as methane is one of the main contributors to global warming. Due to their high accessibility and extensive coverage, satellite-based remote sensing data have become a widely used tool for studying methane emissions. The multispectral satellite Sentinel-2 is particularly prominent for studies related to methane, as it provides shortwave infrared (SWIR) spectral band data that align with the methane absorption line. Previous studies on methane emission detection using Sentinel-2 data rely mainly on synthetic data, which involves simulated methane emission plumes overlaid on real Sentinel-2 backgrounds. However, our analysis reveals that Sentinel-2 images with simulated plumes exhibit different data distributions and visual characteristics from those of real-world data. Moreover, these studies typically use only a single reference image as a clean background along with the plume image, which we find to be insufficient. Finally, the lack of publicly available benchmarks hinders research and development in this domain. To address these challenges, we introduce the MethaneS2CM dataset for methane emission detection, collected from Sentinel-2 multispectral data and the CarbonMapper platform. Our dataset consists of two main products (L1C and L2A) of Sentinel-2, including over 4000 plumes from 43 countries, ranging from 2016 to 2024. The dataset incorporates both long-term and short-term variations in Sentinel-2 images, with each sample consisting of images captured at different time periods. Furthermore, we conduct extensive experiments to evaluate a wide range of existing methane emission detection methods and models based on this dataset. We subsequently propose a deep methane emission classification model, MEECNet. Experimental results demonstrate that MEECNet outperforms all baseline methods. Our dataset is publicly available at https://huggingface.co/datasets/H1deaki/MethaneS2CM.

Diffusion Transformers (DiT) have become the de-facto model for generating high-quality visual content like videos and images. A huge bottleneck is the attention mechanism where complexity scales quadratically with resolution and video length. One logical way to lessen this burden is sparse attention, where only a subset of tokens or patches are included in the calculation. However, existing techniques fail to preserve visual quality at extremely high sparsity levels and might even incur non-negligible compute overheads. To address this concern, we propose Re-ttention, which implements very high sparse attention for visual generation models by leveraging the temporal redundancy of Diffusion Models to overcome the probabilistic normalization shift within the attention mechanism. Specifically, Re-ttention reshapes attention scores based on the prior softmax distribution history in order to preserve the visual quality of the full quadratic attention at very high sparsity levels. Experimental results on T2V/T2I models such as CogVideoX and the PixArt DiTs demonstrate that Re-ttention requires as few as 3.1% of the tokens during inference, outperforming contemporary methods like FastDiTAttn, Sparse VideoGen and MInference.

The fine-tuning of Large Language Models (LLMs) specialized in code generation has seen notable advancements through the use of open-domain coding queries.Despite the successes, existing methodologies like Evol-Instruct encounter performance limitations, impeding further enhancements in code generation tasks.This paper examines the constraints of existing prompt evolution techniques and introduces a novel approach, Instruction Fusion (IF).IF innovatively combines two distinct prompts through a hybridization process, thereby enhancing the evolution of training prompts for code LLMs.Our experimental results reveal that the proposed novel method effectively addresses the shortcomings of prior methods, significantly improving the performance of Code LLMs across five code generation benchmarks, namely HumanEval, Hu-manEval+, MBPP, MBPP+ and MultiPL-E, which underscore the effectiveness of Instruction Fusion in advancing the capabilities of LLMs in code generation.

Neural Architecture Search is a costly practice. The fact that a search space can span a vast number of design choices with each architecture evaluation taking nontrivial overhead makes it hard for an algorithm to sufficiently explore candidate networks. In this paper, we propose Auto-Build, a scheme which learns to align the latent embeddings of operations and architecture modules with the ground-truth performance of the architectures they appear in. By doing so, AutoBuild is capable of assigning interpretable importance scores to architecture modules, such as individual operation features and larger macro operation sequences such that high-performance neural networks can be constructed without any need for search. Through experiments performed on state-of-the-art image classification, segmentation, and Stable Diffusion models, we show that by mining a relatively small set of evaluated architectures, AutoBuild can learn to build high-quality architectures directly or help to reduce search space to focus on relevant areas, finding better architectures that outperform both the original labeled ones and ones found by search baselines. Code available at https://github.com/Ascend-Research/AutoBuild

With large language models (LLMs) achieving remarkable breakthroughs in natural language processing (NLP) domains, recent researchers have actively explored the potential of LLMs for recommendation systems by converting the input data into textual sentences through prompt templates.Although semantic knowledge from LLMs can help enrich the content information of items, to date it is still hard for them to achieve comparable performance to traditional deep learning recommendation models, partly due to a lack of ability to leverage collaborative filtering.In this paper, we propose a novel training-free prompting framework, PepRec, which aims to capture knowledge from both content-based filtering and collaborative filtering to boost recommendation performance with LLMs, while providing interpretation for the recommendation.Experiments based on two real-world datasets from different domains show that PepRec significantly outperforms various traditional deep learning recommendation models and prompt-based recommendation systems.

The reasoning performance of Large Language Models (LLMs) on a wide range of problems critically relies on chain-of-thought prompting, which involves providing a few chain of thought demonstrations as exemplars in prompts. Recent work, e.g., Tree of Thoughts, has pointed out the importance of exploration and self-evaluation in reasoning step selection for complex problem solving. In this paper, we present Boosting of Thoughts (BoT), an automated prompting framework for problem solving with LLMs by iteratively exploring and self-evaluating many trees of thoughts in order to acquire an ensemble of trial-and-error reasoning experiences, which will serve as a new form of prompting to solve the complex problem. Starting from a simple prompt without requiring examples, BoT iteratively explores and evaluates a large collection of reasoning steps, and more importantly, uses error analysis obtained from the LLM on them to explicitly revise prompting, which in turn enhances reasoning step generation, until a final answer is attained. Our experiments with GPT-4 and Llama2 across extensive complex mathematical problems demonstrate that BoT consistently achieves higher or comparable problem-solving rates than other advanced prompting approaches.

Understanding the decision-making process of Graph Neural Networks (GNNs) is crucial to their interpretability. Most existing methods for explaining GNNs typically rely on training auxiliary models, resulting in the explanations remain black-boxed. This paper introduces Graph Output Attribution (GOAt), a novel method to attribute graph outputs to input graph features, creating GNN explanations that are faithful, discriminative, as well as stable across similar samples. By expanding the GNN as a sum of scalar products involving node features, edge features and activation patterns, we propose an efficient analytical method to compute contribution of each node or edge feature to each scalar product and aggregate the contributions from all scalar products in the expansion form to derive the importance of each node and edge. Through extensive experiments on synthetic and real-world data, we show that our method not only outperforms various state-of-the-art GNN explainers in terms of the commonly used fidelity metric, but also exhibits stronger discriminability, and stability by a remarkable margin.

Understanding and explaining the predictions of Graph Neural Networks (GNNs), is crucial for enhancing their safety and trustworthiness. Subgraph-level explanations are gaining attention for their intuitive appeal. However, most existing subgraph-level explainers face efficiency challenges in explaining GNNs due to complex search processes. The key challenge is to find a balance between intuitiveness and efficiency while ensuring transparency. Additionally, these explainers usually induce subgraphs by nodes, which may introduce less-intuitive disconnected nodes in the subgraph-level explanations or omit many important subgraph structures. In this paper, we reveal that inducing subgraph explanations by edges is more comprehensive than other subgraph inducing techniques. We also emphasize the need of determining the subgraph explanation size for each data instance, as different data instances may involve different important substructures. Building upon these considerations, we introduce a training-free approach, named EiG-Search. We employ an efficient linear-time search algorithm over the edge-induced subgraphs, where the edges are ranked by an enhanced gradient-based importance. We conduct extensive experiments on a total of seven datasets, demonstrating its superior performance and efficiency both quantitatively and qualitatively over the leading baselines.

One key challenge to video restoration is to model the transition dynamics of video frames governed by motion. In this work, we propose Turtle to learn the truncated causal history model for efficient and high-performing video restoration. Unlike traditional methods that process a range of contextual frames in parallel, Turtle enhances efficiency by storing and summarizing a truncated history of the input frame latent representation into an evolving historical state. This is achieved through a sophisticated similarity-based retrieval mechanism that implicitly accounts for inter-frame motion and alignment. The causal design in Turtle enables recurrence in inference through state-memorized historical features while allowing parallel training by sampling truncated video clips. We report new state-of-the-art results on a multitude of video restoration benchmark tasks, including video desnowing, nighttime video deraining, video raindrops and rain streak removal, video super-resolution, real-world and synthetic video deblurring, and blind video denoising while reducing the computational cost compared to existing best contextual methods on all these tasks.

Federated learning is a decentralized approach to training machine learning models while preserving data privacy. To accommodate data heterogeneity among clients, a longstanding issue in Federated Learning, many Personalized Federated Learning (PFL) strategies decompose each client model into global modules, which are collaboratively learned by all clients and the server, and local modules, which are only trained locally on private data. While these strategies require every client to participate in training, in reality, many client devices lack sufficient data or computing resources to perform meaningful local training, making it difficult to achieve personalization for every client. In this paper, we present HyperFLoRA, a PFL framework that leverages knowledge learned from training-capable clients to enable the immediate creation of personalized models for training-incapable or new clients. HyperFLoRA uses adapters for personalization to minimize communication costs and client training workload while employing a trainable hypernetwork to generate personalized adapter weights for each client using minimal client statistical information. From experiments conducted on both convolutional and Transformer neural networks, HyperFLoRA can achieve superior model personalization performance for new clients that did not participate in training than conventional PFL methods, while significantly reducing training-related communication costs and client workload.

Modern recommender systems aim to identify items that are most pertinent to a particular user and are particularly useful when an overwhelming number of items are present. Feature embedding is essential to deep recommender systems, which constructs memory-efficient and semantically meaningful representations by mapping high-dimensional sparse feature vectors into low-dimensional dense vectors. Most existing systems assign a unified dimension to all feature fields, regardless of the diverse importance of different features, which usually results in sub-optimal performance and high memory usage. In this paper, we propose a low-cost embedding dimension search approach named DimReg for recommender systems, by assessing information overlapping between the dimensions within each feature field and pruning unimportant and redundant dimensions progressively during model training via a two-level polarization regularizer, while introducing minimum overhead. Moreover, our method does not require retraining after embedding dimension search, which significantly reduces the computational cost and is more friendly to deployment in real-world recommender systems. Extensive experiments conducted on multiple CTR (Click Through Rate) prediction tasks demonstrate that our method can efficiently reduce the model parameters up to 98.6%, and achieve strong recommendation performance outperforming existing automated embedding dimension search methods.

