Autonomous web agents powered by large language models (LLMs) show strong potential for performing goal-oriented tasks such as information retrieval, report generation, and online transactions.These agents mark a key step toward practical embodied reasoning in open web environments.However, existing approaches remain limited in reasoning depth and efficiency: vanilla linear methods fail at multistep reasoning and lack effective backtracking, while other search strategies are coarse-grained and computationally costly.We introduce Branch-and-Browse, a fine-grained web agent framework that unifies structured reasoningacting, contextual memory, and efficient execution.It (i) employs explicit subtask management with tree-structured exploration for controllable multi-branch reasoning, (ii) bootstraps exploration through efficient web state replay with background reasoning, and (iii) leverages a page action memory to share explored actions within and across sessions.On the WebArena benchmark, Branch-and-Browse achieves a task success rate of 35.8% and reduces execution time by up to 40.4% relative to state-of-the-art methods.These results demonstrate that Branch-and-Browse is a reliable and efficient framework for LLM-based web agents.Code is available at https://github.com/ SymbioticLab/Branch-and-Browse.

Yuping Lin, Zitao Li, Yue Xing, Pengfei He, Yingqian Cui, Yaliang Li, Bolin Ding, Jingren Zhou, Jiliang Tang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Leyi Pan, Shuchang Tao, Yunpeng Zhai, Zheyu Fu, Liancheng Fang, Minghua He, Lingzhe Zhang, Zhaoyang Liu, Bolin Ding, Aiwei Liu, Lijie Wen. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Given a table T in a database and a question Q in natural language, the table question answering (TQA) task aims to return an accurate answer to Q based on the content of T. Recent state-of-the-art solutions leverage large language models (LLMs) to obtain high-quality answers. However, most rely on proprietary, large-scale LLMs with costly API access, posing a significant financial barrier. This paper instead focuses on TQA with smaller, open-weight LLMs that can run on a desktop or laptop. This setting is challenging, as such LLMs typically have weaker capabilities than large proprietary models, leading to substantial performance degradation with existing methods. We observe that a key reason for this degradation is that prior approaches often require the LLM to solve a highly sophisticated task using long, complex prompts, which exceed the capabilities of small open-weight LLMs. Motivated by this observation, we present Orchestra, a multi-agent approach that unlocks the potential of accessible LLMs for high-quality, cost-effective TQA. Orchestra coordinates a group of LLM agents, each responsible for a relatively simple task, through a structured, layered workflow to solve complex TQA problems -- akin to an orchestra. By reducing the prompt complexity faced by each agent, Orchestra significantly improves output reliability. We implement Orchestra on top of AgentScope, an open-source multi-agent framework, and evaluate it on multiple TQA benchmarks using a wide range of open-weight LLMs. Experimental results show that Orchestra achieves strong performance even with small- to medium-sized models. For example, with Qwen2.5-14B, Orchestra reaches 72.1% accuracy on WikiTQ, approaching the best prior result of 75.3% achieved with GPT-4; with larger Qwen, Llama, or DeepSeek models, Orchestra outperforms all prior methods and establishes new state-of-the-art results across all benchmarks.

To leverage the advantages of LLM in addressing challenges in the Text-to-SQL task, we present XiYan-SQL, an innovative framework effectively generating and utilizing multiple SQL candidates. It consists of three components: 1) a Schema Filter module filtering and obtaining multiple relevant schemas; 2) a multi-generator ensemble approach generating multiple high-quality and diverse SQL queries; 3) a selection model with a candidate reorganization strategy implemented to obtain the optimal SQL query. Specifically, for the multi-generator ensemble, we employ a multi-task fine-tuning strategy to enhance the capabilities of SQL generation models for the intrinsic alignment between SQL and text, and construct multiple generation models with distinct generation styles by fine-tuning across different SQL formats. The experimental results and comprehensive analysis demonstrate the effectiveness and robustness of our framework. Overall, XiYan-SQL achieves a new SOTA performance of 75.63% on the notable BIRD benchmark, surpassing all previous methods. It also attains SOTA performance on the Spider test set with an accuracy of 89.65%.

Hanxing Ding, Shuchang Tao, Liang Pang, Zihao Wei, Jinyang Gao, Bolin Ding, Huawei Shen, Xueqi Cheng. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

There has been a surge of interest regarding language adaptation of Large Language Models (LLMs) to enhance the processing of texts in low-resource languages. While traditional language models have seen extensive research on language transfer, modern LLMs still necessitate further explorations in language adaptation. In this paper, we present a systematic review of the language adaptation process for LLMs, including vocabulary expansion, continued pre-training, and instruction fine-tuning, which focuses on empirical studies conducted on LLaMA2 and discussions on various settings affecting the model’s capabilities. This study provides helpful insights covering the entire language adaptation process, and highlights the compatibility and interactions between different steps, offering researchers a practical guidebook to facilitate the effective adaptation of LLMs across different languages.

In-context learning (ICL), a predominant trend in instruction learning, aims at enhancing the performance of large language models by providing clear task guidance and examples, improving their capability in task understanding and execution. This paper investigates ICL on Large Vision-Language Models (LVLMs) and explores the policies of multi-modal demonstration selection. Existing research efforts in ICL face significant challenges: First, they rely on pre-defined demonstrations or heuristic selecting strategies based on human intuition, which are usually inadequate for covering diverse task requirements, leading to sub-optimal solutions; Second, individually selecting each demonstration fails in modeling the interactions between them, resulting in information redundancy. Unlike these prevailing efforts, we propose a new exploration-exploitation reinforcement learning framework, which explores policies to fuse multi-modal information and adaptively select adequate demonstrations as an integrated whole. The framework allows LVLMs to optimize themselves by continually refining their demonstrations through self-exploration, enabling the ability to autonomously identify and generate the most effective selection policies for in-context learning. Experimental results verify the superior performance of our approach on four Visual Question-Answering (VQA) datasets, demonstrating its effectiveness in enhancing the generalization capability of few-shot LVLMs.

High-performance Multimodal Large Language Models (MLLMs) rely heavily on data quality. This study introduces a novel data synthesis method, leveraging insights from contrastive learning and image difference captioning to enhance fine-grained image recognition in MLLMs. By analyzing object differences in detailed regions between similar images, we challenge the model to identify both matching and distinct components. Specifically, our method initially create pairs of similar images that highlight object variations. After that, we introduce a Difference Area Generator for object differences identifying, followed by a Difference Captions Generator for differences describing. The outcome is a high-quality dataset of "object replacement" samples, named Img-Diff, which can be expanded as needed due to its automation. We use the generated dataset to fine-tune state-of-the-art (SOTA) MLLMs such as InternVL2, yielding comprehensive improvements across numerous image difference and Visual Question Answering tasks. For instance, the trained models notably surpass the SOTA models GPT-4V and Gemini on the MMVP benchmark. Additionally, we conduct thorough evaluations to confirm the dataset’s diversity, quality, and robustness, presenting several insights on the synthesis of such a contrastive dataset. We release our codes and dataset to encourage further research on multimodal data synthesis and MLLMs’ fundamental capabilities for image understanding.

Vertical Federated Learning (VFL) has emerged as a popular machine learning paradigm, enabling model training between the data and the task parties with different features about the same user set while preserving data privacy. In a production environment, VFL usually involves one task party and one data party. Fair and economically efficient feature trading is crucial to the commercialization of VFL, where the task party is considered the data consumer who buys the data party's features. However, current VFL feature trading practices often price the data party's data as a whole and assume transactions occur before performing VFL. Neglecting the performance gains resulting from traded features may lead to underpayment and overpayment issues. In this study, we propose a bargaining-based feature trading approach in VFL to facilitate economically efficient transactions. Our model incorporates performance gain-based pricing, taking into account the revenue-based optimization objectives of both parties. We analyze the proposed bargaining model under perfect and imperfect performance information settings, proving the existence of an equilibrium that optimizes the parties' objectives. Moreover, we develop performance gain estimation-based bargaining strategies for imperfect performance information scenarios and discuss potential security concerns and solutions. Experiments on three real-world datasets demonstrate the effectiveness of the proposed bargaining model.

Handling long-context inputs is crucial for large language models (LLMs) in tasks such as extended conversations, document summarization, and many-shot in-context learning. While recent approaches have extended the context windows of LLMs and employed perplexity (PPL) as a standard evaluation metric, PPL has proven unreliable for assessing long-context capabilities. The underlying cause of this limitation has remained unclear. In this work, we provide a comprehensive explanation for this issue. We find that PPL overlooks key tokens, which are essential for long-context understanding, by averaging across all tokens and thereby obscuring the true performance of models in long-context scenarios. To address this, we propose \textbf{LongPPL}, a novel metric that focuses on key tokens by employing a long-short context contrastive method to identify them. Our experiments demonstrate that LongPPL strongly correlates with performance on various long-context benchmarks (e.g., Pearson correlation of -0.96), significantly outperforming traditional PPL in predictive accuracy. Additionally, we introduce \textbf{LongCE} (Long-context Cross-Entropy) loss, a re-weighting strategy for fine-tuning that prioritizes key tokens, leading to consistent improvements across diverse benchmarks. In summary, these contributions offer deeper insights into the limitations of PPL and present effective solutions for accurately evaluating and enhancing the long-context capabilities of LLMs. Code is available at https://github.com/PKU-ML/LongPPL.

Through the collaboration of multiple LLM-empowered agents possessing diverse expertise and tools, multi-agent systems achieve impressive progress in solving real-world problems. Given the user queries, the meta-agents, serving as the brain within multi-agent systems, are required to decompose the queries into multiple sub-tasks that can be allocated to suitable agents capable of solving them, so-called agent-oriented planning. In this study, we identify three critical design principles of agent-oriented planning, including solvability, completeness, and non-redundancy, to ensure that each sub-task can be effectively resolved, resulting in satisfactory responses to user queries. These principles further inspire us to propose AOP, a novel framework for agent-oriented planning in multi-agent systems, leveraging a fast task decomposition and allocation process followed by an effective and efficient evaluation via a reward model. According to the evaluation results, the meta-agent is also responsible for promptly making necessary adjustments to sub-tasks and scheduling. Besides, we integrate a feedback loop into AOP to further enhance the effectiveness and robustness of such a problem-solving process. Extensive experiments demonstrate the advancement of AOP in solving real-world problems compared to both single-agent systems and existing planning strategies for multi-agent systems. The source code is available at https://github.com/lalaliat/Agent-Oriented-Planning

This study addresses the challenge of noise in training datasets for Direct Preference Optimization (DPO), a method for aligning Large Language Models (LLMs) with human preferences. We categorize noise into pointwise noise, which includes low-quality data points, and pairwise noise, which encompasses erroneous data pair associations that affect preference rankings. Utilizing Distributionally Robust Optimization (DRO), we enhance DPO's resilience to these types of noise. Our theoretical insights reveal that DPO inherently embeds DRO principles, conferring robustness to pointwise noise, with the regularization coefficient $\beta$ playing a critical role in its noise resistance. Extending this framework, we introduce Distributionally Robustifying DPO (Dr. DPO), which integrates pairwise robustness by optimizing against worst-case pairwise scenarios. The novel hyperparameter $\beta'$ in Dr. DPO allows for fine-tuned control over data pair reliability, providing a strategic balance between exploration and exploitation in noisy training environments. Empirical evaluations demonstrate that Dr. DPO substantially improves the quality of generated text and response accuracy in preference datasets, showcasing enhanced performance in both noisy and noise-free settings.

The emergence of multimodal large models has advanced artificial intelligence, introducing unprecedented levels of performance and functionality. However, optimizing these models remains challenging due to historically isolated paths of model-centric and data-centric developments, leading to suboptimal outcomes and inefficient resource utilization. In response, we present a new sandbox suite tailored for integrated data-model co-development. This sandbox provides a feedback-driven experimental platform, enabling cost-effective iteration and guided refinement of both data and models. Our proposed “Probe-Analyze-Refine” workflow, validated through practical use cases on multimodal tasks such as image-text pre-training with CLIP, image-to-text generation with LLaVA-like models, and text-to-video generation with DiT-based models, yields transferable and notable performance boosts, such as topping the VBench leaderboard. A comprehensive set of over 100 experiments demonstrated the suite’s usability and extensibility, while also uncovering insights into the interplay between data quality, diversity, model behavior, and computational costs. All codes, datasets, and models are open-sourced to foster future research and applications that would otherwise be infeasible due to the lack of a dedicated co-development infrastructure.

Auction plays a crucial role in many modern trading environments, including online advertising and public resource allocation. As the number of competing bidders increases, learning Bayesian Nash Equilibrium (BNE) in auctions faces significant scalability challenges. Existing methods often experience slow convergence in large-scale auctions. For example, in a classic symmetric auction setting, the convergence rate depends on the number of bidders quadratically. To address this issue, we propose the Approximate Best Response Gradient method, a new approach for learning BNE efficiently in auction games. We leverage an analytic solution for gradient estimation to enable efficient gradient computation during optimization. Moreover, we introduce the Best Response Distance objective, which serves as an upper bound of approximation quality to BNE. By optimizing the new objective, our method is proven to achieve a local convergence rate independent of bidder numbers and circumvent the traditional quadratic complexity in the classic symmetric setting. Extensive experiments across various auction formats demonstrate that our approach accelerates convergence and enhances learning efficiency in complex auction settings.

Preference learning is critical for aligning large language models (LLMs) with human values, with the quality of preference datasets playing a crucial role in this process. While existing metrics primarily assess data quality based on either explicit or implicit reward margins, their single-margin focus often leads to contradictory evaluations for the same data. To address this issue, we propose a new metric of alignment potential, $M_{AP}$, which integrates both margins to quantify the gap from the model’s current implicit reward margin to the target explicit reward margin, thereby estimating the model’s potential to align on the preference data. Empirical results demonstrate that training on the data selected by $M_{AP}$ consistently enhances alignment performance, surpassing existing metrics across different base models and optimization objectives. Furthermore, our method can be extended to self-play data generation frameworks, where we use this metric to identify high-quality data within the self-generated content by LLMs. Under this data generation scenario, our method surpasses current state-of-the-art methods across various training settings and demonstrates continuous improvements with increasing dataset size and training iterations.

Aligning large language models (LLMs) with human preferences requires balancing policy optimization with computational stability. While recent offline methods like DPO and SimPO bypass reinforcement learning’s complexity, they face critical limitations: DPO relies on static reference models that degrade with policy updates, and SimPO assumes a uniform target reward margin that ignores instance-wise preference strength. We propose AlphaDPO, an adaptive preference optimization framework that dynamically reparameterizes the reference distribution to address these issues. Our key innovation lies in an implicit reference model \(\hat{\pi}_{\text{ref}} \propto U(y|x)(\pi_\theta/\pi_{\text{ref}})^\alpha\), which interpolates between policy-driven specialization and uniform exploration while enabling instance-adaptive reward margins. Theoretically, we prove AlphaDPO implicitly controls sequential KL divergence between iterative policy updates, ensuring stability even with poorly calibrated reference models. Empirically, AlphaDPO achieves state-of-the-art performance on AlpacaEval 2 (58.7\% LC win rate) and Arena-Hard (35.7\% win rate) across Mistral2-7B, Llama3-8B, and Gemma2-9B, demonstrating robust alignment without multi-stage training. Our work establishes adaptive reference reparameterization as a principled mechanism for preference optimization.

Federated learning (FL) has emerged as a promising paradigm for fine-tuning foundation models using distributed data in a privacy-preserving manner. Under limited computational resources, clients often find it more practical to fine-tune a selected subset of layers, rather than the entire model, based on their task-specific data. In this study, we provide a thorough theoretical exploration of selective layer fine-tuning in FL, emphasizing a flexible approach that allows the clients to adjust their selected layers according to their local data and resources. We theoretically demonstrate that the layer selection strategy has a significant impact on model convergence in two critical aspects: the importance of selected layers and the heterogeneous choices across clients. Drawing from these insights, we further propose a strategic layer selection method that utilizes local gradients and regulates layer selections across clients. Extensive experiments on both image and text datasets demonstrate the effectiveness of the proposed strategy compared with several baselines, highlighting its advances in identifying critical layers that adapt to the client heterogeneity and training dynamics in FL.

Foundation models demand advanced data processing for their vast, multimodal datasets.However, traditional frameworks struggle with the unique complexities of multimodal data.In response, we present Data-Juicer 2.0, a data processing system backed by 100+ data processing operators spanning text, image, video, and audio modalities, supporting more critical tasks including data analysis, synthesis, annotation, and foundation model post-training.With seamless compatibility and dedicated optimization for popular dataset hubs like Hugging Face and computing engines like Ray, it improves upon its predecessor in terms of usability, efficiency, and programmability.It features an easily accessible user interface layer that supports decoupled Python interactions, RESTful APIs, and conversational commands. Its new runtime layer offers adaptive execution across diverse scales and environments, abstracting away system complexities.Extensive empirical evaluations demonstrate Data-Juicer 2.0's remarkable performance and scalability, highlighting its capability to efficiently process TB-level data with 10k+ CPU cores. The system is publicly available and has been widely adopted in diverse research fields and real-world products such as Alibaba Cloud PAI. We actively maintain the system and share practical insights to foster research and applications of next-generation foundation models.

We propose two simple, principled and practical algorithms that enjoy provable scaling laws for the test-time compute of large language models (LLMs). The first one is a two-stage knockout-style algorithm: given an input problem, it first generates multiple candidate solutions, and then aggregate them via a knockout tournament for the final output. Assuming that the LLM can generate a correct solution with non-zero probability and do better than a random guess in comparing a pair of correct and incorrect solutions, we prove theoretically that the failure probability of this algorithm decays to zero exponentially or by a power law (depending on the specific way of scaling) as its test-time compute grows. The second one is a two-stage league-style algorithm, where each candidate is evaluated by its average win rate against multiple opponents, rather than eliminated upon loss to a single opponent. Under analogous but more robust assumptions, we prove that its failure probability also decays to zero exponentially with more test-time compute. Both algorithms require a black-box LLM and nothing else (e.g., no verifier or reward model) for a minimalistic implementation, which makes them appealing for practical applications and easy to adapt for different tasks. Through extensive experiments with diverse models and datasets, we validate the proposed theories and demonstrate the outstanding scaling properties of both algorithms.

Preference learning has become a common approach in various recent methods for aligning large language models with human values. These methods optimize the preference margin between chosen and rejected responses, subject to certain constraints for avoiding over-optimization. In this paper, we report surprising empirical findings that simple ReLU activation can learn meaningful alignments even using \emph{none} of the following: (i) sigmoid-based gradient constraints, (ii) explicit regularization terms. Our experiments show that over-optimization does exist, but a threshold parameter $\gamma$ plays an essential role in preventing it by dynamically filtering training examples. We further provide theoretical analysis demonstrating that ReLU-based Preference Optimization (RePO) corresponds to the convex envelope of the 0-1 loss, establishing its fundamental soundness. Our ``RePO'' method achieves competitive or superior results compared to established preference optimization approaches. We hope this simple baseline will motivate researchers to rethink the fundamental mechanisms behind preference optimization for language model alignment.

Join discovery is a typical task in data lake research. It finds joinable relationships between columns of different tables which is critical for data integration in schema-less data lakes. On the other hand, LLMs have achieved promising results in many natural language tasks recently. They behave well with the in-context learning ability that only needs a few examples and no fine-tuning, showing great advantages that can be applied to join discovery. However, directly applying NLP prompt generation methods to the table modality may lead to performance decline due to differences in data modalities. It means that strategies useful in NLP scenarios cannot be used in data lakes or obtain limited efficacy, leading to sub-optimal prompts. This is because of the differences in the task definition, data modalities and the availability of labeled samples between the two scenarios. Unsuitable prompts fail to harness the full potential of LLMs and may even mislead them into producing incorrect answers. Therefore, a novel Join Discovery System with Comprehensive and Optimized Prompt Engineering (JD-SCOPE) is proposed to deal with join discovery on data lakes. It first constructs unsupervised examples for demonstrations of the prompt and the validation set for hyper-parameters tuning. Then it explores the prompt template automatically. Meanwhile, JD-SCOPE also ensures stable outputs of LLMs, thus avoiding the impact of randomness. In addition, JD-SCOPE can be extended to semantic join discovery to explore column pairs that are domain correlated but not identical (e.g., country code and language code) for a comprehension understanding of data. JD-SCOPE paves a way to ground data lakes with LLMs and harnesses the knowledge and logical reasoning power of LLMs for join discovery. To evaluate join discovery effectively, the JD-Lake dataset is curated and benchmarked with baselines and a series of popular LLMs. The experiments prove that JD-SCOPE alleviates this migration problem well and empirically outperforms alternatives, showing the superiority of JD-SCOPE.

Recent years have witnessed the rapid development of large language models (LLMs). Multi-modal LLMs (MLLMs) extend modality from text to various domains, attracting widespread attention due to their diverse application scenarios. As LLMs and MLLMs rely on vast amounts of model parameters and data to achieve emergent capabilities, the importance of data is gaining increasing recognition. Reviewing recent data-driven works for MLLMs, we find that the development of models and data is not two separate paths but rather interconnected. Vaster and higher-quality data improve MLLM performance, while MLLMs, in turn, facilitate the development of data. The co-development of multi-modal data and MLLMs requires a clear view of 1) at which development stages of MLLMs specific data-centric approaches can be employed to enhance certain MLLM capabilities, and 2) how MLLMs, using these capabilities, can contribute to multi-modal data in specific roles. To promote data-model co-development for MLLM communities, we systematically review existing works on MLLMs from the data-model co-development perspective.

Incremental query processing is widely used in data warehouses and streaming systems. While many optimization techniques are developed to generate incremental query plans, the scheduling support for incremental processing remains preliminary. Typically, execution is triggered with fixed frequencies specified by the user. In this paper, we propose a novel scheduling problem for incremental query execution under a deadline, assuming the resource has a fluctuating and unforeseen price. We propose two naive solutions as well as a prophet scheduler that foresees the future. We present an end-to-end system Agamotto that models future probabilities offline with a Markov Decision Process (MDP) and makes cost-based and dynamic scheduling decisions online. We show how Agamotto can be extended to handle a workflow of dependent queries, so that they can all incrementally execute in an asynchronous fashion. Experiments show that Agamotto consistently outperforms the naive solutions, and the achieved cost is on average 10x closer to the theoretical lower bound provided by the prophet scheduler.

Abstract Approximate query processing (AQP) plays a critical role in modern data analytics. Although machine learning models are used for AQP, existing methods fail to uncover implicit relationships among the underlying data, the aggregate functions in queries, and the query predicates. In this work, we propose a Graph REpresentation Learning-based AQP model (GRELA for short) for answering queries with multiple aggregate functions. GRELA models the aggregate functions and the query predicates as task and clause nodes respectively in a graph and then learns appropriate node representations via its two modules. In particular, the $$\texttt {Encoder}$$ Encoder module coalesces query predicates and underlying data into the representations of clause nodes. The $$\mathbf {\texttt {Graph}}$$ Graph module bridges task nodes and clause nodes such that each task node can aggregate the information from its neighborhood into its representation. Through the inner products of clause and task representations, GRELA is able to make accurate estimates for queries with multiple aggregate functions. Extensive experimental results verify that GRELA outperforms the state-of-the-art AQP methods on different kinds of datasets.

With the increasing application of deep learning to solve scientific problems in biochemistry, molecular federated learning has become popular due to its ability to offer distributed privacy-preserving solutions. However, most existing molecular federated learning methods rely on joint training with public datasets, which are difficult to obtain in practice. These methods also fail to leverage multi-modal molecular representations effectively. To address the above issues, we propose a novel framework, Federated Heterogeneous Contrastive Distillation (FedHCD), which enables to jointly train global models from clients with heterogeneous data modalities, learning tasks, and molecular models. To aggregate data representations of different modalities in a data-free manner, we design a global multi-modal contrastive strategy to align the representation of clients without public dataset. Utilizing intrinsic characteristics of molecular data in different modalities, we tackle the exacerbation of local model drift and data Non-IIDness caused by multi-modal clients. We introduce a multi-view contrastive knowledge transfer to extract features from atoms, substructures, and molecules, solving the issue of information distillation failure due to dimensional biases in different data modalities. Our evaluations on eight real-world molecular datasets and ablation experiments show that FedHCD outperforms other state-of-the-art FL methods, irrespective of whether or not they use public datasets.

Despite the superior performance, Large Language Models (LLMs) require significant computational resources for deployment and use. To overcome this issue, quantization methods have been widely applied to reduce the memory footprint of LLMs as well as increase the inference rate. However, a major challenge is that low-bit quantization methods often lead to performance degradation. It is important to understand how quantization impacts the capacity of LLMs. Different from previous studies focused on overall performance, this work aims to investigate the impact of quantization on emergent abilities, which are important characteristics that distinguish LLMs from small language models. Specifically, we examine the abilities of in-context learning, chain-of-thought reasoning, and instruction-following in quantized LLMs. Our empirical experiments show that these emergent abilities still exist in 4-bit quantization models, while 2-bit models encounter severe performance degradation on the test of these abilities. To improve the performance of low-bit models, we conduct two special experiments: (1) fine-gained impact analysis that studies which components (or substructures) are more sensitive to quantization, and (2) performance compensation through model fine-tuning. Our work derives a series of important findings to understand the impact of quantization on emergent abilities and sheds light on the possibilities of extremely low-bit quantization for LLMs.

Low-rank adaptation (LoRA) is one of the most popular task-specific parameter-efficient fine-tuning (PEFT) methods on pre-trained language models for its good performance and computational efficiency. LoRA injects a product of two trainable rank decomposition matrices over the top of each frozen pre-trained model module. However, when applied in the setting of privacy-preserving federated learning (FL), LoRA may become unstable due to the following facts: 1) the effects of data heterogeneity and multi-step local updates are non-negligible, 2) additive noise enforced on updating gradients to guarantee differential privacy (DP) can be amplified and 3) the final performance is susceptible to hyper-parameters. A key factor leading to these phenomena is the discordance between jointly optimizing the two low-rank matrices by local clients and separately aggregating them by the central server. Thus, this paper proposes an efficient and effective version of LoRA, Federated Freeze A LoRA (FFA-LoRA), to alleviate these challenges and further halve the communication cost of federated fine-tuning LLMs. The core idea of FFA-LoRA is to fix the randomly initialized non-zero matrices and only fine-tune the zero-initialized matrices. Compared to LoRA, FFA-LoRA is motivated by practical and theoretical benefits in privacy-preserved FL. Our experiments demonstrate that FFA-LoRA provides more consistent performance with better computational efficiency over vanilla LoRA in various FL tasks.

Recent studies have demonstrated the great power of Transformer models for time series forecasting. One of the key elements that lead to the transformer's success is the channel-independent (CI) strategy to improve the training robustness. However, the ignorance of the correlation among different channels in CI would limit the model's forecasting capacity. In this work, we design a special Transformer, i.e., **C**hannel **A**ligned **R**obust Blen**d** Transformer (CARD for short), that addresses key shortcomings of CI type Transformer in time series forecasting. First, CARD introduces a channel-aligned attention structure that allows it to capture both temporal correlations among signals and dynamical dependence among multiple variables over time. Second, in order to efficiently utilize the multi-scale knowledge, we design a token blend module to generate tokens with different resolutions. Third, we introduce a robust loss function for time series forecasting to alleviate the potential overfitting issue. This new loss function weights the importance of forecasting over a finite horizon based on prediction uncertainties. Our evaluation of multiple long-term and short-term forecasting datasets demonstrates that CARD significantly outperforms state-of-the-art time series forecasting methods. The code is available at the following repository: https://github.com/wxie9/CARD.

We present EE-LLM, a framework for large-scale training and inference of early-exit large language models (LLMs). While recent works have shown preliminary evidence for the efficacy of early exiting in accelerating LLM inference, EE-LLM makes a foundational step towards scaling up early-exit LLMs by supporting their training and inference with massive 3D parallelism. Built upon Megatron-LM, EE-LLM implements a variety of algorithmic innovations and performance optimizations tailored to early exiting, including a lightweight method that facilitates backpropagation for the early-exit training objective with pipeline parallelism, techniques of leveraging idle resources in the original pipeline schedule for computation related to early-exit layers, and two approaches of early-exit inference that are compatible with KV caching for autoregressive generation. Our analytical and empirical study shows that EE-LLM achieves great training efficiency with negligible computational overhead compared to standard LLM training, as well as outstanding inference speedup without compromising output quality. To facilitate further research and adoption, we release EE-LLM at https://github.com/pan-x-c/EE-LLM.

Auction games have been widely used in plenty of trading environments such as online advertising and real estate. The complexity of real-world scenarios, characterized by diverse auction mechanisms and bidder asymmetries, poses significant challenges in efficiently solving for equilibria. Traditional learning approaches often face limitations due to their specificity to certain settings and high resource demands. Addressing this, we introduce *Auctionformer*, an efficient transformer-based method to solve equilibria of diverse auctions in a unified framework. Leveraging the flexible tokenization schemes, Auctionformer translates varying auction games into a standard token series, making use of renowned Transformer architectures. Moreover, we employ Nash error as the loss term, sidestepping the need for underlying equilibrium solutions and enabling efficient training and inference. Furthermore, a few-shot framework supports adaptability to new mechanisms, reinforced by a self-supervised fine-tuning approach. Extensive experimental results affirm the superior performance of Auctionformer over contemporary methods, heralding its potential for broad real-world applications.

Pre-trained large language models (LLMs) need fine-tuning to improve their responsiveness to natural language instructions. Federated learning offers a way to fine-tune LLMs using the abundant data on end devices without compromising data privacy. Most existing federated fine-tuning methods for LLMs rely on parameter-efficient fine-tuning techniques, which may not reach the performance height possible with full-parameter tuning. However, federated full-parameter tuning of LLMs is a non-trivial problem due to the immense communication cost. This work introduces FedKSeed that employs zeroth-order optimization with a finite set of random seeds. It significantly reduces transmission requirements between the server and clients to just a few random seeds and scalar gradients, amounting to only a few thousand bytes, making federated full-parameter tuning of billion-sized LLMs possible on devices. Building on it, we develop a strategy enabling probability-differentiated seed sampling, prioritizing perturbations with greater impact on model accuracy. Experiments across six scenarios with various LLMs, datasets and data partitions demonstrate that our approach outperforms existing federated LLM fine-tuning methods in both communication efficiency and zero-shot generalization.

In the foundation models era, efficiently processing multi-modal data is crucial. This tutorial covers key techniques for multi-modal data processing and introduces the open-source Data-Juicer system, designed to tackle the complexities of data variety, quality, and scale. Participants will learn how to use Data-Juicer's operators and tools for formatting, mapping, filtering, deduplicating, and selecting multi-modal data efficiently and effectively. They will also be familiar with the Data-Juicer Sandbox Lab, where users can easily experiment with diverse data recipes that represent methodical sequences of operators and streamline the creation of scalable data processing pipelines. This experience solidifies the concepts discussed, as well as provides a space for innovation and exploration, highlighting how data recipes can be optimized and deployed in high-performance distributed environments.

Large language models (LLMs) have demonstrated great capabilities in various natural language understanding and generation tasks.These pre-trained LLMs can be further improved for specific downstream tasks by fine-tuning.However, the adoption of LLM in real-world applications can be hindered by privacy concerns and the resource-intensive nature of model training and fine-tuning.When multiple entities have similar interested tasks but cannot directly share their local data due to privacy regulations, federated learning (FL) is a mainstream solution to leverage the data of different entities.Besides avoiding direct data sharing, FL can also achieve rigorous data privacy protection, model intelligent property protection, and model customization via composition with different techniques.Despite the aforementioned advantages of FL, fine-tuning LLMs in FL settings still lacks adequate support from the existing frameworks and, therefore, faces challenges in optimizing the consumption of significant communication and computational resources, preparing various data for different tasks, and satisfying diverse information protection demands.

Large language models (LLMs) show amazing performance on many domain-specific tasks after fine-tuning with some appropriate data. However, many domain-specific data are privately distributed across multiple owners. Thus, this dilemma raises the interest in how to perform LLM fine-tuning in federated learning (FL). However, confronted with limited computation and communication capacities, FL clients struggle to fine-tune an LLM effectively. To this end, we introduce FedBiOT, a resource-efficient LLM fine-tuning approach to FL. Specifically, our method involves the server generating a compressed LLM and aligning its performance with the full model. Subsequently, the clients fine-tune a lightweight yet important part of the compressed model, referred to as an adapter. Notice that as the server has no access to the private data owned by the clients, the data used for alignment by the server has a different distribution from the one used for fine-tuning by clients. We formulate the problem into a bi-level optimization problem to minimize the negative effect of data discrepancy and derive the updating rules for the server and clients. We conduct extensive experiments on LLaMA-2, empirically showing that the adapter has exceptional performance when reintegrated into the global LLM. The results also indicate that the proposed FedBiOT significantly reduces resource consumption compared to existing benchmarks, all while achieving comparable performance levels.

Data synthesis is a promising solution to share data for various downstream analytic tasks without exposing raw data. However, without a theoretical privacy guarantee, a synthetic dataset would still leak some sensitive information in raw data. As a countermeasure, differential privacy is widely adopted to safeguard data synthesis by strictly limiting the released information. This technique is advantageous yet presents significant challenges in the vertical federated setting, where data attributes are distributed among different data parties. The main challenge lies in maintaining privacy while efficiently and precisely reconstructing the correlation between attributes. In this paper, we propose a novel algorithm called VertiMRF, designed explicitly for generating synthetic data in the vertical setting and providing differential privacy protection for all information shared from data parties. We introduce techniques based on the Flajolet-Martin (FM) sketch for encoding local data satisfying differential privacy and estimating cross-party marginals. We provide theoretical privacy and utility proof for encoding in this multi-attribute data. Collecting the locally generated private Markov Random Field (MRF) and the sketches, a central server can reconstruct a global MRF, maintaining the most useful information. Two critical techniques introduced in our VertiMRF are dimension reduction and consistency enforcement, preventing the noise of FM sketch from overwhelming the information of attributes with large domain sizes when building the global MRF. These two techniques allow flexible and inconsistent binning strategies of local private MRF and the data sketching module, which can preserve information to the greatest extent. We conduct extensive experiments on four real-world datasets to evaluate the effectiveness of VertiMRF. End-to-end comparisons demonstrate the superiority of VertiMRF.