Predicting neural architecture performance is a challenging task and is crucial to neural architecture design and search. Existing approaches either rely on neural performance predictors which are limited to modeling architectures in a predefined design space involving specific sets of operators and connection rules, and cannot generalize to unseen architectures, or resort to Zero-Cost Proxies which are not always accurate. In this paper, we propose GENNAPE, a Generalized Neural Architecture Performance Estimator, which is pretrained on open neural architecture benchmarks, and aims to generalize to completely unseen architectures through combined innovations in network representation, contrastive pretraining, and a fuzzy clustering-based predictor ensemble. Specifically, GENNAPE represents a given neural network as a Computation Graph (CG) of atomic operations which can model an arbitrary architecture. It first learns a graph encoder via Contrastive Learning to encourage network separation by topological features, and then trains multiple predictor heads, which are soft-aggregated according to the fuzzy membership of a neural network. Experiments show that GENNAPE pretrained on NAS-Bench-101 can achieve superior transferability to 5 different public neural network benchmarks, including NAS-Bench-201, NAS-Bench-301, MobileNet and ResNet families under no or minimum fine-tuning. We further introduce 3 challenging newly labelled neural network benchmarks: HiAML, Inception and Two-Path, which can concentrate in narrow accuracy ranges. Extensive experiments show that GENNAPE can correctly discern high-performance architectures in these families. Finally, when paired with a search algorithm, GENNAPE can find architectures that improve accuracy while reducing FLOPs on three families.

Evaluating neural network performance is critical to deep neural network design but a costly procedure. Neural predictors provide an efficient solution by treating architectures as samples and learning to estimate their performance on a given task. However, existing predictors are task-dependent, predominantly estimating neural network performance on image classification benchmarks. They are also search-space dependent; each predictor is designed to make predictions for a specific architecture search space with predefined topologies and set of operations. In this paper, we propose a novel All-in-One Predictor (AIO-P), which aims to pretrain neural predictors on architecture examples from multiple, separate computer vision (CV) task domains and multiple architecture spaces, and then transfer to unseen downstream CV tasks or neural architectures. We describe our proposed techniques for general graph representation, efficient predictor pretraining and knowledge infusion techniques, as well as methods to transfer to downstream tasks/spaces. Extensive experimental results show that AIO-P can achieve Mean Absolute Error (MAE) and Spearman’s Rank Correlation (SRCC) below 1p% and above 0.5, respectively, on a breadth of target downstream CV tasks with or without fine-tuning, outperforming a number of baselines. Moreover, AIO-P can directly transfer to new architectures not seen during training, accurately rank them and serve as an effective performance estimator when paired with an algorithm designed to preserve performance while reducing FLOPs.

Multimodal Sentiment Analysis aims to predict the sentiment of video content.Recent research suggests that multimodal sentiment analysis critically depends on learning a good representation of multimodal information, which should contain both modality-invariant representations that are consistent across modalities as well as modality-specific representations.In this paper, we propose ConFEDE, a unified learning framework that jointly performs contrastive representation learning and contrastive feature decomposition to enhance representation of multimodal information.It decomposes each of the three modalities of a video sample, including text, video frames, and audio, into a similarity feature and a dissimilarity feature, which are learned by a contrastive relation centered around text.We conducted extensive experiments on CH-SIMS, MOSI and MOSEI to evaluate various state-of-the-art multimodal sentiment analysis methods.Experimental results show that ConFEDE outperforms all baselines on these datasets on a range of metrics.

Nested Named Entity Recognition (NNER), as a subarea of Named Entity Recognition, has presented longstanding challenges to researchers. In NNER, one entity may be part of a larger entity, which can occur at multiple levels. These nested structures prevent traditional sequence labeling methods from properly recognizing all entities. While recent research has focused on designing better recognition methods for NNER in various languages, Chinese Nested Named Entity Recognition (CNNER) is still underdeveloped, largely due to a lack of freely available CNNER benchmarks. To support CNNER research, in this paper, we introduce ChiNesE, a CNNER dataset comprising 20,000 sentences from online passages in multiple domains and containing 117,284 entities that fall into 10 categories, of which 43.8% are nested named entities. Based on ChiNesE, we propose Mulco, a novel method that can recognize named entities in nested structures through multiple scopes. Each scope uses a scope-based sequence labeling method that predicts an anchor and the length of a named entity to recognize it. Experimental results show that Mulco outperforms state-of-the-art baseline methods with different recognition schemes on ChiNesE and ACE 2005 Chinese corpus.

Current recommender systems employ large-sized embedding tables with uniform dimensions for all features, leading to overfitting, high computational cost, and suboptimal generalizing performance. Many techniques aim to solve this issue by feature selection or embedding dimension search. However, these techniques typically select a fixed subset of features or embedding dimensions for all instances and feed all instances into one recommender model without considering heterogeneity between items or users. This paper proposes a novel instance-wise Hierarchical Architecture Search framework, iHAS, which automates neural architecture search at the instance level. Specifically, iHAS incorporates three stages: searching, clustering, and retraining. The searching stage identifies optimal instance-wise embedding dimensions across different field features via carefully designed Bernoulli gates with stochastic selection and regularizers. After obtaining these dimensions, the clustering stage divides samples into distinct groups via a deterministic selection approach of Bernoulli gates. The retraining stage then constructs different recommender models, each one designed with optimal dimensions for the corresponding group. We conduct extensive experiments to evaluate the proposed iHAS on two public benchmark datasets from a real-world recommender system. The experimental results demonstrate the effectiveness of iHAS and its outstanding transferability to widely-used deep recommendation models.

Despite the success of deep learning in video understanding tasks, processing every frame in a video is computationally expensive and often unnecessary in real-time applications. Frame selection aims to extract the most informative and representative frames to help a model better understand video content. Existing frame selection methods either individually sample frames based on per-frame importance prediction, without considering interaction among frames, or adopt reinforcement learning agents to find representative frames in succession, which are costly to train and may lead to potential stability issues. To overcome the limitations of existing methods, we propose a Search-Map-Search learning paradigm which combines the advantages of heuristic search and supervised learning to select the best combination of frames from a video as one entity. By combining search with learning, the proposed method can better capture frame interactions while incurring a low inference overhead. Specifically, we first propose a hierarchical search method conducted on each training video to search for the optimal combination of frames with the lowest error on the downstream task. A feature mapping function is then learned to map the frames of a video to the representation of its target optimal frame combination. During inference, another search is performed on an unseen video to select a combination of frames whose feature representation is close to the projected feature representation. Extensive experiments based on several action recognition benchmarks demonstrate that our frame selection method effectively improves performance of action recognition models, and significantly outperforms a number of competitive baselines.

With the increasing complexity and capacity of modern Field-Programmable Gate Arrays (FPGAs), there is a growing demand for efficient FPGA computer-aided design (CAD) tools, particularly in the placement stage. While some previous works, such as RLPlace, have explored the efficacy of single-state Reinforcement Learning (RL) to optimize FPGA placement by framing it as a multi-armed bandit (MAB) problem, numerous AI techniques remain unexplored due to the outstanding engineering challenges to integrate them into the FPGA CAD flow which is based on C++. In this paper, we present VPR-Gym, a Python environment built on OpenAI Gym, that allows seamless integration with various machine learning libraries including PyTorch, TensorFlow, and Nevergrad while enabling the comparison between different AI techniques for FPGA placement. Moreover, we introduce a learning objective that reformulates the FPGA placement task as an optimization problem, thereby expanding the range of AI techniques that can be investigated beyond those for MAB problems. To showcase the capabilities of our platform, we conduct experiments comparing the performance of various MAB algorithms and evolution strategy (ES) algorithms. Our findings demonstrate that the ES approaches exhibit superior performance over the existing MAB approaches, highlighting the effectiveness of VPR-Gym in facilitating AI research to enhance FPGA placement.

Online advertisements are important elements in e-commerce sites, social media platforms, and search engines. With the increasing popularity of mobile browsing, many online ads are displayed with visual information in the form of a cover image in addition to text descriptions to grab the attention of users. Various recent studies have focused on predicting the click rates of online advertisements aware of visual features or composing optimal advertisement elements to enhance visibility. In this paper, we propose Advertisement Style Editing and Attractiveness Enhancement (AdSEE), which explores whether semantic editing to ads images can affect or alter the popularity of online advertisements. We introduce StyleGAN-based facial semantic editing and inversion to ads images and train a click rate predictor attributing GAN-based face latent representations in addition to traditional visual and textual features to click rates. Through a large collected dataset named QQ-AD, containing 20,527 online ads, we perform extensive offline tests to study how different semantic directions and their edit coefficients may impact click rates. We further design a Genetic Advertisement Editor to efficiently search for the optimal edit directions and intensity given an input ad cover image to enhance its projected click rates. Online A/B tests performed over a period of 5 days have verified the increased click-through rates of AdSEE-edited samples as compared to a control group of original ads, verifying the relation between image styles and ad popularity. We open source the code for AdSEE research at https://github.com/LiyaoJiang1998/adsee.

Specialized hardware has become an indispensable component to deep neural network (DNN) acceleration. To keep up with the rapid evolution of neural networks, holistic and automated solutions for jointly optimizing both hardware (HW) architectures and software (SW) mapping have been studied. These studies face two major challenges. First, the combined HW-SW design space is vast, which hinders the finding of optimal or near-optimal designs. This issue is exacerbated for industrial cases when cycle accurate models are used for design evaluation in the joint optimization. Second, HW design is prone to overfitting to the input DNNs used in the HW-SW co-optimization. To address these issues, in this paper, we propose UNICO, an efficient Unified Co-Optimization framework with a novel Robustness metric for better HW generalization. Guided by a high-fidelity surrogate model, UNICO employs multi-objective Bayesian optimization to effectively explore the HW design space, and conducts adaptive, parallel and scalable software mapping search based on successive halving. To reduce HW overfitting, we propose a HW robustness metric by relating a HW configuration’s quality to its sensitivity in software mapping search, and quantitatively incorporate this metric to search for more robust HW design(s). We implement UNICO in open source accelerator platform, and compare it with the state-of-the-art solution HASCO. Experiments show that UNICO significantly outperforms HASCO; it finds design(s) with similar quality to HASCO up to 4 × faster, and eventually converges to better and more robust designs. Finally, we deploy UNICO for optimizing an industrial accelerator, and show that it generates enhanced HW design(s) for key real-world DNNs.

Optimizing Deep Neural Networks (DNNs) to obtain high-quality models for efficient real-world deployment has posed multi-faceted challenges to machine learning engineers. Existing methods either search for neural architectures in heuristic design spaces or apply low-level adjustments to computation primitives to improve inference efficiency on hardware. We present Automated Graph Optimization (AutoGO), a framework to evolve neural networks in a low-level Computation Graph (CG) of primitive operations to improve both its performance and hardware friendliness. Through a tokenization scheme, AutoGO performs variable-sized segment mutations, making both primitive changes and larger-grained changes to CGs. We introduce our segmentation and mutation algorithms, efficient frequent segment mining technique, as well as a pretrained context-aware predictor to estimate the impact of segment replacements. Extensive experimental results show that AutoGO can automatically evolve several typical large convolutional networks to achieve significant task performance improvement and FLOPs reduction on a range of CV tasks, ranging from Classification, Semantic Segmentation, Human Pose Estimation, to Super Resolution, yet without introducing any newer primitive operations. We also demonstrate the lightweight deployment results of AutoGO-optimized super-resolution and denoising U-Nets on a cycle simulator for a Neural Processing Unit (NPU), achieving PSNR improvement and latency/power reduction simultaneously. Code available at https://github.com/Ascend-Research/AutoGO.