Behemoth graphs are often fragmented and separately stored by multiple data owners as distributed subgraphs in many realistic applications. Without harming data privacy, it is natural to consider the subgraph federated learning (subgraph FL) scenario, where each local client holds a subgraph of the entire global graph, to obtain globally generalized graph mining models. To overcome the unique challenge of incomplete information propagation on local subgraphs due to missing cross-subgraph neighbors, previous works resort to the augmentation of local neighborhoods through the joint FL of missing neighbor generators and GNNs. Yet their technical designs have profound limitations regarding the utility, efficiency, and privacy goals of FL. In this work, we propose FedDEP to comprehensively tackle these challenges in subgraph FL. FedDEP consists of a series of novel technical designs: (1) Deep neighbor generation through leveraging the GNN embeddings of potential missing neighbors; (2) Efficient pseudo-FL for neighbor generation through embedding prototyping; and (3) Privacy protection through noiseless edge-local-differential-privacy. We analyze the correctness and efficiency of FedDEP, and provide theoretical guarantees on its privacy. Empirical results on four real-world datasets justify the clear benefits of proposed techniques.

As the prevalence of data analysis grows, safeguarding data privacy has become a paramount concern. Consequently, there has been an upsurge in the development of mechanisms aimed at privacy-preserving data analyses. However, these approaches are task-specific; designing algorithms for new tasks is a cumbersome process. As an alternative, one can create synthetic data that is (ideally) devoid of private information. This paper focuses on privacy-preserving data synthesis (PPDS) by providing a comprehensive overview, analysis, and discussion of the field. Specifically, we put forth a master recipe that unifies two prominent strands of research in PPDS: statistical methods and deep learning (DL)-based methods. Under the master recipe, we further dissect the statistical methods into choices of modeling and representation, and investigate the DL-based methods by different generative modeling principles. To consolidate our findings, we provide comprehensive reference tables, distill key takeaways, and identify open problems in the existing literature. In doing so, we aim to answer the following questions: What are the design principles behind different PPDS methods? How can we categorize these methods, and what are the advantages and disadvantages associated with each category? Can we provide guidelines for method selection in different real-world scenarios? We proceed to benchmark several prominent DL-based methods on the task of private image synthesis and conclude that DP-MERF is an all-purpose approach. Finally, upon systematizing the work over the past decade, we identify future directions and call for actions from researchers.

Recently, how the model performance scales with the training sample size has been extensively studied for large models on vision and language related domains. Nevertheless, the ubiquitous node classification tasks on web-scale graphs were ignored, where the traits of these tasks, such as non-IIDness and transductive setting, are likely to cause different scaling laws and motivate novel techniques to beat the law. Therefore, we first explore the neural scaling law for node classification tasks on three large-scale graphs. Then, we benchmark several state-of-the-art data pruning methods on these tasks, not only validating the possibility of improving the original unsatisfactory power law but also gaining insights into a hard-and-representative principle on picking an effective subset of training nodes. Moreover, we leverage the transductive setting to propose a novel data pruning method, which instantiates our principle in a test set-targeted manner. Our method consistently outperforms related methods on all three datasets. Meanwhile, we utilize a PAC-Bayesian framework to analyze our method, extending prior results to account for both hardness and representativeness. In addition to a promising way to ease GNN training on web-scale graphs, our study offers knowledge of the relationship between training nodes and GNN generalization.

The problem of estimating data properties using sampling frequency histograms has attracted extensive interest in the area of databases. The properties include the number of distinct values (NDV), entropy, and so on. In the field of databases, property estimation is fundamental to complex applications. For example, NDV estimation is the foundation of query optimization, and entropy estimation is the foundation of data compression. Among them, methods originating from statistics exhibit desirable theoretical guarantees but rely on specific assumptions about the distribution of data, resulting in poor performance in real-world applications. Learning-based methods, which use information from training data, are adaptable in the real world but often lack theoretical guarantees or explainability. In addition, a unified framework for estimating these frequency-based estimators with machine learning is lacking. Given the aforementioned challenges, it is natural to wonder if a unified framework with theoretical guarantees can be established for property estimation. The recent literature has presented theoretical studies that propose estimation frameworks based on polynomials. These studies also prove estimation errors with respect to the sample size. Motivated by the above polynomial estimation framework, we propose a learning-based estimation framework with polynomial approximation, which aims to learn the coefficients of the polynomial, providing theoretical guarantees to the learning framework. Through comprehensive experiments on both synthetic and real-world datasets for estimating various data properties like NDV, entropy, and power sum, our results show the superiority of our algorithms over previous estimators.

Large language models (LLMs) have emerged as a new paradigm for Text-to-SQL task. However, the absence of a systematical benchmark inhibits the development of designing effective, efficient and economic LLM-based Text-to-SQL solutions. To address this challenge, in this paper, we first conduct a systematical and extensive comparison over existing prompt engineering methods, including question representation, example selection and example organization, and with these experimental results, we elaborate their pros and cons. Based on these findings, we propose a new integrated solution, named DAIL-SQL, which refreshes the Spider leaderboard with 86.6% execution accuracy and sets a new bar. To explore the potential of open-source LLM, we investigate them in various scenarios, and further enhance their performance with supervised fine-tuning. Our explorations highlight open-source LLMs' potential in Text-to-SQL, as well as the advantages and disadvantages of the supervised fine-tuning. Additionally, towards an efficient and economic LLM-based Text-to-SQL solution, we emphasize the token efficiency in prompt engineering and compare the prior studies under this metric. We hope that our work provides a deeper understanding of Text-to-SQL with LLMs, and inspires further investigations and broad applications.

Many machine learning techniques rely on plenty of training data. However, data are often possessed unequally by different entities, with a large proportion of data being held by a small number of data-rich entities. It can be challenging to incentivize data-rich entities to help train models with others via federated learning (FL) if there are no additional benefits. This difficulty arises because these data-rich entities cannot enjoy the revenue increment generated from the improved performances on tasks controlled by data-limited entities. In this paper, we investigate pricing mechanisms through auctions for FL, focusing on auction scenarios with one data seller and some data-limited entities as buyers. The mechanisms aim to account for buyers' performance gains from the FL and provide equitable monetary compensation to the data seller. We first formulate the task as a performance-based auction mechanism design problem and offer a template that can accommodate multiple kinds of auctions with different desiderata. Utilizing this template, we instantiate different truthful strategies with different goals, including maximizing social welfare and maximizing the seller's profit in auctions. In addition, considering the randomness between the model test performance used in the auction and the actual performance in a production environment, we provide theoretical analyses to quantify the impact of the uncertainty on the social welfare or the seller's profit of auction mechanisms. We provide experimental results based on two datasets with synthetic buyers' valuation to illustrate the truthfulness, social welfare, and data sellers' profit.

Local differential privacy ( LDP ) is a strong privacy standard that has been adopted by popular software systems, including Chrome, iOS, MacOS, and Windows. The main idea is that each individual perturbs their own data locally, and only submits the resulting noisy version to a data aggregator. Although much effort has been devoted to computing various types of aggregates and building machine learning applications under LDP, research on fundamental perturbation mechanisms has not achieved significant improvement in recent years. Towards a more refined result utility, existing works in the literature mainly focus on improving the worst-case guarantee. However, this approach does not necessarily promise a better average performance given the fact that the data in practice obey a certain distribution, which is not known beforehand. In this paper, we propose the advanced adaptive additive ( AAA ) mechanism, which is a distribution-aware approach that addresses the average utility and tackles the classical mean estimation problem. AAA is carried out in a two-step approach: first, as the global data distribution is not available beforehand, the data aggregator selects a random subset of individuals to compute a (noisy) quantized data descriptor; then, in the second step, the data aggregator collects data from the remaining individuals, which are perturbed in a distribution-aware fashion. The perturbation involved in the latter step is obtained by solving an optimization problem, which is formulated with the data descriptor obtained in the former step and the desired properties of task-determined utilities. We provide rigorous privacy proofs and utility analyses, as well as extensive experiments comparing AAA with state-of-the-art mechanisms. The evaluation results demonstrate that the AAA mechanism consistently outperforms existing solutions with a clear margin in terms of result utility, on a wide range of privacy constraints and real-world and synthetic datasets.

Efficient query optimization is crucial for database management systems. Recently, machine learning models have been applied in query optimizers to generate better plans, but the unpredictable performance regressions prevent them from being truly applicable. To be more specific, while a learned query optimizer commonly outperforms the traditional query optimizer on average for a workload of queries, its performance regression seems inevitable for some queries due to model under-fitting and difficulty in generalization. In this paper, we propose a system called Eraser to resolve this problem. Eraser aims at eliminating performance regressions while still attaining considerable overall performance improvement. To this end, Eraser applies a two-stage strategy to estimate the model accuracy for each candidate plan, and helps the learned query optimizer select more reliable plans. The first stage serves as a coarse-grained filter that removes all highly risky plans with feature values that are seen for the first time. The second stage clusters plans in a more fine-grained manner and evaluates each cluster according to the prediction quality of learned query optimizers for selecting the final execution plan. Eraser can be deployed as a plugin on top of any learned query optimizer. We implement Eraser and demonstrate its superiority on PostgreSQL and Spark. In our experiments, Eraser eliminates most of the regressions while bringing very little negative impact on the overall performance of learned query optimizers, no matter whether they perform better or worse than the traditional query optimizer. Meanwhile, it is adaptive to dynamic settings and generally applicable to different database systems.

Cardinality estimation (CardEst) is essential for optimizing query execution plans. Recent ML-based CardEst methods achieve high accuracy but face deployment challenges due to high preparation costs and lack of transferability across databases. In this paper, we propose PRICE, a PRetrained multI-table CardEst model, which addresses these limitations. PRICE takes low-level but transferable features w.r.t. data distributions and query information and elegantly applies self-attention models to learn meta-knowledge to compute cardinality in any database. It is generally and adaptively applicable to any unseen new database to attain high estimation accuracy, while its preparation cost is as little as the basic one-dimensional histogram-based CardEst methods. Moreover, PRICE can be finetuned to further enhance its performance on any specific database. We pretrained PRICE using 30 diverse datasets, completing the process in about 5 hours with a resulting model size of only about 40MB. Evaluations show that PRICE consistently outperforms existing methods, achieving the highest estimation accuracy on several unseen databases and generating faster execution plans with lower overhead. After finetuning with a small volume of database-specific queries, PRICE could even find plans that were very close to the optimal ones. Meanwhile, PRICE is generally applicable to different settings such as data updates, data scaling, and query workload shifts.

Learned databases, or AI4DB techniques, have rapidly developed in the last decade. Deploying machine learning (ML) and AI4DB algorithms into actual databases is the gold standard to examine their performance in practice. However, due to the complexity of database systems, the difference between ML and DB programming paradigms, and the diversity of ML models, the tasks of developing and deploying AI4DB algorithms into databases are prohibitively difficult. Most previous works focus on specific AI4DB algorithms and ML models whose deployment requires close cooperation between ML and DB developers and heavy engineering cost. In this paper, we design and implement PilotScope, an AI4DB middleware with a programming model that largely reduces such difficulties. With a novel abstraction of AI4DB algorithms for, e.g. , knob tuning and query optimization, PilotScope consists of two classes of components, AI4DB drivers and DB interactors , with different programming paradigms and roles in AI4DB tasks. ML developers focus on designing and implementing AI4DB drivers, which are algorithmic workflows that collect statistics from databases, train ML models, make decisions and optimize databases using learned models. AI4DB drivers interact with databases via DB interactors ( e.g. , for collecting data and enforcing actions in databases). DB developers focus on implementing these interactors on one or more database engines, with the interaction details hindered from ML developers. PilotScope supports a variety of AI4DB tasks, and the implementation of an AI4DB algorithm on PilotScope can be deployed in different databases with only minimum modifications. PilotScope is effective in benchmarking these AI4DB algorithms in real-world scenarios. We hope that PilotScope could significantly accelerate iterating AI4DB research and make AI4DB techniques truly applicable in production.

The successes of applying Transformer to graphs have been witnessed on small graphs (e.g., molecular graphs), yet two barriers prevent its adoption on large graphs (e.g., citation networks). First, despite the benefit of the global receptive field, enormous distant nodes might distract the necessary attention of each target node from its neighborhood. Second, training a Transformer model on large graphs is costly due to the node-to-node attention mechanism's quadratic computational complexity. To break down these barriers, we propose a two-view architectureCoarformer, wherein a GNN-based module captures fine-grained local information from the original graph, and a Transformer-based module captures coarse yet long-range information on the coarse graph. We further design a cross-view propagation scheme so that these two views can enhance each other. Our graph isomorphism analysis shows the complementary natures of GNN and Transformer, justifying the motivation and design ofCoarformer. We conduct extensive experiments on real-world datasets, whereCoarformersurpasses any single-view method that solely applies a GNN or Transformer. As an ablation,Coarformeroutperforms straightforward combinations of a GNN model and a Transformer-based model, verifying the effectiveness of our coarse global view and the cross-view propagation scheme. Meanwhile,Coarformerconsumes the least runtime and GPU memory than those combinations.

Nowadays, as privacy concerns continue to rise, federated graph learning (FGL) which generalizes the classic federated learning to graph data has attracted increasing attention. However, while the focus has been on designing collaborative learning algorithms, the potential risks of privacy leakage through the sharing of necessary graph-related information in FGL, such as node embeddings and neighbor generators, have been largely neglected. In this paper, we verify the potential risks of privacy leakage in FGL, and provide insights about the cautions in FGL algorithm design. Specifically, we propose a novel privacy attack algorithm named Privacy Attack on federated Graph learning (PAG) towards reconstructing participants’ private node attributes and the linkage relationships. The participant performing the PAG attack is able to reconstruct the node attributes of the victim by matching the received gradients of the generator, and then train a link prediction model based on its local sub-graph to inductively infer the linkages connected to these reconstructed nodes. We theoretically and empirically demonstrate that under PAG attack, directly sharing the neighbor generators makes the FGL vulnerable to the data reconstruction attack. Furthermore, an investigation into the key factors that can hinder the success of the PAG attack provides insights into corresponding defense strategies and inspires future research into privacy-preserving FGL.

Federated Learning (FL) aims to train machine learning models for multiple clients without sharing their own private data. Due to the heterogeneity of clients’ local data distribution, recent studies explore the personalized FL that learns and deploys distinct local models with the help of auxiliary global models. However, the clients can be heterogeneous in terms of not only local data distribution, but also their computation and communication resources. The capacity and efficiency of personalized models are restricted by the lowest-resource clients, leading to sub-optimal performance and limited practicality of personalized FL. To overcome these challenges, we propose a novel approach named pFedGate for efficient personalized FL by adaptively and efficiently learning sparse local models. With a lightweight trainable gating layer, pFedGate enables clients to reach their full potential in model capacity by generating different sparse models accounting for both the heterogeneous data distributions and resource constraints. Meanwhile, the computation and communication efficiency are both improved thanks to the adaptability between the model sparsity and clients’ resources. Further, we theoretically show that the proposed pFedGate has superior complexity with guaranteed convergence and generalization error. Extensive experiments show that pFedGate achieves superior global accuracy, individual accuracy and efficiency simultaneously over state-of-the-art methods. We also demonstrate that pFedGate performs better than competitors in the novel clients participation and partial clients participation scenarios, and can learn meaningful sparse local models adapted to different data distributions.

Research in the field of hyperparameter optimization (HPO) has been greatly accelerated by existing HPO benchmarks. Nonetheless, existing efforts in benchmarking all focus on HPO for traditional learning paradigms while ignoring federated learning (FL), a promising paradigm for collaboratively learning models from dispersed data. In this paper, we first identify some uniqueness of federated hyperparameter optimization (FedHPO) from various aspects, showing that existing HPO benchmarks no longer satisfy the need to study FedHPO methods. To facilitate the research of FedHPO, we propose and implement a benchmark suite FedHPO-Bench that incorporates comprehensive FedHPO problems, enables flexible customization of the function evaluations, and eases continuing extensions. We conduct extensive experiments based on FedHPO-Bench to provide the community with more insights into FedHPO. We open-sourced FedHPO-Bench at https://github.com/alibaba/FederatedScope/tree/master/benchmark/FedHPOBench.

Federated Learning (FL) aims to train high-quality models in collaboration with distributed clients while not uploading their local data, which attracts increasing attention in both academia and industry. However, there is still a considerable gap between the flourishing FL research and real-world scenarios, mainly caused by the characteristics of heterogeneous devices and its scales. Most existing works conduct evaluations with homogeneous devices, which are mismatched with the diversity and variability of heterogeneous devices in real-world scenarios. Moreover, it is challenging to conduct research and development at scale with heterogeneous devices due to limited resources and complex software stacks. These two key factors are important yet underexplored in FL research as they directly impact the FL training dynamics and final performance, making the effectiveness and usability of FL algorithms unclear. To bridge the gap, in this paper, we propose an efficient and scalable prototyping system for real-world cross-device FL, FS-REAL. It supports heterogeneous device runtime, contains parallelism and robustness enhanced FL server, and provides implementations and extensibility for advanced FL utility features such as personalization, communication compression and asynchronous aggregation. To demonstrate the usability and efficiency of FS-REAL, we conduct extensive experiments with various device distributions, quantify and analyze the effect of the heterogeneous device and various scales, and further provide insights and open discussions about real-world FL scenarios. Our system is released to help to pave the way for further real-world FL research and broad applications involving diverse devices and scales.

In this work, besides improving prediction accuracy, we study whether personalization could bring robustness benefits to backdoor attacks. We conduct the first study of backdoor attacks in the pFL framework, testing 4 widely used backdoor attacks against 6 pFL methods on benchmark datasets FEMNIST and CIFAR-10, a total of 600 experiments. The study shows that pFL methods with partial model-sharing can significantly boost robustness against backdoor attacks. In contrast, pFL methods with full model-sharing do not show robustness. To analyze the reasons for varying robustness performances, we provide comprehensive ablation studies on different pFL methods. Based on our findings, we further propose a lightweight defense method, Simple-Tuning, which empirically improves defense performance against backdoor attacks. We believe that our work could provide both guidance for pFL application in terms of its robustness and offer valuable insights to design more robust FL methods in the future. We open-source our code to establish the first benchmark for black-box backdoor attacks in pFL: https://github.com/alibaba/FederatedScope/tree/backdoor-bench.

We study a natural competitive-information-design variant for the Pandora's Box problem [31], where each box is associated with a strategic information sender who can design what information about the box's prize value to be revealed to the agent when she inspects the box. This variant with strategic boxes is motivated by a wide range of real-world economic applications for Pandora's box. The main contributions of this article are two-fold: (1) we study informational properties of Pandora's Box by analyzing how a box's partial information revelation affects the search agent's optimal decisions; and (2) we fully characterize the pure symmetric equilibrium for the boxes' competitive information revelation, which reveals various insights regarding information competition and the resultant agent utility at equilibrium.

With ubiquitous adoption of machine learning algorithms in web technologies, such as recommendation system and social network, algorithm fairness has become a trending topic, and it has a great impact on social welfare. Among different fairness definitions, path-specific causal fairness is a widely adopted one with great potentials, as it distinguishes the fair and unfair effects that the sensitive attributes exert on algorithm predictions. Existing methods based on path-specific causal fairness either require graph structure as the prior knowledge or have high complexity in the calculation of path-specific effect. To tackle these challenges, we propose a novel casual graph based fair prediction framework which integrates graph structure learning into fair prediction to ensure that unfair pathways are excluded in the causal graph. Furthermore, we generalize the proposed framework to the scenarios where sensitive attributes can be non-root nodes and affected by other variables, which is commonly observed in real-world applications, such as recommendation system, but hardly addressed by existing works. We provide theoretical analysis on the generalization bound for the proposed fair prediction method, and conduct a series of experiments on real-world datasets to demonstrate that the proposed framework can provide better prediction performance and algorithm fairness trade-off.

Some recent works have shown the advantages of reinforcement learning (RL) based learned query optimizers. These works often use the cost (i.e., the estimation of cost model) or the latency (i.e., execution time) as guidance signals for training their learned models. However, cost-based learning underperforms in latency and latency-based learning is time-intensive. In order to bypass such a dilemma, researchers attempt to transfer a learned value network from the cost domain to the latency domain. We recognize critical insights in cost/latency-based training, prompting us to transfer the reward function rather than the value network. Based on this idea, we propose a two-stage RL-based framework, BASE , to bridge the gap between cost and latency. After learning a policy based on cost signals in its first stage, BASE formulates transferring the reward function as a variant of inverse reinforcement learning. Intuitively, BASE learns to calibrate the reward function and updates the policy regarding the calibrated one in a mutually-improved manner. Extensive experiments exhibit the superiority of BASE on two benchmark datasets: Our optimizer outperforms traditional DBMS, using 30% less training time than SOTA methods. Meanwhile, our approach can enhance the efficiency of other learning-based optimizers.

Federated learning (FL) is a general distributed machine learning paradigm that provides solutions for tasks where data cannot be shared directly. Due to the difficulties in communication management and heterogeneity of distributed data and devices, initiating and using an FL algorithm for real-world cross-device scenarios requires significant repetitive effort but may not be transferable to similar projects. To reduce the effort required for developing and deploying FL algorithms, we present FS-Real, an open-source FL platform designed to address the need of a general and efficient infrastructure for real-world cross-device FL. In this paper, we introduce the key components of FS-Real and demonstrate that FS-Real has the following capabilities: 1) reducing the programming burden of FL algorithm development with plug-and-play and adaptable runtimes on Android and other Internet of Things (IoT) devices; 2) handling a large number of heterogeneous devices efficiently and robustly with our communication management components; 3) supporting a wide range of advanced FL algorithms with flexible configuration and extension; 4) alleviating the costs and efforts for deployment, evaluation, simulation, and performance optimization of FL algorithms with automatized tool kits.

Targeting in-memory one-dimensional search keys, we propose a novel DIstribution-driven Learned Index tree ( DILI ), where a concise and computation-efficient linear regression model is used for each node. An internal node's key range is equally divided by its child nodes such that a key search enjoys perfect model prediction accuracy to find the relevant leaf node. A leaf node uses machine learning models to generate searchable data layout and thus accurately predicts the data record position for a key. To construct DILI, we first build a bottom-up tree with linear regression models according to global and local key distributions. Using the bottom-up tree, we build DILI in a top-down manner, individualizing the fanouts for internal nodes according to local distributions. DILI strikes a good balance between the number of leaf nodes and the height of the tree, two critical factors of key search time. Moreover, we design flexible algorithms for DILI to efficiently insert and delete keys and automatically adjust the tree structure when necessary. Extensive experimental results show that DILI outperforms the state-of-the-art alternatives on different kinds of workloads.

For efficient query processing, DBMS query optimizers have for decades relied on delicate cardinality estimation methods. In this work, we propose an Attention-based LEarned Cardinality Estimator ( ALECE for short) for SPJ queries. The core idea is to discover the implicit relationships between queries and underlying dynamic data using attention mechanisms in ALECE's two modules that are built on top of carefully designed featurizations for data and queries. In particular, from all attributes in the database, the data-encoder module obtains organic and learnable aggregations which implicitly represent correlations among the attributes, whereas the query-analyzer module builds a bridge between the query featurizations and the data aggregations to predict the query's cardinality. We experimentally evaluate ALECE on multiple dynamic workloads. The results show that ALECE enables PostgreSQL's optimizer to achieve nearly optimal performance, clearly outperforming its built-in cardinality estimator and other alternatives.

Although remarkable progress has been made by existing federated learning (FL) platforms to provide infrastructures for development, these platforms may not well tackle the challenges brought by various types of heterogeneity. To fill this gap, in this paper, we propose a novel FL platform, named FederatedScope, which employs an event-driven architecture to provide users with great flexibility to independently describe the behaviors of different participants. Such a design makes it easy for users to describe participants with various local training processes, learning goals and backends, and coordinate them into an FL course with synchronous or asynchronous training strategies. Towards an easy-to-use and flexible platform, FederatedScope enables rich types of plug-in operations and components for efficient further development, and we have implemented several important components to better help users with privacy protection, attack simulation and auto-tuning. We have released FederatedScope at https://github.com/alibaba/FederatedScope to promote academic research and industrial deployment of federated learning in a wide range of scenarios.

A recent line of works apply machine learning techniques to assist or rebuild cost-based query optimizers in DBMS. While exhibiting superiority in some benchmarks, their deficiencies, e.g., unstable performance, high training cost, and slow model updating, stem from the inherent hardness of predicting the cost or latency of execution plans using machine learning models. In this paper, we introduce a learning-to-rank query optimizer, called Lero, which builds on top of a native query optimizer and continuously learns to improve the optimization performance. The key observation is that the relative order or rank of plans, rather than the exact cost or latency, is sufficient for query optimization. Lero employs a pairwise approach to train a classifier to compare any two plans and tell which one is better. Such a binary classification task is much easier than the regression task to predict the cost or latency, in terms of model efficiency and accuracy. Rather than building a learned optimizer from scratch, Lero is designed to leverage decades of wisdom of databases and improve the native query optimizer. With its non-intrusive design, Lero can be implemented on top of any existing DBMS with minimal integration efforts. We implement Lero and demonstrate its outstanding performance using PostgreSQL. In our experiments, Lero achieves near optimal performance on several benchmarks. It reduces the plan execution time of the native optimizer in PostgreSQL by up to 70% and other learned query optimizers by up to 37%. Meanwhile, Lero continuously learns and automatically adapts to query workloads and changes in data.

Conversational recommender systems (CRS) endow traditional recommender systems with the capability of dynamically obtaining users’ short-term preferences for items and attributes through interactive dialogues. There are three core challenges for CRS, including the intelligent decisions for what attributes to ask, which items to recommend, and when to ask or recommend, at each conversation turn. Previous methods mainly leverage reinforcement learning (RL) to learn conversational recommendation policies for solving one or two of these three decision-making problems in CRS with separated conversation and recommendation components. These approaches restrict the scalability and generality of CRS and fall short of preserving a stable training procedure. In the light of these challenges, we tackle these three decision-making problems in CRS as a unified policy learning task. In order to leverage different features that are important to each sub-problem and facilitate better unified policy learning in CRS, we propose two novel multi-agent RL-based frameworks, namely Independent and Hierarchical Multi-Agent UNIfied COnversational RecommeNders (IMA-UNICORN and HMA-UNICORN), respectively. In specific, two low-level agents enrich the state representations for attribute prediction and item recommendation, by combining the long-term user preference information from the historical interaction data and the short-term user preference information from the conversation history. A high-level meta agent is responsible for coordinating the low-level agents to adaptively make the final decision. Experimental results on four benchmark CRS datasets and a real-world E-Commerce application show that the proposed frameworks significantly outperform state-of-the-art methods. Extensive analyses further demonstrate the superior scalability of the MARL frameworks on the multi-round conversational recommendation.

Range aggregation is a primitive operation in spatial data applications and there is a growing demand to support such operations over a data federation, where the entire spatial data are separately held by multiple data providers (a.k.a., data silos). Data federations notably increase the amount of data available for data-intensive applications such as smart mobility planning and public health emergency responses. Yet they also challenge the conventional implementation of range aggregation queries because the raw data cannot be shared within the federation and the data partition at each data silo is fixed during query processing. These constraints limit the design space of distributed range aggregation query processing. In this work, we propose approximate algorithms for efficient range aggregation over spatial data federation. We devise novel single-silo sampling algorithms that process queries in parallel and design a level sampling based algorithm which reduces the time complexity of local queries at each data silo to O(log 1/), where is the approximation ratio of the accuracy guarantee. Extensive evaluations with real-world data show that compared with state-of-the-arts, our solutions reduce the time cost and communication cost by up to 85.1x and 5.5x respectively, with average approximate errors of below 2.8%.

Due to the openness of the online platform, recommendation systems are vulnerable to data poisoning attacks, where malicious samples are injected into the training set of the recommendation system to manipulate its recommendation results. Existing attack approaches are either based on heuristic rules or designed against specific recommendation approaches. The former suffers unsatisfactory performance, while the latter requires strong knowledge of the target system. In this paper, we propose a practical poisoning attack approach named LOKI against blackbox recommendation systems. The proposed LOKI utilizes the reinforcement learning algorithm to train the attack agent, which can be used to generate user behavior samples for data poisoning. In real-world recommendation systems, the cost of retraining recommendation models is high, and the interaction frequency between users and a recommendation system is restricted. Thus, we propose to let the agent interact with a recommender simulator instead of the target recommendation system and leverage the transferability of the generated adversarial samples to poison the target system. We also use the influence function to efficiently estimate the influence of injected samples on recommendation results, without re-training the models. Extensive experiments on multiple datasets against four representative recommendation models show that the proposed LOKI outperformances existing method. We also discuss the characteristics of vulnerable users/items, and evaluate whether anomaly detection methods can be used to mitigate the impact of data poisoning attacks.

Recently, privacy issues in web services that rely on users’ personal data have raised great attention. Despite that recent regulations force companies to offer choices for each user to opt-in or opt-out of data disclosure, real-world applications usually only provide an “all or nothing” binary option for users to either disclose all their data or preserve all data with the cost of no personalized service. In this article, we argue that such a binary mechanism is not optimal for both consumers and platforms. To study how different privacy mechanisms affect users’ decisions on information disclosure and how users’ decisions affect the platform’s revenue, we propose a privacy-aware recommendation framework that gives users fine control over their data. In this new framework, users can proactively control which data to disclose based on the tradeoff between anticipated privacy risks and potential utilities. Then we study the impact of different data disclosure mechanisms via simulation with reinforcement learning due to the high cost of real-world experiments. The results show that the platform mechanisms with finer split granularity and more unrestrained disclosure strategy can bring better results for both consumers and platforms than the “all or nothing” mechanism adopted by most real-world applications.

Ciphertexts of an order-preserving encryption (OPE) scheme preserve the order of their corresponding plaintexts. However, OPEs are vulnerable to inference attacks that exploit this preserved order. Differential privacy (DP) has become the de-facto standard for data privacy. One of the most attractive properties of DP is that any post-processing computation, such as inference attacks, performed on the noisy output of a DP algorithm does not degrade its privacy guarantee. In this work, we propose a novel differentially private order preserving encryption scheme, OP ε. Under OP ε, the leakage of order from the ciphertexts is differentially private. Consequently, in the least, OP ε ensures a formal guarantee (a relaxed DP guarantee) even in the face of inference attacks. To the best of our knowledge, this is the first work to combine DP with a OPE. OP ε is based on a novel differentially private order preserving encoding scheme, OPεc, that can be of independent interest in the local DP setting. We demonstrate OP ε's utility in answering range queries via empirical evaluation on four real-world datasets. For instance, OP ε misses only around 4 in every 10K correct records on average for a dataset of size ~732K with an attribute of domain size ~18K and ε= 1.

Masked autoencoder (MAE), as an effective self-supervised learner for computer vision and natural language processing, has been recently applied to molecule representation learning. In this paper, we identify two issues in applying MAE to pre-train Transformer-based models on molecular graphs that existing works have ignored. (1) As only atoms are abstracted as tokens and then reconstructed, the chemical bonds are not decided in the decoded molecule, making molecules with different arrangements of the same atoms indistinguishable. (2) Although a high mask ratio that corresponds to a challenging reconstruction task has been proved beneficial in the vision domain, it cannot be trivially leveraged on molecular graphs as there is less redundancy of information in graph data. To resolve these issues, we propose a novel framework, Molecular Graph Mask AutoEncoder (MGMAE). As the first step in MGMAE, we transform each molecular graph into a heterogeneous atom-bond graph to fully use the bond attributes and design unidirectional position encoding for such graphs. Then we propose a hybrid masking mechanism that exploits the complementary nature between atoms' attributive and spatial features. Meanwhile, we compensate for the mask embedding by a dynamic aggregation representation that exploits the correlations between topologically adjacent tokens. As a result, MGMAE can reconstruct the masked atoms, the masked bonds, and the relative distance among atoms simultaneously, with a high mask ratio. We compare MGMAE with the state-of-the-art methods on various molecular benchmarks and show the competitiveness of MGMAE in both regression and classification tasks.

Data federations notably increase the amount of data available for data-intensive applications such as smart mobility planning and public health emergency responses. Yet they also challenge the conventional implementation of range aggregation queries because the raw data cannot be shared within the federation and the data partition at each data silo is fixed during query processing. In this work, we propose the first-of-its-kind approximate algorithms for efficient range aggregation over spatial data federation. We devise novel single-silo sampling algorithms that process queries in parallel and design a level sampling based algorithm which reduces the time complexity of local queries at each data silo to$O(\log\frac{1}{\epsilon})$, where ∊ is the approximation ratio of the accuracy guarantee. Extensive experiments on real-world dataset validate the efficiency and effectiveness of the solutions.

Sequential recommendation methods play a crucial role in modern recommender systems because of their ability to capture a user's dynamic interest from her/his historical inter-actions. Despite their success, we argue that these approaches usually rely on the sequential prediction task to optimize the huge amounts of parameters. They usually suffer from the data sparsity problem, which makes it difficult for them to learn high-quality user representations. To tackle that, inspired by recent advances of contrastive learning techniques in the computer vision, we propose a novel multi-task framework called Contrastive Learning for Sequential Recommendation (CL4SRec). CL4SRec not only takes advantage of the traditional next item prediction task but also utilizes the contrastive learning framework to derive self-supervision signals from the original user behavior sequences. Therefore, it can extract more meaningful user patterns and further encode the user representations effectively. In addition, we propose three data augmentation approaches to construct self-supervision signals. Extensive experiments on four public datasets demonstrate that CL4SRec achieves state-of-the-art performance over existing baselines by inferring better user representations.

The lottery ticket hypothesis (LTH) states that a randomly initialized dense network contains sub-networks that can be trained in isolation to the performance of the dense network. In this paper, to achieve rapid learning with less computational cost, we explore LTH in the context of meta learning. First, we experimentally show that there are sparse sub-networks, known as meta winning tickets, which can be meta-trained to few-shot classification accuracy to the original backbone. The application of LTH in meta learning enables the adaptation of meta-trained networks on various IoT devices with fewer computation. However, the status quo to identify winning tickets requires iterative training and pruning, which is particularly expensive for finding meta winning tickets. To this end, then we investigate the inter- and intra-layer patterns among different meta winning tickets, and propose a scheme for early detection of a meta winning ticket. The proposed scheme enables efficient training in resource-limited devices. Besides, it also designs a lightweight solution to search the meta winning ticket. Evaluations on standard few-shot classification benchmarks show that we can find competitive meta winning tickets with 20% weights of the original backbone, while incurring only 8%-14% (Conv-4) and 19%-29% (ResNet-12) computation overhead (measured by FLOPs) of the standard winning ticket finding scheme.