Matrix Factorization (MF) is essential to many estimation tasks. Most existing matrix factorization methods focus on least squares matrix factorization (LSMF), which aims to minimize a smooth L2 loss between observations and their dependent matrix measurement variables. In reality, however, L1 loss and check loss are widely used in regression to deal with outliers or observations contaminated by skewed or heavy-tailed noise. Although under certain conditions, linear convergence to the global optimality can be established for matrix factorization under the L2 loss, there is a lack of provably efficient algorithms for solving matrix factorization under non-smooth losses. In this paper, we investigate Quantile Matrix Factorization (QMF), the counterpart of Quantile Regression in matrix estimation, that adopts a tunable check loss and introduces robustness to matrix estimation for skewed and heavy- tailed observations, which are prevalent in reality. To deal with the non-smooth loss, we propose Nesterov- smoothed QMF (NsQMF), extending Nesterov's optimal smooth approximation technique to the matrix factorization setting. We then present an alternating minimization algorithm to solve the smooth NsQMF efficiently. We mathematically prove that solving the smoothed NsQMF is equivalent to solving the original non-smooth QMF problem and that our proposed algorithm achieves linear convergence to the global optimality of QMF. Numerical evaluations verify our theoretical findings and demonstrate that NsQMF significantly outperforms the commonly used LSMF and prior approximate smoothing heuristics for QMF under various noise distributions.

Understanding and modelling the performance of neural architectures is key to Neural Architecture Search (NAS). Performance predictors have seen widespread use in low-cost NAS and achieve high ranking correlations between predicted and ground truth performance in several NAS benchmarks. However, existing predictors are often designed based on network encodings specific to a predefined search space and are therefore not generalizable to other search spaces or new architecture families. In this paper, we propose a general-purpose neural predictor for NAS that can transfer across search spaces, by representing any given candidate Convolutional Neural Network (CNN) with a Computation Graph (CG) that consists of primitive operators. We further combine our CG network representation with Contrastive Learning (CL) and propose a graph representation learning procedure that leverages the structural information of unlabeled architectures from multiple families to train CG embeddings for our performance predictor. Experimental results on NAS-Bench-101, 201 and 301 demonstrate the efficacy of our scheme as we achieve strong positive Spearman Rank Correlation Coefficient (SRCC) on every search space, outperforming several Zero-Cost Proxies, including Synflow and Jacov, which are also generalizable predictors across search spaces. Moreover, when using our proposed general-purpose predictor in an evolutionary neural architecture search algorithm, we can find highperformance architectures on NAS-Bench-101 and find a MobileNetV3 architecture that attains 79.2% top-1 accuracy on ImageNet.

AutoAugment is an automatic method to design data augmentation policies for deep learning, and has achieved significant improvements on computer vision tasks. However, since early AutoAugment approaches cost thousands of GPU hours, there is a recent demand to investigate low-cost search methods that can still find effective augmentation policies. In this paper, we propose a multi-armed bandit algorithm, named Bandit Data Augment (BDA), to efficiently search for optimal and transferable data augmentation policies. We leverage Successive Halving to make the bandit model progressively focus on more promising augmentation operations during the search, leading to sparse selection of operations and more generalizable augmentation policies. We also propose a computationally efficient rewarding scheme to reduce the evaluation cost of augmentation policies. Extensive experiments demonstrate that BDA can achieve comparable or better performance than prior Auto Augment methods on a wide range of models on CIFAR-10/100 and ImageNet benchmarks. Besides, BDA is 555 times and 536 times faster than AutoAugment on CIFAR-10 and ImageNet, respectively. In addition, BDA is 16 times faster than Fast Auto Augment on ImageNet. More importantly, BDA can discover policies that are transferable across datasets and models, and achieve similar performance to policies found directly on the target dataset.

Cross-domain recommendation is an important method to improve recommender system performance, especially when observations in target domains are sparse. However, most existing techniques focus on single-target or dual-target cross-domain recommendation (CDR) and are hard to be generalized to CDR with multiple target domains. In addition, the negative transfer problem is prevalent in CDR, where the recommendation performance in a target domain may not always be enhanced by knowledge learned from a source domain, especially when the source domain has sparse data. In this study, we propose CAT-ART, a multi-target CDR method that learns to improve recommendations in all participating domains through representation learning and embedding transfer. Our method consists of two parts: a self-supervised Contrastive AuToencoder (CAT) framework to generate global user embeddings based on information from all participating domains, and an Attention-based Representation Transfer (ART) framework which transfers domain-specific user embeddings from other domains to assist with target domain recommendation. CAT-ART boosts the recommendation performance in any target domain through the combined use of the learned global user representation and knowledge transferred from other domains, in addition to the original user embedding in the target domain. We conducted extensive experiments on a collected real-world CDR dataset spanning 5 domains and involving a million users. Experimental results demonstrate the superiority of the proposed method over a range of prior arts. We further conducted ablation studies to verify the effectiveness of the proposed components. Our collected dataset will be open-sourced to facilitate future research in the field of multi-domain recommender systems and user modelling.

Text clustering, as one of the most fundamental challenges in unsupervised learning, aims at grouping semantically similar text segments without relying on human annotations. With the rapid development of deep learning, deep clustering has achieved significant advantages over traditional clustering methods. Despite the effectiveness, most existing deep text clustering methods rely heavily on representations pre-trained in general domains, which may not be the most suitable solution for clustering in specific target domains. To address this issue, we propose CEIL, a novel Classification-Enhanced Iterative Learning framework for short text clustering, which aims at generally promoting the clustering performance by introducing a classification objective to iteratively improve feature representations. In each iteration, we first adopt a language model to retrieve the initial text representations, from which the clustering results are collected using our proposed Category Disentangled Contrastive Clustering (CDCC) algorithm. After strict data filtering and aggregation processes, samples with clean category labels are retrieved, which serve as supervision information to update the language model with the classification objective via a prompt learning approach. Finally, the updated language model with improved representation ability is used to enhance clustering in the next iteration. Extensive experiments demonstrate that the CEIL framework significantly improves the clustering performance over iterations, and is generally effective on various clustering algorithms. Moreover, by incorporating CEIL on CDCC, we achieve the state-of-the-art clustering performance on a wide range of short text clustering benchmarks outperforming other strong baseline methods.

Although conceptualization has been widely studied in semantics and knowledge representation, it is still challenging to find the most accurate concept terms to tag fast-growing social media content. This is partly attributed to the fact that most traditional knowledge bases contain general terms of the world, such as trees and cars, which are not interesting to users, and do not have the defining power for social media content. Another reason is that the intricate use of tense, negation and grammar in social media content may change the logic or emphasis of the content, thus focusing on different main ideas. In this paper, we present TAG, a high-quality concept matching dataset consisting of 10,000 labeled pairs of fine-grained concepts and web-styled natural language sentences, mined from open-domain social media content. The concepts we provide are the trending terms on social media and have the right granularity to define user interests, e.g., highly educated actors instead of just actors. In the meantime, TAG offers a concept graph which interconnects these fine-grained concepts and entities to provide contextual information. We evaluate a wide range of neural text matching models as well as pre-trained language models for the concept matching task on TAG, and point out their insufficiency to tag social media content to characterize its main idea. We further propose a novel graph-graph matching framework that demonstrates superior abstraction and generalization performance by better utilizing both the structural information in the concept graph and logic interactions between semantic units in the natural language sentence via syntactic dependency parsing.

Cross-domain recommendation can help alleviate the data sparsity issue in traditional sequential recommender systems. In this paper, we propose the RecGURU algorithm framework to generate a Generalized User Representation (GUR) incorporating user information across domains in sequential recommendation, even when there is minimum or no common users in the two domains. We propose a self-attentive autoencoder to derive latent user representations, and a domain discriminator, which aims to predict the origin domain of a generated latent representation. We propose a novel adversarial learning method to train the two modules to unify user embeddings generated from different domains into a single global GUR for each user. The learned GUR captures the overall preferences and characteristics of a user and thus can be used to augment the behavior data and improve recommendations in any single domain in which the user is involved. Extensive experiments have been conducted on two public cross-domain recommendation datasets as well as a large dataset collected from real-world applications. The results demonstrate that RecGURU boosts performance and outperforms various state-of-the-art sequential recommendation and cross-domain recommendation methods. The collected data will be released to facilitate future research.

It is common practice for many large e-commerce operators to analyze daily logged transaction data to predict customer purchase behavior, which may potentially lead to more effective recommendations and increased sales. Traditional recommendation techniques based on collaborative filtering, although having gained success in video and music recommendation, are not sufficient to fully leverage the diverse information contained in the implicit user behavior on e-commerce platforms. In this article, we analyze user action records in the Alibaba Mobile Recommendation dataset from the Alibaba Tianchi Data Lab, as well as the Retailrocket recommender system dataset from the Retail Rocket website. To estimate the probability that a user will purchase a certain item tomorrow, we propose a new model called Time-decayed Multifaceted Factorizing Personalized Markov Chains (Time-decayed Multifaceted-FPMC), taking into account multiple types of user historical actions not only limited to past purchases but also including various behaviors such as clicks, collects and add-to-carts. Our model also considers the time-decay effect of the influence of past actions. To learn the parameters in the proposed model, we further propose a unified framework named Bayesian Sparse Factorization Machines. It generalizes the theory of traditional Factorization Machines to a more flexible learning structure and trains the Time-decayed Multifaceted-FPMC with the Markov Chain Monte Carlo method. Extensive evaluations based on multiple real-world datasets demonstrate that our proposed approaches significantly outperform various existing purchase recommendation algorithms.

Neural architecture search (NAS) has achieved remarkable results in deep neural network design. Differentiable architecture search converts the search over discrete architectures into a hyperparameter optimization problem which can be solved by gradient descent. However, questions have been raised regarding the effectiveness and generalizability of gradient methods for solving non-convex architecture hyperparameter optimization problems. In this paper, we propose L2NAS, which learns to intelligently optimize and update architecture hyperparameters via an actor neural network based on the distribution of high-performing architectures in the search history. We introduce a quantile-driven training procedure which efficiently trains L2NAS in an actor-critic framework via continuous-action reinforcement learning. Experiments show that L2NAS achieves state-of-the-art results on NAS-Bench-201 benchmark as well as DARTS search space and Once-for-All MobileNetV3 search space. We also show that search policies generated by L2NAS are generalizable and transferable across different training datasets with minimal fine-tuning.