In order to develop effective sequential recommenders, a series of sequence representation learning (SRL) methods are proposed to model historical user behaviors. Most existing SRL methods rely on explicit item IDs for developing the sequence models to better capture user preference. Though effective to some extent, these methods are difficult to be transferred to new recommendation scenarios, due to the limitation by explicitly modeling item IDs. To tackle this issue, we present a novel universal sequence representation learning approach, named UniSRec. The proposed approach utilizes the associated description text of items to learn transferable representations across different recommendation scenarios. For learning universal item representations, we design a lightweight item encoding architecture based on parametric whitening and mixture-of-experts enhanced adaptor. For learning universal sequence representations, we introduce two contrastive pre-training tasks by sampling multi-domain negatives. With the pre-trained universal sequence representation model, our approach can be effectively transferred to new recommendation domains or platforms in a parameter-efficient way, under either inductive or transductive settings. Extensive experiments conducted on real-world datasets demonstrate the effectiveness of the proposed approach. Especially, our approach also leads to a performance improvement in a cross-platform setting, showing the strong transferability of the proposed universal SRL method. The code and pre-trained model are available at: https://github.com/RUCAIBox/UniSRec.

As Internet users attach importance to their own privacy, and a number of laws and regulations go into effect in most countries, Internet products need to provide users with privacy protection. As one of the feasible solutions to provide such privacy protection, federated learning has rapidly gained popularity in both academia and industry in recent years. In this tutorial, we will start off with some real-world tasks to illustrate the topic of federated learning, and cover some basic concepts and important scenarios including cross-device and cross-silo settings. Along with it, we will give several demonstrations with popular federated learning frameworks. We will also show how to do the automatic hyperparameter tuning with federated learning to significantly save their efforts in practice. Then we dive into three parallel hot topics, Personalized Federated Learning, Federated Graph Learning, and Attack in Federated Learning. For each of them, we will motivate it with real-world applications, illustrate the state-of-the-art methods, and discuss their pros and cons using concrete examples. As the last part, we will point out some future research directions.

In data mining, estimating the number of distinct values (NDV) is a fundamental problem with various applications. Existing methods for estimating NDV can be broadly classified into two categories: i) scanning-based methods, which scan the entire data and maintain a sketch to approximate NDV; and ii) sampling-based methods, which estimate NDV using sampling data rather than accessing the entire data warehouse. Scanning-based methods achieve a lower approximation error at the cost of higher I/O and more time. Sampling-based estimation is preferable in applications with a large data volume and a permissible error restriction due to its higher scalability. However, while the sampling-based method is more effective on a single machine, it is less practical in a distributed environment with massive data volumes. For obtaining the final NDV estimators, the entire sample must be transferred throughout the distributed system, incurring a prohibitive communication cost when the sample rate is significant. This paper proposes a novel sketch-based distributed method that achieves sub-linear communication costs for distributed sampling-based NDV estimation under mild assumptions. Our method leverages a sketch-based algorithm to estimate the sample's frequency of frequency in the distributed streaming model, which is compatible with most classical sampling-based NDV estimators. Additionally, we provide theoretical evidence for our method's ability to minimize communication costs in the worst-case scenario. Extensive experiments show that our method saves orders of magnitude in communication costs compared to existing sampling- and sketch-based methods.

The incredible development of federated learning (FL) has benefited various tasks in the domains of computer vision and natural language processing, and the existing frameworks such as TFF and FATE has made the deployment easy in real-world applications. However, federated graph learning (FGL), even though graph data are prevalent, has not been well supported due to its unique characteristics and requirements. The lack of FGL-related framework increases the efforts for accomplishing reproducible research and deploying in real-world applications. Motivated by such strong demand, in this paper, we first discuss the challenges in creating an easy-to-use FGL package and accordingly present our implemented package FederatedScope-GNN (FS-G), which provides (1) a unified view for modularizing and expressing FGL algorithms; (2) comprehensive DataZoo and ModelZoo for out-of-the-box FGL capability; (3) an efficient model auto-tuning component; and (4) off-the-shelf privacy attack and defense abilities. We validate the effectiveness of FS-G by conducting extensive experiments, which simultaneously gains many valuable insights about FGL for the community. Moreover, we employ FS-G to serve the FGL application in real-world E-commerce scenarios, where the attained improvements indicate great potential business benefits. We publicly release FS-G, as submodules of FederatedScope, at https://github.com/alibaba/FederatedScope to promote FGL's research and enable broad applications that would otherwise be infeasible due to the lack of a dedicated package.

Recently, Neural Architecture Search (NAS) for GNN has received increasing popularity as it can seek an optimal architecture for a given new graph. However, the optimal architecture is applied to all the instances (i.e., nodes, in the context of graph) equally, which might be insufficient to handle the diverse local patterns ingrained in a graph, as shown in this paper and some very recent studies. Thus, we argue the necessity of node-wise architecture search for GNN. Nevertheless, node-wise architecture cannot be realized by trivially applying NAS methods node by node due to the scalability issue and the need for determining test nodes' architectures. To tackle these challenges, we propose a framework wherein the parametric controllers decide the GNN architecture for each node based on its local patterns. We instantiate our framework with depth, aggregator and resolution controllers, and then elaborate on learning the backbone GNN model and the controllers to encourage their cooperation. Empirically, we justify the effects of node-wise architecture through the performance improvements introduced by the three controllers, respectively. Moreover, our proposed framework significantly outperforms state-of-the-art methods on five of the ten real-world datasets, where the diversity of these datasets has hindered any graph convolution-based method to lead on them simultaneously. This result further confirms that node-wise architecture can help GNNs become versatile models.

Personalized Federated Learning (pFL), which utilizes and deploys distinct local models, has gained increasing attention in recent years due to its success in handling the statistical heterogeneity of FL clients. However, standardized evaluation and systematical analysis of diverse pFL methods remain a challenge. Firstly, the highly varied datasets, FL simulation settings and pFL implementations prevent easy and fair comparisons of pFL methods. Secondly, the current pFL literature diverges in the adopted evaluation and ablation protocols. Finally, the effectiveness and robustness of pFL methods are under-explored in various practical scenarios, such as the generalization to new clients and the participation of resource-limited clients. To tackle these challenges, we propose the first comprehensive pFL benchmark, pFL-Bench, for facilitating rapid, reproducible, standardized and thorough pFL evaluation. The proposed benchmark contains more than 10 dataset variants in various application domains with a unified data partition and realistic heterogeneous settings; a modularized and easy-to-extend pFL codebase with more than 20 competitive pFL method implementations; and systematic evaluations under containerized environments in terms of generalization, fairness, system overhead, and convergence. We highlight the benefits and potential of state-of-the-art pFL methods and hope pFL-Bench enables further pFL research and broad applications that would otherwise be difficult owing to the absence of a dedicated benchmark. The code is released at https://github.com/alibaba/FederatedScope/tree/master/benchmark/pFL-Bench.

Vertical Federated Learning (VFL), that trains federated models over vertically partitioned data, has emerged as an important learning paradigm. However, existing VFL methods are facing two challenges: (1) scalability when # participants grows to even modest scale and (2) diminishing return w.r.t. # participants: not all participants are equally important and many will not introduce quality improvement in a large consortium. Inspired by these two challenges, in this paper, we ask: How can we select l out of m participants, where l ≪ m, that are most important?We call this problem Vertically Federated Participant Selection, and model it with a principled mutual information-based view. Our first technical contribution is VF-MINE—a Vertically Federated Mutual INformation Estimator—that uses one of the most celebrated algorithms in database theory—Fagin’s algorithm as a building block. Our second contribution is to further optimize VF-MINE to enable VF-PS, a group testing-based participant selection framework. We empirically show that vertically federated participation selection can be orders of magnitude faster than training a full-fledged VFL model, while being able to identify the most important subset of participants that often lead to a VFL model of similar quality.

Graph Neural Networks (GNNs) have received extensive research attention for their promising performance in graph machine learning. Despite their extraordinary predictive accuracy, existing approaches, such as GCN and GPRGNN, are not robust in the face of homophily changes on test graphs, rendering these models vulnerable to graph structural attacks and with limited capacity in generalizing to graphs of varied homophily levels. Although many methods have been proposed to improve the robustness of GNN models, most of these techniques are restricted to the spatial domain and employ complicated defense mechanisms, such as learning new graph structures or calculating edge attentions. In this paper, we study the problem of designing simple and robust GNN models in the spectral domain. We propose EvenNet, a spectral GNN corresponding to an even-polynomial graph filter. Based on our theoretical analysis in both spatial and spectral domains, we demonstrate that EvenNet outperforms full-order models in generalizing across homophilic and heterophilic graphs, implying that ignoring odd-hop neighbors improves the robustness of GNNs. We conduct experiments on both synthetic and real-world datasets to demonstrate the effectiveness of EvenNet. Notably, EvenNet outperforms existing defense models against structural attacks without introducing additional computational costs and maintains competitiveness in traditional node classification tasks on homophilic and heterophilic graphs.

Limited by the statistical-based machine learning framework, a spurious correlation is likely to appear in existing knowledge-aware recommendation methods. It refers to a knowledge fact that appears causal to the user behaviors (inferred by the recommender) but is not in fact. For tackling this issue, we present a novel approach to discovering and alleviating the potential spurious correlations from a counterfactual perspective. To be specific, our approach consists of two counterfactual generators and a recommender. The counterfactual generators are designed to generate counterfactual interactions via reinforcement learning, while the recommender is implemented with two different graph neural networks to aggregate the information from KG and user-item interactions respectively. The counterfactual generators and recommender are integrated in a mutually collaborative way. With this approach, the recommender helps the counterfactual generators better identify potential spurious correlations and generate high-quality counterfactual interactions, while the counterfactual generators help the recommender weaken the influence of the potential spurious correlations simultaneously. Extensive experiments on three real-world datasets have shown the effectiveness of the proposed approach by comparing it with a number of competitive baselines. Our implementation code is available at: https://github.com/RUCAIBox/CGKR.

Sample-based estimation, which uses a sample to estimate population parameters (e.g., SUM, COUNT, and AVG), has various applications in database systems. A sampler defines how samples are drawn from a population. Various samplers have been proposed (e.g., uniform sampler, stratified sampler, and measure-biased sampler), since there is no single sampler that works well in all cases. To overcome the "one size does not fit all" challenge, we study how to combine multiple samplers to estimate population parameters, and propose SamComb, a novel bandit-based sampler combination framework. Given a set of samplers, a budget, and a population parameter, SamComb can automatically decide how much budget should be allocated to each sampler so that the combined estimation achieves the highest accuracy. We model this sampler combination problem as a multi-armed bandit (MAB) problem and propose effective approaches to balance the exploration and exploitation trade-off in a principled way. We provide theoretical guarantees for our approaches and conduct extensive experiments on both synthetic and real datasets. The results show that there is a strong need to combine multiple samplers, in order to obtain accurate estimations without the knowledge about population predicates and distributions, and SamComb is an effective framework to achieve this goal.

Recommender systems provide essential web services by learning users’ personal preferences from collected data. However, in many cases, systems also need to forget some training data. From the perspective of privacy, users desire a tool to erase the impacts of their sensitive data from the trained models. From the perspective of utility, if a system’s utility is damaged by some bad data, the system needs to forget such data to regain utility. While unlearning is very important, it has not been well-considered in existing recommender systems. Although there are some researches have studied the problem of machine unlearning, existing methods can not be directly applied to recommendation as they are unable to consider the collaborative information.

Although neural networks have achieved great successes in various machine learning tasks, people can hardly know what neural networks learn from data due to their black-box nature. The lack of such explainability is one of the limitations of neural networks when applied in domains, e.g., healthcare and finance, that demand transparency and accountability. Moreover, explainability is beneficial for guiding a neural network to learn the causal patterns that can extrapolate out-of-distribution (OOD) data, which is critical in real-world applications and has surged as a hot research topic.

User-User interaction recommendation, or interaction recommendation, is an indispensable service in social platforms, where the system automatically predicts with whom a user wants to interact. In real-world social platforms, we observe that user interactions may occur in diverse scenarios, and new scenarios constantly emerge, such as new games or sales promotions. There are two challenges in these emerging scenarios: (1) The behavior of users on the emerging scenarios could be different from existing ones due to the diversity among scenarios; (2) Emerging scenarios may only have scarce user behavioral data for model learning. Towards these two challenges, we present KoMen, a Domain Knowledge Guided Meta-learning framework for Interaction Recommendation. KoMen first learns a set of global model parameters shared among all scenarios and then quickly adapts the parameters for an emerging scenario based on its similarities with the existing ones. There are two highlights of KoMen: (1) KoMen customizes global model parameters by incorporating domain knowledge of the scenarios (e.g., a taxonomy that organizes scenarios by their purposes and functions), which captures scenario inter-dependencies with very limited training. (2) KoMen learns the scenario-specific parameters through a mixture-of-expert architecture, which reduces model variance resulting from data scarcity while still achieving the expressiveness to handle diverse scenarios. Extensive experiments demonstrate that KoMen achieves state-of-the-art performance on a public benchmark dataset and a large-scale real industry dataset. Remarkably, KoMen improves over the best baseline w.r.t. weighted ROC-AUC by 2.14% and 2.03% on the two datasets, respectively. Our code is available at: https://github.com/Veronicium/koMen.

Recently, Product Question Answering (PQA) on E-Commerce platforms has attracted increasing attention as it can act as an intelligent online shopping assistant and improve the customer shopping experience. Its key function, automatic answer generation for product-related questions, has been studied by aiming to generate content-preserving while question-related answers. However, an important characteristic of PQA, i.e., personalization, is neglected by existing methods. It is insufficient to provide the same “completely summarized” answer to all customers, since many customers are more willing to see personalized answers with customized information only for themselves, by taking into consideration their own preferences toward product aspects or information needs. To tackle this challenge, we propose a novel Personalized Answer GEneration method with multi-perspective preference modeling, which explores historical user-generated contents to model user preference for generating personalized answers in PQA. Specifically, we first retrieve question-related user history as external knowledge to model knowledge-level user preference. Then, we leverage the Gaussian Softmax distribution model to capture latent aspect-level user preference. Finally, we develop a persona-aware pointer network to generate personalized answers in terms of both content and style by utilizing personal user preference and dynamic user vocabulary. Experimental results on real-world E-Commerce QA datasets demonstrate that the proposed method outperforms existing methods by generating informative and customized answers and show that answer generation in E-Commerce can benefit from personalization.

Citation count prediction is an important task for estimating the future impact of research papers. Most of the existing works utilize the information extracted from the paper itself. In this article, we focus on how to utilize another kind of useful data signal (i.e., peer review text) to improve both the performance and interpretability of the prediction models. Specially, we propose a novel aspect-aware capsule network for citation count prediction based on review text. It contains two major capsule layers, namely the feature capsule layer and the aspect capsule layer, with two different routing approaches, respectively. Feature capsules encode the local semantics from review sentences as the input of aspect capsule layer, whereas aspect capsules aim to capture high-level semantic features that will be served as final representations for prediction. Besides the predictive capacity, we also enhance the model interpretability with two strategies. First, we use the topic distribution of the review text to guide the learning of aspect capsules so that each aspect capsule can represent a specific aspect in the review. Then, we use the learned aspect capsules to generate readable text for explaining the predicted citation count. Extensive experiments on two real-world datasets have demonstrated the effectiveness of the proposed model in both performance and interpretability.

Machine Learning methods have been adopted for a wide range of real-world applications, ranging from social networks, online image/video-sharing platforms, and e-commerce to education, healthcare, etc. However, in practice, a large amount of effort is required to tune several components of machine learning methods, including data representation, hyperparameter, and model architecture, in order to achieve a good performance. To alleviate the required tunning efforts, Automated Machine Learning (AutoML), which can automate the process of applying machine learning methods, has been studied in both academy and industry recently. In this tutorial, we will introduce the main research topics of AutoML, including Hyperparameter Optimization, Neural Architecture Search, and Meta-Learning. Two emerging topics of AutoML, Automatic Feature Generation and Machine Learning Guided Database, will also be discussed since they are important components for real-world applications. For each topic, we will motivate it with application examples from industry, illustrate the state-of-the-art methodologies, and discuss some future research directions based on our experience from industry and the trends in academy.

To improve user experience and profits of corporations, modern industrial recommender systems usually aim to select the items that are most likely to be interacted with (e.g., clicks and purchases). However, they overlook the fact that users may purchase the items even without recommendations. The real effective items are the ones that can contribute to purchase probability uplift. To select these effective items, it is essential to estimate the causal effect of recommendations. Nevertheless, it is difficult to obtain the real causal effect since we can only recommend or not recommend an item to a user at one time. Furthermore, previous works usually rely on the randomized controlled trial (RCT) experiment to evaluate their performance. However, it is usually not practicable in the recommendation scenario due to its expensive experimental cost. To tackle these problems, in this paper, we propose a causal collaborative filtering (CausCF) method inspired by the widely adopted collaborative filtering (CF) technique. It is based on the idea that similar users not only have a similar taste on items but also have similar treatment effects under recommendations. CausCF extends the classical matrix factorization to the tensor factorization with three dimensions---user, item, and treatment. Furthermore, we also employ regression discontinuity design (RDD) to evaluate the precision of the estimated causal effects from different models. With the testable assumptions, RDD analysis can provide an unbiased causal conclusion without RCT experiments. Through dedicated experiments on both offline and online experiments, we demonstrate the effectiveness of our proposed CausCF on the causal effect estimation and ranking performance improvement.

Machine learning methods have been adopted for various real-world applications, ranging from social networks, online image/video-sharing platforms, and e-commerce to education, healthcare, etc. However, several components of machine learning methods, including data representation, hyperparameter and model architecture, can largely affect their performance in practice. Moreover, the explosions of data scale and model size make the optimization of these components more and more time-consuming for machine learning developers. To tackle these challenges, Automated Machine Learning (AutoML) aims to automate the process of applying machine learning methods to solve real-world application tasks, reducing the time of tuning machine learning methods while maintaining good performance. In this tutorial, we will introduce the main research topics of AutoML, including Hyperparameter Optimization, Neural Architecture Search and Meta-Learning. Two emerging topics of AutoML, DNN-based Feature Generation and Machine Learning Guided Database, will also be discussed as they are important components for real-world applications. For each topic, we will motivate it with examples from industry, illustrate the state-of-the-art methods, and discuss their pros and cons from both perspectives of industry and academy. We will also discuss some future research directions based on our experience from industry and the trends in academy.

As it becomes prevalent that user information exists in multiple platforms or services, cross-domain recommendation has been an important task in industry. Although it is well known that users tend to show different preferences in different domains, existing studies seldom model how domain biases affect user preferences. Focused on this issue, we develop a casual-based approach to mitigating the domain biases when transferring the user information cross domains. To be specific, this paper presents a novel debiasing learning based cross-domain recommendation framework with causal embedding. In this framework, we design a novel Inverse-Propensity-Score (IPS) estimator designed for cross-domain scenario, and further propose three kinds of restrictions for propensity score learning. Our framework can be generally applied to various recommendation algorithms for cross-domain recommendation. Extensive experiments on both public and industry datasets have demonstrated the effectiveness of the proposed framework.

High-order interactive features capture the correlation between different columns and thus are promising to enhance various learning tasks on ubiquitous tabular data. To automate the generation of interactive features, existing works either explicitly traverse the feature space or implicitly express the interactions via intermediate activations of some designed models. These two kinds of methods show that there is essentially a trade-off between feature interpretability and search efficiency. To possess both of their merits, we propose a novel method named Feature Interaction Via Edge Search (FIVES), which formulates the task of interactive feature generation as searching for edges on the defined feature graph. Specifically, we first present our theoretical evidence that motivates us to search for useful interactive features with increasing order. Then we instantiate this search strategy by optimizing both a dedicated graph neural network (GNN) and the adjacency tensor associated with the defined feature graph. In this way, the proposed FIVES method simplifies the time-consuming traversal as a typical training course of GNN and enables explicit feature generation according to the learned adjacency tensor. Experimental results on both benchmark and real-world datasets show the advantages of FIVES over several state-of-the-art methods. Moreover, the interactive features identified by FIVES are deployed on the recommender system of Taobao, a worldwide leading e-commerce platform. Results of an online A/B testing further verify the effectiveness of the proposed method FIVES, and we further provide FIVES as AI utilities for the customers of Alibaba Cloud.

Conversational recommender systems (CRS) enable the traditional recommender systems to explicitly acquire user preferences towards items and attributes through interactive conversations. Reinforcement learning (RL) is widely adopted to learn conversational recommendation policies to decide what attributes to ask, which items to recommend, and when to ask or recommend, at each conversation turn. However, existing methods mainly target at solving one or two of these three decision-making problems in CRS with separated conversation and recommendation components, which restrict the scalability and generality of CRS and fall short of preserving a stable training procedure. In the light of these challenges, we propose to formulate these three decision-making problems in CRS as a unified policy learning task. In order to systematically integrate conversation and recommendation components, we develop a dynamic weighted graph based RL method to learn a policy to select the action at each conversation turn, either asking an attribute or recommending items. Further, to deal with the sample efficiency issue, we propose two action selection strategies for reducing the candidate action space according to the preference and entropy information. Experimental results on two benchmark CRS datasets and a real-world E-Commerce application show that the proposed method not only significantly outperforms state-of-the-art methods but also enhances the scalability and stability of CRS.

The maximum independent set problem is one of the most fundamental problems in graph algorithms and has been widely studied in social networks. The weighted version of this problem, where each vertex is assigned a nonnegative weight, also receives a lot of attention due to its potential applications in many areas. However, many nice properties and fast algorithms for the unweighted version can not be extended to the weighted version. In this paper, we study the structural properties of this problem, giving some sufficient conditions for a vertex being or not being in a maximum weighted independent set. These properties provide a suite of reduction rules that includes and generalizes almost all frequently used reduction rules for this problem. These rules can efficiently find partial solutions and greatly reduce the instances, especially for sparse graphs. Based on them, we also propose a simple exact yet practical algorithm. To demonstrate the efficiency of our algorithm, we compare it with state-of-the-art algorithms on several well-known datasets from the real world. The experimental results reveal that our exact algorithm is not only faster than existing algorithms but also can exactly solve more hard instances with 1,000 seconds. For remaining infeasible instances, our reduction rules can also improve existing heuristic algorithms by producing higher-quality solutions using less time.

Local differential privacy (LDP) is a well-established privacy protection scheme for collecting sensitive data, which has been integrated into major platforms such as iOS, Chrome, and Windows. The main idea is that each individual randomly perturbs her data on her local device, and only uploads the noisy version to an untrusted data aggregator. This paper focuses on the collection of streaming data consisting of regular updates, e.g. , daily app usage. Such streams, when aggregated over a large population, often exhibit strong autocorrelations , e.g. , the average usage of an app usually does not change dramatically from one day to the next. To our knowledge, this property has been largely neglected in existing LDP mechanisms. Consequently, data collected with current LDP methods often exhibit unrealistically violent fluctuations due to the added noise, drowning the overall trend, as shown in our experiments. This paper proposes a novel correlated Gaussian mechanism ( CGM ) for enforcing (ϵ, δ)-LDP on streaming data collection, which reduces noise by exploiting public-known autocorrelation patterns of the aggregated data. This is done through non-trivial modifications to the core of the underlying Gaussian Mechanism; in particular, CGM injects temporally correlated noise, computed through an optimization program that takes into account the given autocorrelation pattern, data value range, and utility metric. CGM comes with formal proof of correctness, and consumes negligible computational resources. Extensive experiments using real datasets from different application domains demonstrate that CGM achieves consistent and significant utility gains compared to the baseline method of repeatedly running the underlying one-shot LDP mechanism.

Matrix factorization (MF) approximates unobserved ratings in a rating matrix, whose rows correspond to users and columns correspond to items to be rated, and has been serving as a fundamental building block in recommendation systems. This paper comprehensively studies the problem of matrix factorization in different federated learning (FL) settings, where a set of parties want to cooperate in training but refuse to share data directly. We first propose a generic algorithmic framework for various settings of federated matrix factorization (FMF) and provide a theoretical convergence guarantee. We then systematically characterize privacy-leakage risks in data collection, training, and publishing stages for three different settings and introduce privacy notions to provide end-to-end privacy protections. The first one is vertical federated learning (VFL), where multiple parties have the ratings from the same set of users but on disjoint sets of items. The second one is horizontal federated learning (HFL), where parties have ratings from different sets of users but on the same set of items. The third setting is local federated learning (LFL), where the ratings of the users are only stored on their local devices. We introduce adapted versions of FMF with the privacy notions guaranteed in the three settings. In particular, a new private learning technique called embedding clipping is introduced and used in all the three settings to ensure differential privacy. For the LFL setting, we combine differential privacy with secure aggregation to protect the communication between user devices and the server with a strength similar to the local differential privacy model, but much better accuracy. We perform experiments to demonstrate the effectiveness of our approaches.

End-to-end AutoML has attracted intensive interests from both academia and industry, which automatically searches for ML pipelines in a space induced by feature engineering, algorithm/model selection, and hyper-parameter tuning. Existing AutoML systems, however, suffer from scalability issues when applying to application domains with large, high-dimensional search spaces. We present VOLCANOML, a scalable and extensible framework that facilitates systematic exploration of large AutoML search spaces. VOLCANOML introduces and implements basic building blocks that decompose a large search space into smaller ones, and allows users to utilize these building blocks to compose an execution plan for the AutoML problem at hand. VOLCANOML further supports a Volcano-style execution model - akin to the one supported by modern database systems - to execute the plan constructed. Our evaluation demonstrates that, not only does VOLCANOML raise the level of expressiveness for search space decomposition in AutoML, it also leads to actual findings of decomposition strategies that are significantly more efficient than the ones employed by state-of-the-art AutoML systems such as auto-sklearn.

Policy-aware differential privacy (DP) frameworks such as Blowfish privacy enable more accurate query answers than standard DP. In this work, we build the first policy-aware DP system for interactive data exploration, BlowfishDB, that aims to (i) provide bounded and flexible privacy guarantees to the data curators of sensitive data and (ii) support accurate and efficient data exploration by data analysts. However, the specification and processing of customized privacy policies incur additional performance cost, especially for datasets with a large domain. To address this challenge, we propose dynamic Blowfish privacy which allows for the dynamic generation of smaller privacy policies and their data representations at query time. BlowfishDB ensures same levels of accuracy and privacy as one would get working on the static privacy policy. In this demonstration of BlowfishDB, we show how a data curator can fine-tune privacy policies for a sensitive dataset and how a data analyst can retrieve accuracy-bounded query answers efficiently without being a privacy expert.

Estimating the number of distinct values (NDV) in a column is useful for many tasks in database systems, such as columnstore compression and data profiling. In this work, we focus on how to derive accurate NDV estimations from random (online/offline) samples. Such efficient estimation is critical for tasks where it is prohibitive to scan the data even once. Existing sample-based estimators typically rely on heuristics or assumptions and do not have robust performance across different datasets as the assumptions on data can easily break. On the other hand, deriving an estimator from a principled formulation such as maximum likelihood estimation is very challenging due to the complex structure of the formulation. We propose to formulate the NDV estimation task in a supervised learning framework, and aim to learn a model as the estimator. To this end, we need to answer several questions: i) how to make the learned model workload agnostic; ii) how to obtain training data; iii) how to perform model training. We derive conditions of the learning framework under which the learned model isworkload agnostic, in the sense that the model/estimator can be trained with synthetically generated training data, and then deployed into any data warehouse simply as,e.g., user-defined functions (UDFs), to offer efficient (within microseconds on CPU) and accurate NDV estimations forunseen tables and workloads.We compare the learned estimator with the state-of-the-art sample-based estimators on nine real-world datasets to demonstrate its superior estimation accuracy. We publish our code for training data generation, model training, and the learned estimator online for reproducibility.

Interactive response time is important in analytical pipelines for users to explore a sufficient number of possibilities and make informed business decisions. We consider a forecasting pipeline with large volumes of high-dimensional time series data. Real-time forecasting can be conducted in two steps. First, we specify the part of data to be focused on and the measure to be predicted by slicing, dicing, and aggregating the data. Second, a forecasting model is trained on the aggregated results to predict the trend of the specified measure. While there are a number of forecasting models available, the first step is the performance bottleneck. A natural idea is to utilize sampling to obtain approximate aggregations in real time as the input to train the forecasting model. Our scalable real-time forecasting system FlashP (Flash Prediction) is built based on this idea, with two major challenges to be resolved in this paper: first, we need to figure out how approximate aggregations affect the fitting of forecasting models, and forecasting results; and second, accordingly, what sampling algorithms we should use to obtain these approximate aggregations and how large the samples are. We introduce a new sampling scheme, called GSW sampling, and analyze error bounds for estimating aggregations using GSW samples. We introduce how to construct compact GSW samples with the existence of multiple measures to be analyzed. We conduct experiments to evaluate our solution its alternatives on real data.

With the rapid development of smartphones, spatial crowdsourcing platforms are getting popular. A foundational research of spatial crowdsourcing is to allocate micro-tasks to suitable crowd workers. Many existing studies focus on the offline scenario, where all the spatiotemporal information of micro-tasks and crowd workers is given. In this paper, we focus on the online scenario and identify a more practical micro-task allocation problem, called theGlobalOnlineMicro-taskAllocation in spatial crowdsourcing(GOMA) problem. We first extend the state-of-the-art algorithm for the online maximum weighted bipartite matching problem to the GOMA problem as the baseline algorithm. Although the baseline algorithm provides a theoretical guarantee for the worst case, its average performance in practice is not good enough since the worst case happens with a very low probability in the real world. Thus, we consider the average performance of online algorithms,a.k.a.random order model. We propose a two-phase-based framework, based on which we present the TGOA algorithm with a$\frac{1}{4}$-competitive ratio under the random order model. To improve its efficiency, we further design the TGOA-Greedy and TGOA-OP algorithm following this framework, which runs faster than the TGOA algorithm with a competitive ratio of$\frac{1}{8}$and$\frac{1}{4}$, respectively. We also revisit the average performance of Greedy, which has long been considered as the worst due to its unbounded competitive ratio in the worst case. Finally, we verify the effectiveness and efficiency of the proposed methods through extensive experiments on synthetic and real datasets.

Heterogeneous information networks (HINs) have been ubiquitous in the real-world. HIN embeddings, which encode various information of the networks into low-dimensional vectors, can facilitate a wide range of applications on graph-structured data. Existing HIN embedding methods include random walk based methods that may not fully utilize the edge semantics and knowledge graph embedding methods that restrict the expression ability of topological information. In this paper, we propose a novel adaptive embedding framework, which integrates these two kinds of methods to preserve both topological information and relational information. By incorporating an assistant knowledge graph embedding model, the proposed framework performs efficient biased random walk under the guidance of edge semantics.

Nowadays, crowd sensing becomes increasingly more popular due to the ubiquitous usage of mobile devices. However, the quality of such human-generated sensory data varies significantly among different users. To better utilize sensory data, the problem of truth discovery, whose goal is to estimate user quality and infer reliable aggregated results through quality-aware data aggregation, has emerged as a hot topic. Although the existing truth discovery approaches can provide reliable aggregated results, they fail to protect the private information of individual users. Moreover, crowd sensing systems typically involve a large number of participants, making encryption or secure multi-party computation based solutions difficult to deploy. To address these challenges, in this paper, we propose an efficient privacy-preserving truth discovery mechanism with theoretical guarantees of both utility and privacy. The key idea of the proposed mechanism is to perturb data from each user independently and then conduct weighted aggregation among users’ perturbed data. The proposed approach is able to assign user weights based on information quality, and thus the aggregated results will not deviate much from the true results even when large noise is added. We adapt local differential privacy definition to this privacy-preserving task and demonstrate the proposed mechanism can satisfy local differential privacy while preserving high aggregation accuracy. We formally quantify utility and privacy trade-off and further verify the claim by experiments on both synthetic data and a real-world crowd sensing system.

Graph convolutional networks (GCNs) are a powerful deep learning approach for graph-structured data. Recently, GCNs and subsequent variants have shown superior performance in various application areas on real-world datasets. Despite their success, most of the current GCN models are shallow, due to the \emph{over-smoothing} problem. In this paper, we study the problem of designing and analyzing deep graph convolutional networks. We propose the GCNII, an extension of the vanilla GCN model with two simple yet effective techniques: \emph{Initial residual} and \emph{Identity mapping}. We provide theoretical and empirical evidence that the two techniques effectively relieves the problem of over-smoothing. Our experiments show that the deep GCNII model outperforms the state-of-the-art methods on various semi- and full-supervised tasks.

Large pre-trained language models such as BERT have shown their effectiveness in various natural language processing tasks. However, the huge parameter size makes them difficult to be deployed in real-time applications that require quick inference with limited resources. Existing methods compress BERT into small models while such compression is task-independent, i.e., the same compressed BERT for all different downstream tasks. Motivated by the necessity and benefits of task-oriented BERT compression, we propose a novel compression method, AdaBERT, that leverages differentiable Neural Architecture Search to automatically compress BERT into task-adaptive small models for specific tasks. We incorporate a task-oriented knowledge distillation loss to provide search hints and an efficiency-aware loss as search constraints, which enables a good trade-off between efficiency and effectiveness for task-adaptive BERT compression. We evaluate AdaBERT on several NLP tasks, and the results demonstrate that those task-adaptive compressed models are 12.7x to 29.3x faster than BERT in inference time and 11.5x to 17.0x smaller in terms of parameter size, while comparable performance is maintained.