Neural architecture search automates neural network design and has achieved state-of-the-art results in many deep learning applications. While recent literature has focused on designing networks to maximize accuracy, little work has been conducted to understand the compatibility of architecture design spaces to varying hardware. In this paper, we analyze the neural blocks used to build Once-for-All (MobileNetV3), ProxylessNAS and ResNet families, in order to understand their predictive power and inference latency on various devices, including Huawei Kirin 9000 NPU, RTX 2080 Ti, AMD Threadripper 2990WX, and Samsung Note10. We introduce a methodology to quantify the friendliness of neural blocks to hardware and the impact of their placement in a macro network on overall network performance via only end-to-end measurements. Based on extensive profiling results, we derive design insights and apply them to hardware-specific search space reduction. We show that searching in the reduced search space generates better accuracy-latency Pareto frontiers than searching in the original search spaces, customizing architecture search according to the hardware. Moreover, insights derived from measurements lead to notably higher ImageNet top-1 scores on all search spaces investigated.

Query feeds recommendation is a new recommended paradigm in mobile search applications, where a stream of queries need to be recommended to improve user engagement. It requires a great quantity of attractive queries for recommendation. A conventional solution is to retrieve queries from a collection of past queries recorded in user search logs. However, these queries usually have poor readability and limited coverage of article content, and are thus not suitable for the query feeds recommendation scenario. Furthermore, to deploy the generated queries for recommendation, human validation, which is costly in practice, is required to filter unsuitable queries. In this paper, we propose TitIE, a query mining system to generate valuable queries using the titles of documents. We employ both an extractive text generator and an abstractive text generator to generate queries from titles. To improve the acceptance rate during human validation, we further propose a model-based scoring strategy to pre-select the queries that are more likely to be accepted during human validation. Finally, we propose a novel dual learning approach to jointly learn the generation model and the selection model by making full use of the unlabeled corpora under a semi-supervised scheme, thereby simultaneously improving the performance of both models. Results from both offline and online evaluations demonstrate the superiority of our approach.

Despite the empirical success of neural architecture search (NAS) in deep learning applications, the optimality, reproducibility and cost of NAS schemes remain hard to assess. In this paper, we propose Generative Adversarial NAS (GA-NAS) with theoretically provable convergence guarantees, promoting stability and reproducibility in neural architecture search. Inspired by importance sampling, GA-NAS iteratively fits a generator to previously discovered top architectures, thus increasingly focusing on important parts of a large search space. Furthermore, we propose an efficient adversarial learning approach, where the generator is trained by reinforcement learning based on rewards provided by a discriminator, thus being able to explore the search space without evaluating a large number of architectures. Extensive experiments show that GA-NAS beats the best published results under several cases on three public NAS benchmarks. In the meantime, GA-NAS can handle ad-hoc search constraints and search spaces. We show that GA-NAS can be used to improve already optimized baselines found by other NAS methods, including EfficientNet and ProxylessNAS, in terms of ImageNet accuracy or the number of parameters, in their original search space.

Human activity recognition (HAR) based on mobile sensors plays an important role in ubiquitous computing. However, the rise of data regulatory constraints precludes collecting private and labeled signal data from personal devices at scale. Thanks to the growth of computational power on mobile devices, federated learning has emerged as a decentralized alternative solution to model training, which iteratively aggregates locally updated models into a shared global model, therefore being able to leverage decentralized, private data without central collection. However, the effectiveness of federated learning for HAR is affected by the fact that each user has different activity types and even a different signal distribution for the same activity type. Furthermore, it is uncertain if a single global model trained can generalize well to individual users or new users with heterogeneous data. In this paper, we propose Meta-HAR, a federated representation learning framework, in which a signal embedding network is meta-learned in a federated manner, while the learned signal representations are further fed into a personalized classification network at each user for activity prediction. In order to boost the representation ability of the embedding network, we treat the HAR problem at each user as a different task and train the shared embedding network through a Model-Agnostic Meta-learning framework, such that the embedding network can generalize to any individual user. Personalization is further achieved on top of the robustly learned representations in an adaptation procedure. We conducted extensive experiments based on two publicly available HAR datasets as well as a newly created HAR dataset. Results verify that Meta-HAR is effective at maintaining high test accuracies for individual users, including new users, and significantly outperforms several baselines, including Federated Averaging, Reptile and even centralized learning in certain cases. Our collected dataset will be open-sourced to facilitate future development in the field of sensor-based human activity recognition.

Deep learning models for human activity recognition (HAR) based on sensor data have been heavily studied recently. However, the generalization ability of deep models on complex real-world HAR data is limited by the availability of high-quality labeled activity data, which are hard to obtain. In this article, we design a similarity embedding neural network that maps input sensor signals onto real vectors through carefully designed convolutional and Long Short-Term Memory (LSTM) layers. The embedding network is trained with a pairwise similarity loss, encouraging the clustering of samples from the same class in the embedded real space, and can be effectively trained on a small dataset and even on a noisy dataset with mislabeled samples. Based on the learned embeddings, we further propose both nonparametric and parametric approaches for activity recognition. Extensive evaluation based on two public datasets has shown that the proposed similarity embedding network significantly outperforms state-of-the-art deep models on HAR classification tasks, is robust to mislabeled samples in the training set, and can also be used to effectively denoise a noisy dataset.

The competitive performance of neural machine translation (NMT) critically relies on large amounts of training data. However, acquiring high-quality translation pairs requires expert knowledge and is costly. Therefore, how to best utilize a given dataset of samples with diverse quality and characteristics becomes an important yet understudied question in NMT. Curriculum learning methods have been introduced to NMT to optimize a model's performance by prescribing the data input order, based on heuristics such as the assessment of noise and difficulty levels. However, existing methods require training from scratch, while in practice most NMT models are pre-trained on big data already. Moreover, as heuristics, they do not generalize well. In this paper, we aim to learn a curriculum for improving a pre-trained NMT model by re-selecting influential data samples from the original training set and formulate this task as a reinforcement learning problem. Specifically, we propose a data selection framework based on Deterministic Actor-Critic, in which a critic network predicts the expected change of model performance due to a certain sample, while an actor network learns to select the best sample out of a random batch of samples presented to it. Experiments on several translation datasets show that our method can further improve the performance of NMT when original batch training reaches its ceiling, without using additional new training data, and significantly outperforms several strong baseline methods.

The widespread deployment of machine learning applications in ubiquitous environments has sparked interests in exploiting the vast amount of data stored on mobile devices. To preserve data privacy, Federated Learning has been proposed to learn a shared model by performing distributed training locally on participating devices and aggregating the local models into a global one. However, due to the limited network connectivity of mobile devices, it is not practical for federated learning to perform model updates and aggregation on all participating devices in parallel. Besides, data samples across all devices are usually not independent and identically distributed (IID), posing additional challenges to the convergence and speed of federated learning. In this paper, we propose Favor, an experience-driven control framework that intelligently chooses the client devices to participate in each round of federated learning to counterbalance the bias introduced by non-IID data and to speed up convergence. Through both empirical and mathematical analysis, we observe an implicit connection between the distribution of training data on a device and the model weights trained based on those data, which enables us to profile the data distribution on that device based on its uploaded model weights. We then propose a mechanism based on deep Q-learning that learns to select a subset of devices in each communication round to maximize a reward that encourages the increase of validation accuracy and penalizes the use of more communication rounds. With extensive experiments performed in PyTorch, we show that the number of communication rounds required in federated learning can be reduced by up to 49% on the MNIST dataset, 23% on FashionMNIST, and 42% on CIFAR-10, as compared to the Federated Averaging algorithm.

Deep neural networks have recently become a popular solution to keyword spotting systems, which enable the control of smart devices via voice. In this paper, we apply neural architecture search to search for convolutional neural network models that can help boost the performance of keyword spotting based on features extracted from acoustic signals while maintaining an acceptable memory footprint. Specifically, we use differentiable architecture search techniques to search for operators and their connections in a predefined cell search space. The found cells are then scaled up in both depth and width to achieve competitive performance. We evaluated the proposed method on Google's Speech Commands Dataset and achieved a state-of-the-art accuracy of over 97% on the setting of 12-class utterance classification commonly reported in the literature.

Concepts naturally constitute an abstraction for fine-grained entities and knowledge in the open domain. They enable search engines and recommendation systems to enhance user experience by discovering high-level abstraction of a search query and the user intent behind it. In this paper, we study the problem of query conceptualization, which is to find the most appropriate matching concepts for any given search query from a large pool of pre-defined concepts. We propose a coarse-to-fine approach to first reduce the search space for each query through a shortlisting scheme and then identify the matching concepts using pre-trained language models, which are meta-tuned to our query-concept matching task. Our shortlisting scheme involves using a GRU-based Relevant Words Generator (RWG) to first expand and complete the context of the given query and then shortlisting the candidate concepts through a scoring mechanism based on word overlaps. To accurately identify the most appropriate matching concepts for a query, even when the concepts may have zero verbatim overlaps with the query, we meta-fine-tune a BERT pairwise text-matching model under the Reptile meta-learning algorithm, which achieves zero-shot transfer learning on the conceptualization problem. Our two-stage framework can be trained with data completely derived from a search click graph, without requiring any human labelling efforts. For evaluation, we have constructed a large click graph based on more than $7$ million instances of the click history recorded in Tencent QQ browser and performed the query conceptualization task based on a large ontology with $159,148$ unique concepts. Results from a range of evaluation methods, including an offline evaluation procedure on the click graph, human evaluation, online A/B testing and case studies, have demonstrated the superiority of our approach over a number of competitive pre-trained language models and fine-tuned neural network baselines.

Understanding what online users may pay attention to on the web is key to content recommendation and search services. These services will benefit from a highly structured and web-scale ontology of entities, concepts, events, topics and categories. While existing knowledge bases and taxonomies embody a large volume of entities and categories, we argue that they fail to discover properly grained concepts, events and topics in the language style of online users. Neither is a logically structured ontology maintained among these notions. In this paper, we present GIANT, a mechanism to construct a user-centered, web-scale, structured ontology, containing a large number of natural language phrases conforming to user attentions at various granularities, mined from the vast volume of web documents and search click logs. Various types of edges are also constructed to maintain a hierarchy in the ontology. We present our detailed techniques used in GIANT, and evaluate the proposed models and methods as compared to a variety of baselines, as well as deploy the resulted Attention Ontology in real-world applications, involving over a billion users, to observe its effect on content recommendation. The online performance of the ontology built by GIANT proves that it can significantly improve the click-through rate in news feeds recommendation.

The ability to ask questions is important in both human and machine intelligence. Learning to ask questions helps knowledge acquisition, improves question-answering and machine reading comprehension tasks, and helps a chatbot to keep the conversation flowing with a human. Existing question generation models are ineffective at generating a large amount of high-quality question-answer pairs from unstructured text, since given an answer and an input passage, question generation is inherently a one-to-many mapping. In this paper, we propose Answer-Clue-Style-aware Question Generation (ACS-QG), which aims at automatically generating high-quality and diverse question-answer pairs from unlabeled text corpus at scale by imitating the way a human asks questions. Our system consists of: i) an information extractor, which samples from the text multiple types of assistive information to guide question generation; ii) neural question generators, which generate diverse and controllable questions, leveraging the extracted assistive information; and iii) a neural quality controller, which removes low-quality generated data based on text entailment. We compare our question generation models with existing approaches and resort to voluntary human evaluation to assess the quality of the generated question-answer pairs. The evaluation results suggest that our system dramatically outperforms state-of-the-art neural question generation models in terms of the generation quality, while being scalable in the meantime. With models trained on a relatively smaller amount of data, we can generate 2.8 million quality-assured question-answer pairs from a million sentences found in Wikipedia.

Extracting events accurately from vast news corpora and organize events logically is critical for news apps and search engines, which aim to organize news information collected from the Internet and present it to users in the most sensible forms. Intuitively speaking, an event is a group of news documents that report the same news incident possibly in different ways. In this article, we describe our experience of implementing a news content organization system at Tencent to discover events from vast streams of breaking news and to evolve news story structures in an online fashion. Our real-world system faces unique challenges in contrast to previous studies on topic detection and tracking (TDT) and event timeline or graph generation, in that we (1) need to accurately and quickly extract distinguishable events from massive streams of long text documents, and (2) must develop the structures of event stories in an online manner, in order to guarantee a consistent user viewing experience. In solving these challenges, we propose Story Forest , a set of online schemes that automatically clusters streaming documents into events, while connecting related events in growing trees to tell evolving stories. A core novelty of our Story Forest system is EventX , a semi-supervised scheme to extract events from massive Internet news corpora. EventX relies on a two-layered, graph-based clustering procedure to group documents into fine-grained events. We conducted extensive evaluations based on (1) 60 GB of real-world Chinese news data, (2) a large Chinese Internet news dataset that contains 11,748 news articles with truth event labels, and (3) the 20 News Groups English dataset, through detailed pilot user experience studies. The results demonstrate the superior capabilities of Story Forest to accurately identify events and organize news text into a logical structure that is appealing to human readers.

Big data analytics are practiced in many fields to extract insights from massive amounts of data. With exponential growth in both the volume and variety of data, analytic queries have expanded from those executed in a single datacenter to those requiring inputs from multiple datacenters that are geographically separate or even globally distributed. Unfortunately, the software stack that supports data analytics is designed originally for a cluster environment and is not tailored to execute global analytic queries, where resources such as inter-datacenter networks may vary on the fly. Existing optimization strategies that determine the query execution plan before its execution are not able to adapt to resource variations at query runtime. In this article, we present Turbo, a lightweight and non-intrusive global analytics system that can dynamically adjust query execution plans for geo-distributed analytics in the presence of time-varying resources and network bandwidth across datacenters. Turbo uses machine learning to accurately predict the time cost of a query execution plan so that dynamic adjustments can be made to it when necessary. Turbo is non-intrusive in the sense that it does not require modifications to the existing software stack for data analytics. We have implemented a real-world prototype of Turbo, and evaluated it on a cluster of 33 instances across eight regions in the Google Cloud Platform. Our experimental results have shown that Turbo can achieve an accuracy of 95 percent for estimating time costs, and can reduce the query completion time by 41 percent.

To learn the underlying parent-child influence relationships between nodes in a diffusion network, most existing approaches require timestamps that pinpoint the exact time when node infections occur in historical diffusion processes. In many real-world diffusion processes like the spread of epidemics, monitoring such infection temporal information is often expensive and difficult. In this work, we study how to carry out diffusion network inference without infection timestamps, using only the final infection statuses of nodes in each historical diffusion process, which are more readily accessible in practice. Our main result is a probabilistic model that can find for each node an appropriate number of most probable parent nodes, who are most likely to have generated the historical infection results of the node. Extensive experiments on both synthetic and real-world networks are conducted, and the results verify the effectiveness and efficiency of our approach.

Identifying the relationship between two articles, e.g., whether two articles published from different sources describe the same breaking news, is critical to many document understanding tasks.Existing approaches for modeling and matching sentence pairs do not perform well in matching longer documents, which embody more complex interactions between the enclosed entities than a sentence does.To model article pairs, we propose the Concept Interaction Graph to represent an article as a graph of concepts.We then match a pair of articles by comparing the sentences that enclose the same concept vertex through a series of encoding techniques, and aggregate the matching signals through a graph convolutional network.To facilitate the evaluation of long article matching, we have created two datasets, each consisting of about 30K pairs of breaking news articles covering diverse topics in the open domain.Extensive evaluations of the proposed methods on the two datasets demonstrate significant improvements over a wide range of state-of-the-art methods for natural language matching.

The need to scale up machine learning, in the presence of a rapid growth of data both in volume and in variety, has sparked broad interests to develop distributed machine learning systems, typically based on parameter servers. However, since these systems are based on a dedicated cluster of physical or virtual machines, they have posed non-trivial cluster management overhead to machine learning practitioners and data scientists. In addition, there exists an inherent mismatch between the dynamically varying resource demands during a model training job and the inflexible resource provisioning model of current cluster-based systems.In this paper, we propose SIREN, an asynchronous distributed machine learning framework based on the emerging serverless architecture, with which stateless functions can be executed in the cloud without the complexity of building and maintaining virtual machine infrastructures. With SIREN, we are able to achieve a higher level of parallelism and elasticity by using a swarm of stateless functions, each working on a different batch of data, while greatly reducing system configuration overhead. Furthermore, we propose a scheduler based on Deep Reinforcement Learning to dynamically control the number and memory size of the stateless functions that should be used in each training epoch. The scheduler learns from the training process itself, in pursuit for the minimum possible training time given a cost. With our real-world prototype implementation on AWS Lambda, extensive experimental results have shown that SIREN can reduce model training time by up to 44%, as compared to traditional machine learning training benchmarks on AWS EC2 at the same cost.

Related search query recommendation is a standard feature in many modern search engines. Interesting and relevant queries often increase the active time of users and improve the overall search experience. However, conventional approaches based on tag extraction, keywords matching or click graph link analysis suffer from the common problem of limited coverage and generalizability, which means the system could only make suggestions for a small portion of well-formed search queries.

Most current distributed machine learning systems try to scale up model training by using a data-parallel architecture that divides the computation for different samples among workers. We study distributed machine learning from a different motivation, where the information about the same samples, e.g., users and objects, are owned by several parities that wish to collaborate but do not want to share raw data with each other.

Concepts embody the knowledge of the world and facilitate the cognitive processes of human beings. Mining concepts from web documents and constructing the corresponding taxonomy are core research problems in text understanding and support many downstream tasks such as query analysis, knowledge base construction, recommendation, and search. However, we argue that most prior studies extract formal and overly general concepts from Wikipedia or static web pages, which are not representing the user perspective.

We study the problem of search query inference from web documents, where a short, comprehensive natural language query is inferred from a long article. Search query generation or inference is of great value to search engines and recommenders in terms of locating potential target users and ranking content. Despite being closely related to other NLP tasks like abstract generation and keyword extraction, we point out that search query inference is, in fact, a new problem, in that the generated natural language query, which consists of a few words, is expected to be comprehensive enough to lead to the click-through of the corresponding document. Therefore, query generation requires an accurate inference of query words, as well as a deeper level of understanding on document semantic structures. Toward this end, we propose a novel generative model called the Graph-augmented Sequence to Attention (G-S2A) network. Adopting an Encoder-Decoder architecture, G-S2A incorporates a sentence-level Graph Convolutional Network (GCN), a keyword-level GCN, as well as a hierarchical recurrent neural network (RNN) into the encoder to generate structural document representations. An attentional Transformer decoder is then applied to combine different types of encoded features to generate a target query. On a query-document dataset from a real-world search engine, our model outperforms several neural generative models on a wide range of metrics.

Automatic question generation is an important technique that can improve the training of question answering, help chatbots to start or continue a conversation with humans, and provide assessment materials for educational purposes. Existing neural question generation models are not sufficient mainly due to their inability to properly model the process of how each word in the question is selected, i.e., whether repeating the given passage or being generated from a vocabulary. In this paper, we propose our Clue Guided Copy Network for Question Generation (CGC-QG), which is a sequence-to-sequence generative model with copying mechanism, yet employing a variety of novel components and techniques to boost the performance of question generation. In CGC-QG, we design a multi-task labeling strategy to identify whether a question word should be copied from the input passage or be generated instead, guiding the model to learn the accurate boundaries between copying and generation. Furthermore, our input passage encoder takes as input, among a diverse range of other features, the prediction made by a clue word predictor, which helps identify whether each word in the input passage is a potential clue to be copied into the target question. The clue word predictor is designed based on a novel application of Graph Convolutional Networks onto a syntactic dependency tree representation of each passage, thus being able to predict clue words only based on their context in the passage and their relative positions to the answer in the tree. We jointly train the clue prediction as well as question generation with multi-task learning and a number of practical strategies to reduce the complexity. Extensive evaluations show that our model significantly improves the performance of question generation and out-performs all previous state-of-the-art neural question generation models by a substantial margin.

Modern stream processing frameworks, such as Spark Streaming, are designed to support a wide variety of stream processing applications, such as real-time data analytics in social networks. As the volume of data to be processed increases rapidly, there is a pressing need for processing them across multiple geo-distributed datacenters. However, these frameworks are not designed to take limited and varying inter-datacenter bandwidth into account, leading to longer query latencies. In this paper, we present the design and implementation of an extended Spark Streaming framework to automatically and optimally schedule tasks, select data flow routes and determine micro-batch sizes across geo-distributed datacenters in wide-area networks. To make these decisions, we propose a sparsity-regularized ADMM algorithm to efficiently solve a nonconvex optimization problem, based on readily measurable operating traces. Toward incremental real-world deployment, we take a non-intrusive approach to support flexible routing of micro-batches by adding a new DStream transformation we have developed to the existing Spark Streaming framework. As a result, our implementation can enforce scheduling decisions by modifying application workflows only. We have deployed our implementation on Amazon EC2 with emulated bandwidth constraints, and our experimental results on various types of queries have demonstrated the effectiveness of our proposed framework, as compared to the existing Spark Streaming scheduler and other data-locality-based heuristics.

User action modeling and prediction has long been a topic of importance to recommender systems and user profiling. The quality of the model or accuracy of prediction plays a vital role in related applications like recommendation, advertisement displaying, searching, etc. For large scale systems with a massive number of users, beside the pure prediction performance, there are other practical factors like training and prediction latency, memory overhead, that must be optimized to ensure smooth operation of the system. We propose a fast linear computational framework to handle a vast number of second order crossed features with dimensionality reduction. By leveraging the training and serving system architecture, we shift heavy calculation burden from online serving to offline preprocessing, at the cost of a reasonable amount of memory overhead. The experiments on a 15-day data trace from Tencent MyApp shows that our proposed framework can achieve comparable prediction performance to much complex models like the field-aware factorization machine (FFM) while being served in 2 ms with a reasonable amount of memory overhead.