Accurately characterizing the user's current interest is the core of recommender systems. However, users' interests are dynamic and affected by intent factors and preference factors. The intent factors imply users' current needs and change among different visits. The preference factors are relatively stable and learned continuously over time. Existing works either resort to the sequential recommendation to model the current browsing intent and historical preference separately or just mix up these two factors during online learning. In this paper, we propose a novel learning strategy named FLIP to decouple the learning of intent and preference under the online settings. The learning of the intent is considered as a meta-learning task and fast adaptive to the current browsing; the learning of the preference is based on the calibrated user intent and constantly updated over time. We conducted experiments on two public datasets and a real-world recommender system. When equipping it with modern recommendation methods, significant improvements are demonstrated over strong baselines.

Local differential privacy (LDP) enables private data sharing and analytics without the need for a trusted data collector. Error-optimal primitives (for, e.g., estimating means and item frequencies) under LDP have been well studied. For analytical tasks such as range queries, however, the best known error bound is dependent on the domain size of private data, which is potentially prohibitive. This deficiency is inherent as LDP protects the same level of indistinguishability between any pair of private data values for each data downer.In this paper, we utilize an extension of -LDP called MetricLDP or E-LDP, where a metric E defines heterogeneous privacy guarantees for different pairs of private data values and thus provides a more flexible knob than does to relax LDP and tune utility-privacy trade-offs. We show that, under such privacy relaxations, for analytical workloads such as linear counting, multi-dimensional range counting queries, and quantile queries, we can achieve significant gains in utility. In particular, for range queries under E-LDP where the metric E is the L1distance function scaled by , we design mechanisms with errors independent on the domain sizes; instead, their errors depend on the metric E, which specifies in what granularity the private data is protected. We believe that the primitives we design for ELDP will be useful in developing mechanisms for other analytical tasks, and encourage the adoption of LDP in practice.

Graph Neural Networks (GNN) are an emerging field for learning on non-Euclidean data. Recently, there has been increased interest in designing GNN that scales to large graphs. Most existing methods use "graph sampling" or "layer-wise sampling" techniques to reduce training time; However, these methods still suffer from degrading performance and scalability problems when applying to graphs with billions of edges. In this paper, we present GBP, a scalable GNN that utilizes a localized bidirectional propagation process from both the feature vector and the training/testing nodes. Theoretical analysis shows that GBP is the first method that achieves sub-linear time complexity for both the precomputation and the training phases. An extensive empirical study demonstrates that GBP achieves state-of-the-art performance with significantly less training/testing time. Most notably, GBP is able to deliver superior performance on a graph with over 60 million nodes and 1.8 billion edges in less than 2,000 seconds on a single machine.

Computing the gradient of model hyperparameters, i.e., hypergradient, enables a promising and natural way to solve the hyperparameter optimization task. However, gradient-based methods could lead to suboptimal solutions due to the non-convex nature of optimization in a complex hyperparameter space. In this study, we propose a hyperparameter mutation (HPM) algorithm to explicitly consider a learnable trade-off between using global and local search, where we adopt a population of student models to simultaneously explore the hyperparameter space guided by hypergradient and leverage a teacher model to mutate the underperforming students by exploiting the top ones. The teacher model is implemented with an attention mechanism and is used to learn a mutation schedule for different hyperparameters on the fly. Empirical evidence on synthetic functions is provided to show that HPM outperforms hypergradient significantly. Experiments on two benchmark datasets are also conducted to validate the effectiveness of the proposed HPM algorithm for training deep neural networks compared with several strong baselines.

Recently, deep learning has made significant progress in the task of sequential recommendation. Existing neural sequential recommenders typically adopt a generative way trained with Maximum Likelihood Estimation (MLE). When context information (called factor) is involved, it is difficult to analyze when and how each individual factor would affect the final recommendation performance.

Online recommendation systems make use of a variety of information sources to provide users the items that users are potentially interested in. However, due to the openness of the online platform, recommendation systems are vulnerable to data poisoning attacks. Existing attack approaches are either based on simple heuristic rules or designed against specific recommendations approaches. The former often suffers unsatisfactory performance, while the latter requires strong knowledge of the target system. In this paper, we focus on a general next-item recommendation setting and propose a practical poisoning attack approach named LOKI against blackbox recommendation systems. The proposed LOKI utilizes the reinforcement learning algorithm to train the attack agent, which can be used to generate user behavior samples for data poisoning. In real-world recommendation systems, the cost of retraining recommendation models is high, and the interaction frequency between users and a recommendation system is restricted. Given these real-world restrictions, we propose to let the agent interact with a recommender simulator instead of the target recommendation system and leverage the transferability of the generated adversarial samples to poison the target system. We also propose to use the influence function to efficiently estimate the influence of injected samples on the recommendation results, without re-training the models within the simulator. Extensive experiments on two datasets against four representative recommendation models show that the proposed LOKI achieves better attacking performance than existing methods.

When collecting information, local differential privacy (LDP) alleviates privacy concerns of users because their private information is randomized before being sent it to the central aggregator. LDP imposes large amount of noise as each user executes the randomization independently. To address this issue, recent work introduced an intermediate server with the assumption that this intermediate server does not collude with the aggregator. Under this assumption, less noise can be added to achieve the same privacy guarantee as LDP, thus improving utility for the data collection task. This paper investigates this multiple-party setting of LDP. We analyze the system model and identify potential adversaries. We then make two improvements: a new algorithm that achieves a better privacy-utility tradeoff; and a novel protocol that provides better protection against various attacks. Finally, we perform experiments to compare different methods and demonstrate the benefits of using our proposed method.

A configuration of training refers to the combinations of feature engineering, learner, and its associated hyperparameters. Given a set of configurations and a large dataset randomly split into training and testing set, we study how to efficiently identify the best configuration with approximately the highest testing accuracy when trained from the training set. To guarantee small accuracy loss, we develop a solution using confidence interval (CI)-based progressive sampling and pruning strategy. Compared to using full data to find the exact best configuration, our solution achieves more than two orders of magnitude speedup, while the returned top configuration has identical or close test accuracy.

Differential privacy has emerged as the main definition for private data analysis and machine learning. The global model of differential privacy, which assumes that users trust the data collector, provides strong privacy guarantees and introduces small errors in the output. In contrast, applications of differential privacy in commercial systems by Apple, Google, and Microsoft, use the local model. Here, users do not trust the data collector, and hence randomize their data before sending it to the data collector. Unfortunately, local model is too strong for several important applications and hence is limited in its applicability. In this work, we propose a framework based on trusted processors and a new definition of differential privacy called Oblivious Differential Privacy, which combines the best of both local and global models. The algorithms we design in this framework show interesting interplay of ideas from the streaming algorithms, oblivious algorithms, and differential privacy.

Multi-dimensional analytical (MDA) queries are often issued against a fact table with predicates on (categorical or ordinal) dimensions and aggregations on one or more measures. In this paper, we study the problem of answering MDA queries under local differential privacy (LDP). In the absence of a trusted agent, sensitive dimensions are encoded in a privacy-preserving (LDP) way locally before being sent to the data collector. The data collector estimates the answers to MDA queries, based on the encoded dimensions. We propose several LDP encoders and estimation algorithms, to handle a large class of MDA queries with different types of predicates and aggregation functions. Our techniques are able to answer these queries with tight error bounds and scale well in high-dimensional settings (i.e., error is polylogarithmic in dimension sizes). We conduct experiments on real and synthetic data to verify our theoretical results, and compare our solution with marginal-estimation based solutions.

Differential privacy has emerged as the de facto standard for privacy definitions, and been used by, e.g ., Apple, Google, Uber, and Microsoft, to collect sensitive information about users and to build privacy-preserving analytics engines. However, most of such advanced privacy-protection techniques are not accessible to mid-size companies and app developers in the cloud. We demonstrate a lightweight middleware DPSAaS , which provides d ifferentially p rivate data-sharing-and-analytics functionality a s cloud services. We focus on multi-dimensional analytical (MDA) queries under local differential privacy (LDP) in this demo. MDA queries against a fact table have predicates on (categorical or ordinal) dimensions and aggregate one or more measures. In the absence of a trusted agent, sensitive dimensions and measures are encoded in a privacy-preserving way locally using our LDP data sharing service, before being sent to the data collector. The data collector estimates the answers to MDA queries from the encoded data, using our data analytics service. We will highlight the design decisions of DPSAaS and twists made to LDA algorithms to fit the design, in order to smoothly connect DPSAaS to the data processing platform and analytics engines, and to facilitate efficient large-scale processing.

A statistical hypothesis test determines whether a hypothesis should be rejected based on samples from populations. In particular, randomized controlled experiments (or A/B testing) that compare population means using, e.g., t-tests, have been widely deployed in technology companies to aid in making data-driven decisions. Samples used in these tests are collected from users and may contain sensitive information. Both the data collection and the testing process may compromise individuals’ privacy. In this paper, we study how to conduct hypothesis tests to compare population means while preserving privacy. We use the notation of local differential privacy (LDP), which has recently emerged as the main tool to ensure each individual’s privacy without the need of a trusted data collector. We propose LDP tests that inject noise into every user’s data in the samples before collecting them (so users do not need to trust the data collector), and draw conclusions with bounded type-I (significance level) and type-II errors (1 - power). Our approaches can be extended to the scenario where some users require LDP while some are willing to provide exact data. We report experimental results on real-world datasets to verify the effectiveness of our approaches.

Soliciting answers from online users is an efficient and effective solution to many challenging tasks. Due to the variety in the quality of users, it is important to infer their ability to provide correct answers during aggregation. Therefore, truth discovery methods can be used to automatically capture the user quality and aggregate user-contributed answers via a weighted combination. Despite the fact that truth discovery is an effective tool for answer aggregation, existing work falls short of the protection towards the privacy of participating users. To fill this gap, we propose perturbation-based mechanisms that provide users with privacy guarantees and maintain the accuracy of aggregated answers. We first present a one-layer mechanism, in which all the users adopt the same probability to perturb their answers. Aggregation is then conducted on perturbed answers but the aggregation accuracy could drop accordingly. To improve the utility, a two-layer mechanism is proposed where users are allowed to sample their own probabilities from a hyper distribution. We theoretically compare the one-layer and two-layer mechanisms, and prove that they provide the same privacy guarantee while the two-layer mechanism delivers better utility. This advantage is brought by the fact that the two-layer mechanism can utilize the estimated user quality information from truth discovery to reduce the accuracy loss caused by perturbation, which is confirmed by experimental results on real-world datasets. Experimental results also demonstrate the effectiveness of the proposed two-layer mechanism in privacy protection with tolerable accuracy loss in aggregation.

Commercial DBMSs, such as Microsoft SQL Server, cater to diverse workloads including transaction processing, decision support, and operational analytics. They also support variety in physical design structures such as B+ tree and columnstore. The benefits of B+ tree for OLTP workloads and columnstore for decision support workloads are well-understood. However, the importance of hybrid physical designs, consisting of both columnstore and B+ tree indexes on the same database, is not well-studied --- a focus of this paper. We first quantify the trade-offs using carefully-crafted micro-benchmarks. This micro-benchmarking indicates that hybrid physical designs can result in orders of magnitude better performance depending on the workload. For complex real-world applications, choosing an appropriate combination of columnstore and B+ tree indexes for a database workload is challenging. We extend the Database Engine Tuning Advisor for Microsoft SQL Server to recommend a suitable combination of B+ tree and columnstore indexes for a given workload. Through extensive experiments using industry-standard benchmarks and several real-world customer workloads, we quantify how a physical design tool capable of recommending hybrid physical designs can result in orders of magnitude better execution costs compared to approaches that rely either on columnstore-only or B+ tree-only designs.

Efficiently estimating the inclusion coefficient - the fraction of values of one column that are contained in another column - is useful for tasks such as data profiling and foreign-key detection. We present a new estimator, BML, for inclusion coefficient based on Hyperloglog sketches that results in significantly lower error compared to the state-of-the art approach that uses Bottom-k sketches. We evaluate the error of the BML estimator using experiments on industry benchmarks such as TPC-H and TPC-DS, and several real-world databases. As an independent contribution, we show how Hyperloglog sketches can be maintained incrementally with data deletions using only a constant amount of additional memory.

Analysts need interactive speed for exploratory analysis, but big data systems are often slow. With sampling, data systems can produce approximate answers fast enough for exploratory visualization, at the cost of accuracy and trust. We propose optimistic visualization, which approaches these issues from a user experience perspective. This method lets analysts explore approximate results interactively, and provides a way to detect and recover from errors later. Pangloss implements these ideas. We discuss design issues raised by optimistic visualization systems. We test this concept with five expert visualizers in a laboratory study and three case studies at Microsoft. Analysts reported that they felt more confident in their results, and used optimistic visualization to check that their preliminary results were correct.

The collection and analysis of telemetry data from user's devices is routinely performed by many software companies. Telemetry collection leads to improved user experience but poses significant risks to users' privacy. Locally differentially private (LDP) algorithms have recently emerged as the main tool that allows data collectors to estimate various population statistics, while preserving privacy. The guarantees provided by such algorithms are typically very strong for a single round of telemetry collection, but degrade rapidly when telemetry is collected regularly. In particular, existing LDP algorithms are not suitable for repeated collection of counter data such as daily app usage statistics. In this paper, we develop new LDP mechanisms geared towards repeated collection of counter data, with formal privacy guarantees even after being executed for an arbitrarily long period of time. For two basic analytical tasks, mean estimation and histogram estimation, our LDP mechanisms for repeated data collection provide estimates with comparable or even the same accuracy as existing single-round LDP collection mechanisms. We conduct empirical evaluation on real-world counter datasets to verify our theoretical results. Our mechanisms have been deployed by Microsoft to collect telemetry across millions of devices.

In this paper, we reflect on the state of the art of Approximate Query Processing. Although much technical progress has been made in this area of research, we are yet to see its impact on products and services. We discuss two promising avenues to pursue towards integrating Approximate Query Processing into data platforms.

The popularity of Online To Offline (O2O) service platforms has spurred the need for online task assignment in real-time spatial data, where streams of spatially distributed tasks and workers are matched in real time such that the total number of assigned pairs is maximized. Existing online task assignment models assume that each worker is either assigned a task immediately or waits for a subsequent task at a fixed location once she/he appears on the platform. Yet in practice a worker may actively move around rather than passively wait in place if no task is assigned. In this paper, we define a new problem Flexible Two-sided Online task Assignment (FTOA). FTOA aims to guide idle workers based on the prediction of tasks and workers so as to increase the total number of assigned worker-task pairs. To address the FTOA problem, we face two challenges: (i) How to generate guidance for idle workers based on the prediction of the spatiotemporal distribution of tasks and workers? (ii) How to leverage the guidance of workers' movements to optimize the online task assignment? To this end, we propose a novel two-step framework, which integrates offline prediction and online task assignment. Specifically, we estimate the distributions of tasks and workers per time slot and per unit area, and design an online task assignment algorithm, Prediction-oriented Online task Assignment in Real-time spatial data (POLAR-OP). It yields a 0.47-competitive ratio, which is nearly twice better than that of the state-of-the-art. POLAR-OP also reduces the time complexity to process each newly-arrived task/worker to O(1). We validate the effectiveness and efficiency of our methods via extensive experiments on both synthetic datasets and real-world datasets from a large-scale taxi-calling platform.

Recent years have witnessed the proliferation of mobile crowd sensing (MCS) systems that leverage the public crowd equipped with various mobile devices (e.g., smartphones, smartglasses, smartwatches) for large scale sensing tasks. Because of the importance of incentivizing worker participation in such MCS systems, several auction-based incentive mechanisms have been proposed in past literature. However, these mechanisms fail to consider the preservation of workers' bid privacy. Therefore, different from prior work, we propose a differentially private incentive mechanism that preserves the privacy of each worker's bid against the other honest-but-curious workers. The motivation of this design comes from the concern that a worker's bid usually contains her private information that should not be disclosed. We design our incentive mechanism based on the single-minded reverse combinatorial auction. Specifically, we design a differentially private, approximately truthful, individual rational, and computationally efficient mechanism that approximately minimizes the platform's total payment with a guaranteed approximation ratio. The advantageous properties of the proposed mechanism are justified through not only rigorous theoretical analysis but also extensive simulations.

With the rapid development of smartphones, spatial crowdsourcing platforms are getting popular. A foundational research of spatial crowdsourcing is to allocate micro-tasks to suitable crowd workers. Most existing studies focus on offline scenarios, where all the spatiotemporal information of micro-tasks and crowd workers is given. However, they are impractical since micro-tasks and crowd workers in real applications appear dynamically and their spatiotemporal information cannot be known in advance. In this paper, to address the shortcomings of existing offline approaches, we first identify a more practical micro-task allocation problem, called the Global Online Micro-task Allocation in spatial crowdsourcing (GOMA) problem. We first extend the state-of-art algorithm for the online maximum weighted bipartite matching problem to the GOMA problem as the baseline algorithm. Although the baseline algorithm provides theoretical guarantee for the worst case, its average performance in practice is not good enough since the worst case happens with a very low probability in real world. Thus, we consider the average performance of online algorithms, a.k.a online random order model.We propose a two-phase-based framework, based on which we present the TGOA algorithm with 1 over 4 -competitive ratio under the online random order model. To improve its efficiency, we further design the TGOA-Greedy algorithm following the framework, which runs faster than the TGOA algorithm but has lower competitive ratio of 1 over 8. Finally, we verify the effectiveness and efficiency of the proposed methods through extensive experiments on real and synthetic datasets.