A large number of deep learning models have been proposed for the text matching problem, which is at the core of various typical natural language processing (NLP) tasks. However, existing deep models are mainly designed for the semantic matching between a pair of short texts, such as paraphrase identification and question answering, and do not perform well on the task of relevance matching between short-long text pairs. This is partially due to the fact that the essential characteristics of short-long text matching have not been well considered in these deep models. More specifically, these methods fail to handle extreme length discrepancy between text pieces and neither can they fully characterize the underlying structural information in long text documents. In this paper, we are especially interested in relevance matching between a piece of short text and a long document, which is critical to problems like query-document matching in information retrieval and web searching. To extract the structural information of documents, an undirected graph is constructed, with each vertex representing a keyword and the weight of an edge indicating the degree of interaction between keywords. Based on the keyword graph, we further propose a Multiresolution Graph Attention Network to learn multi-layered representations of vertices through a Graph Convolutional Network (GCN), and then match the short text snippet with the graphical representation of the document with an attention mechanism applied over each layer of the GCN. Experimental results on two datasets demonstrate that our graph approach outperforms other state-of-the-art deep matching models.

Operating at a large scale, data analytics has become an essential tool for gaining insights from operational data, such as user online activities. With the volume of data growing exponentially, data analytic jobs have expanded from a single datacenter to multiple geographically distributed datacenters. Unfortunately, designed originally for a single datacenter, the software stack that supports data analytics is oblivious to on-the-fly resource variations on inter-datacenter networks, which negatively affects the performance of analytic queries. Existing solutions that optimize query execution plans before their execution are not able to quickly adapt to resource variations at query runtime.

Short Text Matching plays an important role in many natural language processing tasks such as information retrieval, question answering, and conversational system. Conventional text matching methods rely on predefined templates and rules, which are not applicable to short text with limited numebr of words and limit their ability to generalize to unobserved data. Many recent efforts have been made to apply deep neural network models to natural language processing tasks, which reduces the cost of feature engineering. In this paper, we present the design of Multi-Channel Information Crossing , a multi-channel convolutional neural network model for text matching, with additional attention mechanisms from sentence and text semantics. MIX compares text snippets at varied granularities to form a series of multi-channel similarity matrices, which are crossed with another set of carefully designed attention matrices to expose the rich structures of sentences to deep neural networks. We implemented MIX and deployed the system on Tencent's Venus distributed computation platform. Thanks to carefully engineered multi-channel information crossing, evaluation results suggest that MIX outperforms a wide range of state-of-the-art deep neural network models by at least 11.1% in terms of the normalized discounted cumulative gain ([email protected]), on the English WikiQA dataset. Moreover, we also performed online A/B tests with real users on the search service of Tencent QQ Browser. Results suggest that MIX raised the number of clicks on the returned results by 5.7%, due to an increased accuracy in query-document matching, which demonstrates the superior performance of MIX in production environments.

Semantic matching of natural language sentences or identifying the relationship between two sentences is a core research problem underlying many natural language tasks. Depending on whether training data is available, prior research has proposed both unsupervised distance-based schemes and supervised deep learning schemes for sentence matching. However, previous approaches either omit or fail to fully utilize the ordered, hierarchical, and flexible structures of language objects, as well as the interactions between them. In this paper, we propose Hierarchical Sentence Factorization---a technique to factorize a sentence into a hierarchical representation, with the components at each different scale reordered into a "predicate-argument" form. The proposed sentence factorization technique leads to the invention of: 1) a new unsupervised distance metric which calculates the semantic distance between a pair of text snippets by solving a penalized optimal transport problem while preserving the logical relationship of words in the reordered sentences, and 2) new multi-scale deep learning models for supervised semantic training, based on factorized sentence hierarchies. We apply our techniques to text-pair similarity estimation and text-pair relationship classification tasks, based on multiple datasets such as STSbenchmark, the Microsoft Research paraphrase identification (MSRP) dataset, the SICK dataset, etc. Extensive experiments show that the proposed hierarchical sentence factorization can be used to significantly improve the performance of existing unsupervised distance-based metrics as well as multiple supervised deep learning models based on the convolutional neural network (CNN) and long short-term memory (LSTM).

Matrix factorization is a popular approach to solving matrix estimation problems based on partial observations. Existing matrix factorization is based on least squares and aims to yield a low-rank matrix to interpret the conditional sample means given the observations. However, in many real applications with skewed and extreme data, least squares cannot explain their central tendency or tail distributions, yielding undesired estimates. In this paper, we propose expectile matrix factorization by introducing asymmetric least squares, a key concept in expectile regression analysis, into the matrix factorization framework. We propose an efficient algorithm to solve the new problem based on alternating minimization and quadratic programming. We prove that our algorithm converges to a global optimum and exactly recovers the true underlying low-rank matrices when noise is zero. For synthetic data with skewed noise and a real-world dataset containing web service response times, the proposed scheme achieves lower recovery errors than the existing matrix factorization method based on least squares in a wide range of settings.

We describe our experience of implementing a news content organization system at Tencent that discovers events from vast streams of breaking news and evolves news story structures in an online fashion. Our real-world system has distinct requirements in contrast to previous studies on topic detection and tracking (TDT) and event timeline or graph generation, in that we 1) need to accurately and quickly extract distinguishable events from massive streams of long text documents that cover diverse topics and contain highly redundant information, and 2) must develop the structures of event stories in an online manner, without repeatedly restructuring previously formed stories, in order to guarantee a consistent user viewing experience. In solving these challenges, we propose Story Forest, a set of online schemes that automatically clusters streaming documents into events, while connecting related events in growing trees to tell evolving stories. We conducted extensive evaluation based on 60 GB of real-world Chinese news data, although our ideas are not language-dependent and can easily be extended to other languages, through detailed pilot user experience studies. The results demonstrate the superior capability of Story Forest to accurately identify events and organize news text into a logical structure that is appealing to human readers, compared to multiple existing algorithm frameworks.

Erasure coding has been used in storage systems to enhance data durability at a lower storage overhead. However, these systems suffer from long access latency tails due to a lack of flexible load balancing mechanisms and passively launched degraded reads when the original storage node of the requested data becomes a hotspot. We provide a new perspective to load balancing in coded storage systems by proactively and intelligently launching degraded reads and propose a variety of schemes to make optimal decisions either per request or across requests statistically. Experiments on a 98-machine cluster based on the request traces of 12 million objects collected from Windows Azure Storage (WAS) show that our schemes can reduce the median latency by 44.7% and the 95th-percentile tail latency by 77.8% in coded storage systems.

In this paper, we study a new type of spatial sparse recovery problem, that is to infer the fine-grained spatial distribution of certain density data in a region only based on the aggregate observations recorded for each of its subregions. One typical example of this spatial sparse recovery problem is to infer spatial distribution of cellphone activities based on aggregate mobile traffic volumes observed at sparsely scattered base stations. We propose a novel Constrained Spatial Smoothing (CSS) approach, which exploits the local continuity that exists in many types of spatial data to perform sparse recovery via finite-element methods, while enforcing the aggregated observation constraints through an innovative use of the ADMM algorithm. We also improve the approach to further utilize additional geographical attributes. Extensive evaluations based on a large dataset of phone call records and a demographical dataset from the city of Milan show that our approach significantly outperforms various state-of-the-art approaches, including Spatial Spline Regression (SSR).

We study the problem of personalized Quality of Service (QoS) estimation for web services. State-of-the-art methods use matrix factorization or collaborative prediction to estimate web service response times and throughput for each user based on partial measurements collected from past invocations. We point out that in reality, both the response times and through-put of web services follow highly skewed distributions. In this case, the conditional mean QoS estimates generated by traditional matrix completion approaches can be heavily biased toward a few outliers, leading to poor web service recommendation performance. In this paper, we propose the Quantile Matrix Factorization (QMF) technique for web service recommendation by introducing quantile regression into the matrix factorization framework. We propose a novel and efficient algorithm based on Iterative Reweighted Least Squares (IRLS) to solve the QMF problem involving a non-smooth objective function. We further extend the proposed QMF approach to take into account user and service side attributes. Extensive evaluation based on a large-scale QoS dataset has shown that our schemes significantly outperform various state-of-the-art web service QoS estimation schemes in terms of personalized recommendation performance.

Virtualized datacenter networks have been deployed in production platforms, e.g., Amazon VPC and VMware's NVP, to offer the flexibility of network management to enterprise-level clients. A common characteristic of these platforms is that they adopt software switches, such as Open vSwitch (OvS), instead of hardware switches to transfer data between VMs. Although group communication is common in enterprise applications, the unique characteristics of software switches have posed new challenges to the design of multicast protocols. How logical multicast can be optimally performed with software switches is still not well understood. In this paper, we observe that unlike hardware switches, the per-stream output rate in a software switch critically depends on the packet processing overhead of flow cloning. We study the optimal OvS multicast topology with or without the help of additional dedicated software switches called service nodes, and formulate the throughput maximization as a new class of degree-supervised combinatorial graph problems due to the presence of flow cloning costs. We propose a lineartime optimal solution that translates into simple forwarding rules installed at each software switch. Through emulation-based OvS profiling and extensive simulation results, we demonstrate that our proposed logical multicast solutions can significantly improve session throughput with the ability to handle load balancing and latency issues, as compared to the state-of-the-art in the literature.

Network latency prediction is important for server selection and quality-of-service estimation in real-time applications on the Internet. Traditional network latency prediction schemes attempt to estimate the latencies between all pairs of nodes in a network based on sampled round-trip times, through either Euclidean embedding or matrix factorization. However, these schemes become less effective in terms of estimating the latencies of personal devices, due to unstable and time-varying network conditions, triangle inequality violation and the unknown ranks of latency matrices. In this paper, we propose a matrix completion approach to network latency estimation. Specifically, we propose a new class of low-rank matrix completion algorithms, which predicts the missing entries in an extracted “network feature matrix” by iteratively minimizing a weighted Schatten-p norm to approximate the rank. Simulations on true low-rank matrices show that our new algorithm achieves better and more robust performance than multiple state-of-the-art matrix completion algorithms in the presence of noise. We further enhance latency estimation based on multiple “frames” of latency matrices measured in the past, and extend the proposed matrix completion scheme to the case of 3-D tensor completion. Extensive performance evaluations driven by real-world latency measurements collected from the Seattle platform show that our proposed approaches significantly outperform various state-of-the-art network latency estimation techniques, especially for networks that contain personal devices.

Data transfers happen frequently in server clusters for software and application deployment, and in parallel computing clusters to transmit intermediate results in batches among servers between computation stages. This paper presents Cooper, an optimized prototype system to speedup multi-batch data transfers among a cluster of servers, leveraging a theoretically proven optimal algorithm called “coded permutation gossip,” which employs a simple random topology control scheme to best utilize bandwidth and decentralized random linear network coding to maximize the useful information transmitted. On a process-level coding-transfer pipeline, we investigate the best block division, batch division and inter-batch scheduling strategies to minimize the broadcast finish time in a realistic setting. For batch-based transfers, we propose a scheduling algorithm with low overhead that overlaps the transfers of consecutive batches and temporarily prioritizes later batches, to further reduce the broadcast finish time. We describe an asynchronous and distributed implementation of Cooper and have deployed it on Amazon EC2 for evaluation. Based on results from real experiments, we show that Cooper can almost double the speed of data transfers in computing clusters, as compared to state-of-the-art content distribution tools like BitTorrent, at a low CPU overhead.

We study the video recommendation problem based on a large amount of user viewing logs instead of explicit ratings. As viewing records are implicitly suggest user preferences, existing matrix factorization methods fail to generate discriminative recommendations based on such one-class positive samples. We propose a scalable approach called separating-plane matrix factorization (SPMF) to make effective recommendations based on positive implicit feedback, with a learning complexity that is comparable to traditional matrix factorization. With extensive offline evaluation in Tencent Data Warehouse (TDW) based on a large amount of data, we show that our approach outperforms a wide range of state-of-the-art methods. We also deployed our system in the QQ Browser App of Tencent and performed online A/B testing with real users. Results suggest that our approach increased the video click through rate by $23% over implicit-feedback collaborative filtering (IFCF), a scheme available in Apache Spark's MLlib.

Online real-estate information systems such as Zillow and Trulia have gained increasing popularity in recent years. One important feature offered by these systems is the online home price estimate through automated data-intensive computation based on housing information and comparative market value analysis. State-of-the-art approaches model house prices as a combination of a latent land desirability surface and a regression from house features. However, by using uniformly damping kernels, they are unable to handle irregularly shaped regions or capture land value discontinuities within the same region due to the existence of implicit sub-communities, which are common in real-world scenarios. In this paper, we explore the novel application of recent advances in spatial functional analysis to house price modeling and propose the Hierarchical Spatial Functional Model (HSFM), which decomposes house values into land desirability at both the global scale and hidden local scales as well as the feature regression component. We propose statistical learning algorithms based on finite-element spatial functional analysis and spatial constrained clustering to train our model. Extensive evaluations based on housing data in a major Canadian city show that our proposed approach can reduce the mean relative house price estimation error down to 6.60%.

Erasure coding has been applied in many cloud storage systems to enhance reliability at a lower storage cost than replication. While a large amount of prior work aims to enhance recovery performance and reliability, the overall access delay in coded storage still needs to be optimized. As most production systems adopt a systematic code and place the original copy of each block on only one server to be read normally, it is harder to balance server loads and more likely to incur latency tails in coded storage than in three-way replication, where a block can be read from any of the 3 servers storing the block. In this paper, we propose to reduce the access latency in coded storage systems by moving blocks with anti-correlated demands onto same servers for statistical load balancing. We formulate the optimal block placement as a problem similar to Min-k-Partition, propose a local block migration scheme, and derive an approximation ratio as a function of demand variation across blocks. Based on request traces from Windows Azure Storage, we demonstrate that our scheme can significantly reduce the access latency with only a few block moves, especially when the request demand is skewed.

Many distributed interactive multimedia applications, such as live video conferencing and video sharing, require each participating client to transmit its captured video stream to other clients via relay servers. We consider connecting multiple clients through multiple relay servers and study the server selection problem from a dense pool of content delivery network edge locations and datacenters to reduce the end-to-end delays between clients. To achieve scalability in the presence of a large number of candidate servers, we formulate server selection as a geometric problem in a delay space instead of in a graph, which turns out to be an extension of the well-known Euclidean k-median problem. We propose practical approximation schemes when using only one or two servers with theoretical worst-case guarantees as well as fast heuristics when using k servers. We demonstrate the benefit of our optimized multiserver selection schemes through extensive evaluation based on real-world traces collected from the PlanetLab and Seattle platforms, containing personal mobile devices as well as real network experiments based on a prototype implementation.

Many video streaming applications operate their geo-distributed services in the cloud, taking advantage of superior connectivities between datacenters to push content closer to users or to relay live video traffic between end users at a higher throughput. In the meantime, inter-datacenter networks also carry high volumes of other types of traffic, including service replication and data backups, e.g., for storage and email services. It is an important research topic to optimally engineer and schedule inter-datacenter traffic, taking into account the stringent latency requirements of video flows when transmitted along inter-datacenter links shared with other types of traffic. Since inter-datacenter networks are usually overprovisioned, unlike prior work that mainly aims to maximize link utilization, we propose a delay-optimized traffic routing scheme to explicitly differentiate path selection for different sessions according to their delay sensitivities, leading to a software-defined inter-datacenter networking overlay implemented at the application layer. We show that our solution can yield sparse path selection by only solving linear programs, and thus, in contrast to prior traffic engineering solutions, does not lead to overly fine-grained traffic splitting, further reducing packet resequencing overhead and the number of forwarding rules to be installed in each forwarding unit. Real-world experiments based on a deployment on six globally distributed Amazon EC2 datacenters have shown that our system can effectively prioritize and improve the delay performance of inter-datacenter video flows at a low cost.

Distributed resource allocation and sharing can often be formulated as a utility maximization problem, with the objective being the sum of user utilities minus a coupled cost. A traditional distributed solution to such problems, called “consistency pricing,” decouples the objective function via dual decomposition, which is then iteratively solved by the subgradient method. However, such gradient-based approaches may require many iterations of message passing to converge, which may not be sufficient in large-scale real-time applications. In this paper, we propose a new fixed-point-like distributed solution to resource sharing problems with coupled objective functions. While preserving the simple pricing interpretation, our approach speeds up convergence by exploiting the structural difference between user utilities and the coupled cost function. We theoretically analyze the asynchronous algorithm convergence conditions based on contraction mapping. Through a detailed case study of cloud bandwidth reservation based on real-world workload traces, we demonstrate the benefits of the proposed algorithm over state-of-the-art distributed optimization techniques including gradient descent, dual decomposition, and ADMM. In addition, we also extend the proposed algorithm to approach a more general class of consensus optimization problems with not only a coupled objective function, but also a certain class of coupled constraints.

We use randomized network coding (RNC) with simple connection topology control to approach the theoretical limit on finish time of disseminating k blocks in a server cluster of n nodes. Unlike prior gossip literature which relies on completely random contact, we prove that with RNC, any receiver selection following a simple permutation rule can achieve a broadcast completion time of k + n and that a time-varying random ring topology achieves a completion time of k + o(k) + O(logn), both with high probability. Since the theoretical limit on finish time is k + [log2n], our simple permutation algorithms achieve absolutely optimal (not only order-optimal) block pipelining for the k blocks. Our results hold for both one-to-all (broadcast) and all-to-all transfers. We demonstrate the usefulness of the proposed organized network coded gossip with an application to content distribution in cluster computing systems like MapReduce, and discuss practical block dividing strategies to hide the negative effect of computation overhead of network coding.

With an increasing popularity of real-time applications, such as live chat and gaming, latency prediction between personal devices including mobile devices becomes an important problem. Traditional approaches recover all-pair latencies in a network from sampled measurements using either Euclidean embedding or matrix factorization. However, these approaches targeting static or mean network latency prediction are insufficient to predict personal device latencies, due to unstable and time-varying network conditions, triangle inequality violation and unknown rank of latency matrices. In this paper, by analyzing latency measurements from the Seattle platform, we propose new methods for both static latency estimation as well as the dynamic estimation problem given 3D latency matrices sampled over time. We propose a distance-feature decomposition algorithm that can decompose latency matrices into a distance component and a network feature component, and further leverage the structured pattern inherent in the 3D sampled data to increase estimation accuracy. Extensive evaluations driven by real-world traces show that our proposed approaches significantly outperform various state-of-the-art latency prediction techniques.

Infrastructure-as-a-Service clouds offer diverse pricing options, including on-demand and reserved instances with various discounts to attract different cloud users. A practical problem facing cloud users is how to minimize their costs by choosing among different pricing options based on their own demands. In this paper, we propose a new cloud brokerage service that reserves a large pool of instances from cloud providers and serves users with price discounts. The broker optimally exploits both pricing benefits of longterm instance reservations and multiplexing gains. We propose dynamic strategies for the broker to make instance reservations with the objective of minimizing its service cost. These strategies leverage dynamic programming and approximation algorithms to rapidly handle large volumes of demand. Our extensive simulations driven by large-scale Google cluster-usage traces have shown that significant price discounts can be realized via the broker.

We use random linear network coding with simple connection topology control to approach the theoretical limit on finish time of disseminating k blocks in a server cluster of n nodes. Unlike existing gossip schemes which rely on completely random contact, we prove that with random linear network coding, any receiver selection following a simple permutation rule can achieve a broadcast finish time of k + n and that a time-varying random permutation topology achieves a finish time of k + o (k) + O (log n), both with high probability. Since the theoretical limit on finish time is k + log2 n, our simple permutation algorithms achieve absolutely optimal (not only order-optimal) block pipelining for k blocks. Our results hold for both one-to-all (broadcast) and all-to-all transfers. We demonstrate the usefulness of the proposed organized network coded gossip with an application to content distribution in cluster computing systems like MapReduce.

This paper proposes new methods for multi-server placement and topology control in multi-party video conferences. Given a large server pool available from CDN infrastructures and datacenter networks, our lightweight methods can rapidly determine the network topology and select the best physical servers to deploy virtualized server instances on, with the objective of minimizing the mean end-to-end delay between clients. We propose D-Grouping, a ping-based clustering algorithm, which is used in combination with convex optimization to determine the network topology and fine-tune server selection. To verify the proposed methods, we present extensive simulation studies based on the ping traces collected from 518 PlanetLab nodes as well as real-world experiment results based on a prototype implementation.

Infrastructure-as-a-Service clouds offer diverse pricing options, including on-demand and reserved instances with various discounts to attract different cloud users. A practical problem facing cloud users is how to minimize their costs by choosing among different pricing options based on their own demands. In this paper, we propose a new cloud brokerage service that reserves a large pool of instances from cloud providers and serves users with price discounts. The broker optimally exploits both pricing benefits of long-term instance reservations and multiplexing gains. We propose dynamic strategies for the broker to make instance reservations with the objective of minimizing its service cost. These strategies leverage dynamic programming and approximate algorithms to rapidly handle large volumes of demand. Our extensive simulations driven by large-scale Google cluster-usage traces have shown that significant price discounts can be realized via the broker.

In modern large-scale systems, fast distributed resource allocation and utility maximization are becoming increasingly important. Traditional solutions to such problems rely on primal/dual decomposition and gradient methods, whose convergence is sensitive to the choice of the stepsize and may not be sufficient to satisfy the requirement of large-scale real-time applications. We propose a new iterative approach to distributed resource allocation in coupled systems. Without complicating message-passing, the new approach is robust to parameter choices and expedites convergence by exploiting problem structures. We theoretically analyze the asynchronous algorithm convergence conditions, and empirically evaluate its benefits in a case of cloud network resource reservation based on real-world data.

Current-generation cloud computing is offered with usage-based pricing, with no bandwidth capacity guarantees, which is however unappealing to bandwidth-intensive applications such as video-on-demand (VoD). We consider a new type of service where VoD providers, such as Netflix and Hulu, make reservations for bandwidth guarantees from the cloud at negotiable prices to support continuous media streaming. We argue that it is beneficial to multiplex such bandwidth reservations in the market using a profit-making broker while controlling the performance risks. We ask the question-in such a market, how much should each VoD provider pay for bandwidth reservation? We prove that the market has a unique Nash equilibrium where the bandwidth reservation price for a VoD provider critically depends on its demand correlation to the market. Real-world traces verify that our theory can significantly lower the market price for cloud bandwidth reservation.

There has been a recent trend that video-on-demand (VoD) providers such as Netflix are leveraging resources from cloud services for multimedia streaming. In this paper, we consider the scenario that a VoD provider can make reservations for bandwidth guarantees from cloud service providers to guarantee the streaming performance in each video channel. We propose a predictive resource auto-scaling system that dynamically books the minimum bandwidth resources from multiple data centers for the VoD provider to match its short-term demand projections. We exploit the anti-correlation between the demands of video channels for statistical multiplexing and for hedging the risk of under-provision. The optimal load direction from channels to data centers is derived with provable performance. We further provide suboptimal solutions that balance bandwidth and storage costs. The system is backed up by a demand predictor that forecasts the demand expectation, volatility and correlations based on learning. Extensive simulations are conducted driven by the workload traces from a commercial VoD system.

In a public cloud, bandwidth is traditionally priced in a pay-as-you-go model. Reflecting the recent trend of augmenting cloud computing with bandwidth guarantees, we consider a novel model of cloud bandwidth allocation and pricing when explicit bandwidth reservation is enabled. We argue that a tenant's utility depends not only on its bandwidth usage, but more importantly on the portion of its demand that is satisfied with a performance guarantee. Our objective is to determine the optimal policy for pricing cloud bandwidth reservations, in order to maximize social welfare, i.e., the sum of the expected profits that can be made by all tenants and the cloud provider, even with the presence of demand uncertainty. The problem turns out to be a large-scale network optimization problem with a coupled objective function. We propose two new distributed solutions --- based on chaotic equation updates and cutting-plane methods --- that prove to be more efficient than existing solutions based on consistency pricing and subgradient methods. In addition, we address the practical challenge of forecasting demand statistics, required by our optimization problem as input. We propose a factor model for near-future demand prediction, and test it on a real-world video workload dataset. All included, we have designed a fully computerized trading environment for cloud bandwidth reservations, which operates effectively at a fine granularity of as small as ten minutes in our trace-driven simulations.

We consider the problem of distributing k blocks from a source to N nodes in a peer-to-peer (P2P) network with both node upload and download capacity constraints. As k scales up, we prove for homogeneous networks that if network coding is allowed, randomly matching senders and receivers in each time slot asymptotically achieves the maximum downloading rate at each node. For heterogeneous networks with network coding allowed, we show that a fair and optimal downloading rate at each node can be asymptotically approached, if in each time slot, each node randomly allocates its upload bandwidth to its receivers that have available download bandwidth. We also give a performance lower bound of the above randomized coded dissemination when both k and N scale under certain conditions. These results demonstrate that with network coding, simple randomized receiver selection and rate allocation suffice to achieve P2P broadcast capacity, forming a theoretical foundation for mesh-based P2P networks with network coding.

Peer-assisted on-demand video streaming services are extremely large-scale distributed systems on the Internet. Automated demand forecast and performance prediction, if implemented, can help with capacity planning and quality control so that sufficient server bandwidth can always be supplied to each video channel without incurring wastage. In this paper, we use time-series analysis techniques to automatically predict the online population, the peer upload and the server bandwidth demand in each video channel, based on the learning of both human factors and system dynamics from online measurements. The proposed mechanisms are evaluated on a large dataset collected from a commercial Internet video-on-demand system.

Video-on-demand (VoD) servers are usually over-provisioned for peak demands, incurring a low average resource efficiency. However, bandwidth shortage may still occur for individual videos as they share and contend for server resources. In this position paper, we propose a predictive workload management system for VoD servers targeting bandwidth. The system draws belief about future demand as well as demand volatility based on demand history using time series forecasting techniques. The prediction enables dynamic and efficient server bandwidth reservation with QoS guarantees. More importantly, we use a hedging technique similar to investment portfolio management and distribute workloads to multiple servers exploiting demand anti-correlation. The proposed system consolidates the workloads, enhances resource utilization, while in the meantime effectively controlling risk of server overload. The proposed methods are evaluated based on real-world VoD traces.

Bandwidth usage in large-scale Video on Demand (VoD) systems varies rapidly over time, due to unpredictable dynamics in user demand and network conditions. Such bandwidth volatility makes it hard to provision the exact amount of server resources that matches the demand in each video channel, posing significant challenges to achieving quality assurance and efficient resource allocation at the same time. In this paper, we seek to statistically model time-varying traffic volatility in VoD servers, leveraging heteroscedastic models first used to interpret economic time series, with the goal of forecasting not only traffic patterns but also traffic volatility. We present the application of volatility forecast to efficient resource allocation that provides probabilistic service level guarantees to user groups. We also discuss volatility reduction from diversification, and its implications to new strategies for cost-effective server management. Our study is based on monitoring the workload of a large-scale commercial VoD system widely deployed on the Internet.

With random network coding, network nodes between the source and receivers are able to not only relay and replicate data packets, but also code them using randomly generated coding coefficients. From a theoretical perspective, it has been recognized that network coding maximizes the network flow rates in multicast sessions in directed acyclic network graphs. To date, random network coding has seen practical and real-world applications in peer-to-peer (P2P) networks, in which overlay network topologies are formed among participating end hosts, called “peers.” Due to uncertainties and dynamics involved with peer arrivals and departures, these network topologies are usually randomly generated in practice, and are referred to as “random mesh” topologies. Unlike structured topologies such as trees, random mesh topologies are practical to be implemented, and are resilient to the level of volatility typically experienced in peer-to-peer networks. It has been shown, from both theoretical and practical perspectives, that random network coding leads to performance benefits in these peer-to-peer networks with random mesh topologies. This paper presents a survey of existing results with respect to practical applications of random network coding in peer-to-peer networks. We focus on bulk content distribution and media streaming systems, as well as the computational overhead introduced by random network coding in modern off-the-shelf servers and mobile devices. Throughout the paper, we also show theoretical insights on why random network coding may become beneficial in practice.

Most current-generation P2P content distribution protocols use fine-granularity blocks to distribute content to all the peers in a decentralized fashion. Such protocols often suffer from a significant degree of imbalance in block distributions, especially when the users are highly dynamic. As certain blocks become rare or even unavailable, content availability and download efficiency are adversely affected. Randomized network coding may improve block diversity and availability in P2P networks, as coded blocks are equally innovative and useful to peers. However, the computational complexity of network coding mandates that, in reality, network coding needs to be performed within segments, each containing a subset of blocks. In this paper, we quantitatively evaluate how network coding may improve content availability, block diversity, and download performance in the presence of churn, as the number of blocks in each segment for coding varies. Based on stochastic models and a differential equation approach, we explore the fundamental tradeoff between the resilience gain of network coding to peer dynamics and its inherent coding complexity. We conclude that a small number of blocks in each segment is sufficient to realize the major benefits of network coding, with acceptable coding cost.

There exists a certain level of ambiguity regarding whether network coding can further improve download performance in P2P content distribution systems, as compared to commonly applied heuristics such as rarest first protocols. In this paper, we revisit the problem of broadcasting multiple data blocks from a single source in an overlay network using gossip-like protocols. Our new finding reveals that the marginal benefit of network coding critically depends on the dynamics of network topologies. We show that although network coding is optimal as a block selection mechanism, simple non-coding protocols are close to optimal in complete and random graphs, leading to marginal benefits of network coding. However, network coding demonstrates salient benefits in clustered and time-varying topologies, which are common in real-world systems with ISP-locality mechanisms implemented. Through both theoretical analysis and simulation results, we unveil the underlying reasons behind discrepancies in the power of network coding under different scenarios.

Quality control and resource optimization are challenging problems in peer-assisted video streaming systems, due to their large scales and unreliable peer behavior. Such systems are also prone to per- formance degradation in the event of drastic demand changes, such as flash crowds and large-scale simultaneous peer departures. In this paper, we demonstrate the deficiency of state-of-the-art video streaming systems by analyzing real-world traces from UUSee, a popular commercial P2P media streaming system based in China, during the 2008 Beijing Olympics. We show how simple machine learning techniques combined with periodic collection of statistics can be used for automated monitoring and diagnosis of peer-assisted video streaming systems. With such a framework, it is possible to es- timate performance given certain resource usage patterns, making resource utilization more efficient. It also enables the prediction of large-scale performance degradation due to irregular demand pat- terns. #e effectiveness of our proposed framework is validated with extensive trace-driven evaluations.

In most large-scale peer-to-peer (P2P) applications, it is necessary to collect vital statistics data - sometimes referred to as logs - from up to millions of peers. Traditional solutions involve sending large volumes of such data to centralized logging servers, which are not scalable. In addition, they may not be able to retrieve statistics data from departed peers in dynamic peer-to-peer systems. In this paper, we solve this dilemma through an indirect collection mechanism that distributes data using random network coding across the network, from which servers proactively pull such statistics. By buffering data in a decentralized fashion with only a small portion of peer resources, we show that our new mechanism provides a "buffering" zone and a "smoothing" factor to collect large volumes of statistics, with appropriate resilience to peer dynamics and scalability to a large peer population.

Most current-generation P2P content distribution protocols use line-granularity blocks to distribute content to all the peers in a decentralized fashion. Such protocols often suffer from a significant degree of imbalance in block distributions, such that certain blocks become rare or even unavailable, adversely affecting content availability. It has been pointed out that randomized network coding may improve block availability in P2P networks, as coded blocks are equally innovative and useful to peers. However, the computational complexity of network coding mandates that, in reality, network coding needs to be performed within segments, each containing a subset of blocks. In this paper, using both theoretical analysis and simulations, we quantitatively evaluate how segment-based network coding may improve resilience to peer dynamics and content availability. The objective of this paper is to explore the fundamental tradeoff between the resilience gain of network coding and its inherent coding complexity. We introduce a differential equations approach to quantify the resilience gain of network coding as a function of the number of blocks in a segment, as well as various other tunable parameters. We conclude that a small number of blocks in each segment is sufficient to realize the major benefits of network coding, with acceptable coding complexity.