Data volumes are growing exponentially for our decision-support systems making it challenging to ensure interactive response time for ad-hoc queries without increasing cost of hardware. Aggregation queries with Group By that produce an aggregate value for every combination of values in the grouping columns are the most important class of ad-hoc queries. As small errors are usually tolerable for such queries, approximate query processing (AQP) has the potential to answer them over very large datasets much faster.

We present a system that approximates the answer to complex ad-hoc queries in big-data clusters by injecting samplers on-the-fly and without requiring pre-existing samples. Improvements can be substantial when big-data queries take multiple passes over data and when samplers execute early in the query plan. We present a new, universe, sampler which is able to sample multiple join inputs. By incorporating samplers natively into a cost-based query optimizer, we automatically generate plans with appropriate samplers at appropriate locations. We devise an accuracy analysis method using which we ensure that query plans with samplers will not miss groups and that aggregate values are within a small ratio of their true value. An implementation on a cluster with tens of thousands of machines shows that queries in the TPC-DS benchmark use a median of 2X fewer resources. In contrast, approaches that construct input samples even when given 10X the size of the input to store samples improve only 22% of the queries, i.e., a median speed up of 0X.

We describe the design and implementation of the new Live Query Statistics (LQS) feature in Microsoft SQL Server 2016. The functionality includes the display of overall query progress as well as progress of individual operators in the query execution plan. We describe the overall functionality of LQS, give usage examples and detail all areas where we had to extend the current state-of-the-art to build the complete LQS feature. Finally, we evaluate the effect these extensions have on progress estimation accuracy with a series of experiments using a large set of synthetic and real workloads.

Recently, with the development of mobile Internet and smartphones, the online minimum bipartite matching in real time spatial data (OMBM) problem becomes popular. Specifically, given a set of service providers with specific locations and a set of users who dynamically appear one by one, the OMBM problem is to find a maximum-cardinality matching with minimum total distance following that once a user appears, s/he must be immediately matched to an unmatched service provider, which cannot be revoked, before subsequent users arrive. To address this problem, existing studies mainly focus on analyzing the worst-case competitive ratios of the proposed online algorithms, but study on the performance of the algorithms in practice is absent. In this paper, we present a comprehensive experimental comparison of the representative algorithms of the OMBM problem. Particularly, we observe a surprising result that the simple and efficient greedy algorithm, which has been considered as the worst due to its exponential worst-case competitive ratio, is significantly more effective than other algorithms. We investigate the results and further show that the competitive ratio of the worst case of the greedy algorithm is actually just a constant, 3.195, in the average-case analysis. We try to clarify a 25-year misunderstanding towards the greedy algorithm and justify that the greedy algorithm is not bad at all. Finally, we provide a uniform implementation for all the algorithms of the OMBM problem and clarify their strengths and weaknesses, which can guide practitioners to select appropriate algorithms for various scenarios.

An enterprise information worker is often aware of a few example tuples that should be present in the output of the query. Query discovery systems have been developed to discover project-join queries that contain the given example tuples in their output. However, they require the output to exactly contain all the example tuples and do not perform any ranking. To address this limitation, we study the problem of efficiently discovering top-k project join queries which approximately contain the given example tuples in their output. We extend our algorithms to incrementally produce results as soon as the user finishes typing/modifying a cell. Our experiments on real-life and synthetic datasets show that our proposed solution is significantly more efficient compared with applying state-of-the-art algorithms.

The problem of designing error optimal differentially private algorithms is well studied. Recent work applying differential privacy to real world settings have used variants of differential privacy that appropriately modify the notion of neighboring databases. The problem of designing error optimal algorithms for such variants of differential privacy is open. In this paper, we show a novel transformational equivalence result that can turn the problem of query answering under differential privacy with a modified notion of neighbors to one of query answering under standard differential privacy, for a large class of neighbor definitions. We utilize the Blowfish privacy framework that generalizes differential privacy. Blowfish uses a policy graph to instantiate different notions of neighboring databases. We show that the error incurred when answering a workload W on a database x under a Blowfish policy graph G is identical to the error required to answer a transformed workload f G ( W ) on database g G ( x ) under standard differential privacy, where f G and g G are linear transformations based on G. Using this result, we develop error efficient algorithms for releasing histograms and multidimensional range queries under different Blowfish policies. We believe the tools we develop will be useful for finding mechanisms to answer many other classes of queries with low error under other policy graphs.

Large-scale data center networks are complex---comprising several thousand network devices and several hundred thousand links---and form the critical infrastructure upon which all higher-level services depend on. Despite the built-in redundancy in data center networks, performance issues and device or link failures in the network can lead to user-perceived service interruptions. Therefore, determining and localizing user-impacting availability and performance issues in the network in near real time is crucial. Traditionally, both passive and active monitoring approaches have been used for failure localization. However, data from passive monitoring is often too noisy and does not effectively capture silent or gray failures, whereas active monitoring is potent in detecting faults but limited in its ability to isolate the exact fault location depending on its scale and granularity.

Privacy definitions provide ways for trading-off the privacy of individuals in a statistical database for the utility of downstream analysis of the data. In this paper, we present Blowfish, a class of privacy definitions inspired by the Pufferfish framework, that provides a rich interface for this trade-off. In particular, we allow data publishers to extend differential privacy using a policy, which specifies (a) secrets, or information that must be kept secret, and (b) constraints that may be known about the data. While the secret specification allows increased utility by lessening protection for certain individual properties, the constraint specification provides added protection against an adversary who knows correlations in the data (arising from constraints). We formalize policies and present novel algorithms that can handle general specifications of sensitive information and certain count constraints. We show that there are reasonable policies under which our privacy mechanisms for k-means clustering, histograms and range queries introduce significantly lesser noise than their differentially private counterparts. We quantify the privacy-utility trade-offs for various policies analytically and empirically on real datasets.

An enterprise information worker is often aware of a few example tuples (but not the entire result) that should be present in the output of the query. We study the problem of discovering the minimal project join query that contains the given example tuples in its output. Efficient discovery of such queries is challenging. We propose novel algorithms to solve this problem. Our experiments on real-life datasets show that the proposed solution is significantly more efficient compared with na\"{i}ve adaptations of known techniques.

We aim to provide table answers to keyword queries using a knowledge base. For queries referring to multiple entities, like "Washington cities population" and "Mel Gibson movies", it is better to represent each relevant answer as a table which aggregates a set of entities or joins of entities within the same table scheme or pattern. In this paper, we study how to find highly relevant patterns in a knowledge base for user-given keyword queries to compose table answers. A knowledge base is modeled as a directed graph called knowledge graph, where nodes represent its entities and edges represent the relationships among them. Each node/edge is labeled with type and text. A pattern is an aggregation of subtrees which contain all keywords in the texts and have the same structure and types on node/edges. We propose efficient algorithms to find patterns that are relevant to the query for a class of scoring functions. We show the hardness of the problem in theory, and propose path-based indexes that are affordable in memory. Two query-processing algorithms are proposed: one is fast in practice for small queries (with small numbers of patterns as answers) by utilizing the indexes; and the other one is better in theory, with running time linear in the sizes of indexes and answers, which can handle large queries better. We also conduct extensive experimental study to compare our approaches with a naive adaption of known techniques.

A large portion of real world data is either text or structured (e.g., relational) data. Moreover, such data objects are often linked together (e.g., structured specification of products linking with the corresponding product descriptions and customer comments). Even for text data such as news data, typed entities can be extracted with entity extraction tools. The EventCube project constructs TextCube and TopicCube from interconnected structured and text data (or from text data via entity extraction and dimension building), and performs multidimensional search and analysis on such datasets, in an informative, powerful, and user-friendly manner. This proposed EventCube demo will show the power of the system not only on the originally designed ASRS (Aviation Safety Report System) data sets, but also on news datasets collected from multiple news agencies, and academic datasets constructed from the DBLP and web data. The system has high potential to be extended in many powerful ways and serve as a general platform for search, OLAP (online analytical processing) and data mining on integrated text and structured data. After the system demo in the conference, the system will be put on the web for public access and evaluation.

With the development of positioning technology, movement data has become widely available nowadays. An important task in movement data analysis is to mine the relationships among moving objects based on their spatiotemporal interactions. Among all relationship types, attraction and avoidance are arguably the most natural ones. However, rather surprisingly, there is no existing method that addresses the problem of mining significant attraction and avoidance relationships in a well-defined and unified framework. In this paper, we propose a novel method to measure the significance value of relationship between any two objects by examining the background model of their movements via permutation test. Since permutation test is computationally expensive, two effective pruning strategies are developed to reduce the computation time. Furthermore, we show how the proposed method can be extended to efficiently answer the classic threshold query: given an object, retrieve all the objects in the database that have relationships, whose significance values are above certain threshold, with the query object. Empirical studies on both synthetic data and real movement data demonstrate the effectiveness and efficiency of our method.

In recent years, many new applications, such as sensor network monitoring and moving object search, show a growing amount of importance of uncertain data management and mining. In this paper, we study the problem of discovering threshold-based frequent closed item sets over probabilistic data. Frequent item set mining over probabilistic database has attracted much attention recently. However, existing solutions may lead an exponential number of results due to the downward closure property over probabilistic data. Moreover, it is hard to directly extend the successful experiences from mining exact data to a probabilistic environment due to the inherent uncertainty of data. Thus, in order to obtain a reasonable result set with small size, we study discovering frequent closed item sets over probabilistic data. We prove that even a sub-problem of this problem, computing the frequent closed probability of an item set, is #P-Hard. Therefore, we develop an efficient mining algorithm based on depth-first search strategy to obtain all probabilistic frequent closed item sets. To reduce the search space and avoid redundant computation, we further design several probabilistic pruning and bounding techniques. Finally, we verify the effectiveness and efficiency of the proposed methods through extensive experiments.

Query substitution is an important problem in information retrieval. Much work focuses on how to find substitutes for any given query. In this paper, we study how to efficiently process a keyword query whose substitutes are defined by a given taxonomy. This problem is challenging because each term in a query can have a large number of substitutes, and the original query can be rewritten into any of their combinations. We propose to build an additional index (besides inverted index) to efficiently process queries. For a query workload, we formulate an optimization problem which chooses the additional index structure, aiming at minimizing the query evaluation cost, under given index space constraints. We show the NP-hardness of the problem, and propose a pseudo-polynomial time algorithm using dynamic programming, as well as an 1 over 4(1-1/e)-approximation algorithm to solve the problem. Experimental results show that, with only 10% additional index space, our approach can greatly reduce the query evaluation cost.

We propose a novel system TEXplorer that integrates keyword-based object ranking with the aggregation and exploration power of OLAP in a text database with rich structured attributes available, e.g., a product review database. TEXplorer can be implemented within a multi-dimensional text database, where each row is associated with structural dimensions (attributes) and text data (e.g., a document). The system utilizes the text cube data model, where a cell aggregates a set of documents with matching values in a subset of dimensions. Cells in a text cube capture different levels of summarization of the documents, and can represent objects at different conceptual levels.

We are interested in scalable mining of a non-redundant set of significant recurrent rules from a sequence database. Recurrent rules have the form “whenever a series of precedent events occurs, eventually a series of consequent events occurs”. They are intuitive and characterize behaviors in many domains. An example is the domain of software specification, in which the rules capture a family of properties beneficial to program verification and bug detection. We enhance a past work on mining recurrent rules by Lo, Khoo, and Liu to perform mining more scalably. We propose a new set of pruning properties embedded in a new mining algorithm. Performance and case studies on benchmark synthetic and real datasets show that our approach is much more efficient and outperforms the state-of-the-art approach in mining recurrent rules by up to two orders of magnitude.

The main challenge of designing classification algorithms for sensor networks is the lack of labeled sensory data, due to the high cost of manual labeling in the harsh locales where a sensor network is normally deployed. Moreover, delivering all the sensory data to the sink would cost enormous energy. Therefore, although some classification techniques can deal with limited label information, they cannot be directly applied to sensor networks since they are designed for centralized databases. To address these challenges, we propose a hierarchical aggregate classification (HAC) protocol which can reduce the amount of data sent by each node while achieving accurate classification in the face of insufficient label information. In this protocol, each sensor node locally makes cluster analysis and forwards only its decision to the parent node. The decisions are aggregated along the tree, and eventually the global agreement is achieved at the sink node. In addition, to control the tradeoff between the communication energy and the classification accuracy, we design an extended version of HAC, called the constrained hierarchical aggregate classification (cHAC) protocol. cHAC can achieve more accurate classification results compared with HAC, at the cost of more energy consumption. The advantages of our schemes are demonstrated through the experiments on not only synthetic data but also a real testbed.

Data cubes play an essential role in data analysis and decision support. In a data cube, data from a fact table is aggregated on subsets of the table's dimensions, forming a collection of smaller tables called cuboids. When the fact table includes sensitive data such as salary or diagnosis, publishing even a subset of its cuboids may compromise individuals' privacy. In this paper, we address this problem using differential privacy (DP), which provides provable privacy guarantees for individuals by adding noise to query answers. We choose an initial subset of cuboids to compute directly from the fact table, injecting DP noise as usual; and then compute the remaining cuboids from the initial set. Given a fixed privacy guarantee, we show that it is NP-hard to choose the initial set of cuboids so that the maximal noise over all published cuboids is minimized, or so that the number of cuboids with noise below a given threshold (precise cuboids) is maximized. We provide an efficient procedure with running time polynomial in the number of cuboids to select the initial set of cuboids, such that the maximal noise in all published cuboids will be within a factor (ln|L| + 1)^2 of the optimal, where |L| is the number of cuboids to be published, or the number of precise cuboids will be within a factor (1 - 1/e) of the optimal. We also show how to enforce consistency in the published cuboids while simultaneously improving their utility (reducing error). In an empirical evaluation on real and synthetic data, we report the amounts of error of different publishing algorithms, and show that our approaches outperform baselines significantly.

Set intersection is a fundamental operation in information retrieval and database systems. This paper introduces linear space data structures to represent sets such that their intersection can be computed in a worst-case efficient way. In general, given k (preprocessed) sets, with totally n elements, we will show how to compute their intersection in expected time [EQUATION], where r is the intersection size and w is the number of bits in a machine-word. In addition, we introduce a very simple version of this algorithm that has weaker asymptotic guarantees but performs even better in practice; both algorithms outperform the state of the art techniques for both synthetic and real data sets and workloads.

Driven by the emerging network applications, querying and mining uncertain graphs has become increasingly important. In this paper, we investigate a fundamental problem concerning uncertain graphs, which we call the distance-constraint reachability (DCR) problem: Given two vertices s and t, what is the probability that the distance from s to t is less than or equal to a user-defined threshold d in the uncertain graph? Since this problem is #P-Complete, we focus on efficiently and accurately approximating DCR online. Our main results include two new estimators for the probabilistic reachability. One is a Horvitz-Thomson type estimator based on the unequal probabilistic sampling scheme, and the other is a novel recursive sampling estimator, which effectively combines a deterministic recursive computational procedure with a sampling process to boost the estimation accuracy. Both estimators can produce much smaller variance than the direct sampling estimator, which considers each trial to be either 1 or 0. We also present methods to make these estimators more computationally efficient. The comprehensive experiment evaluation on both real and synthetic datasets demonstrates the efficiency and accuracy of our new estimators.

The need for accurate SQL progress estimation in the context of decision support administration has led to a number of techniques proposed for this task. Unfortunately, no single one of these progress estimators behaves robustly across the variety of SQL queries encountered in practice, meaning that each technique performs poorly for a significant fraction of queries. This paper proposes a novel estimator selection framework that uses a statistical model to characterize the sets of conditions under which certain estimators outperform others, leading to a significant increase in estimation robustness. The generality of this framework also enables us to add a number of novel "special purpose" estimators which increase accuracy further. Most importantly, the resulting model generalizes well to queries very different from the ones used to train it. We validate our findings using a large number of industrial real-life and benchmark workloads.

With the maturity and wide availability of GPS, wireless, telecommunication, and Web technologies, massive amounts of object movement data have been collected from various moving object targets, such as animals, mobile devices, vehicles, and climate radars. Analyzing such data has deep implications in many applications, such as, ecological study, traffic control, mobile communication management, and climatological forecast. In this article, we focus our study on animal movement data analysis and examine advanced data mining methods for discovery of various animal movement patterns. In particular, we introduce a moving object data mining system, MoveMine, which integrates multiple data mining functions, including sophisticated pattern mining and trajectory analysis. In this system, two interesting moving object pattern mining functions are newly developed: (1) periodic behavior mining and (2) swarm pattern mining . For mining periodic behaviors, a reference location-based method is developed, which first detects the reference locations, discovers the periods in complex movements, and then finds periodic patterns by hierarchical clustering. For mining swarm patterns, an efficient method is developed to uncover flexible moving object clusters by relaxing the popularly-enforced collective movement constraints. In the MoveMine system, a set of commonly used moving object mining functions are built and a user-friendly interface is provided to facilitate interactive exploration of moving object data mining and flexible tuning of the mining constraints and parameters. MoveMine has been tested on multiple kinds of real datasets, especially for MoveBank applications and other moving object data analysis. The system will benefit scientists and other users to carry out versatile analysis tasks to analyze object movement regularities and anomalies. Moreover, it will benefit researchers to realize the importance and limitations of current techniques and promote future studies on moving object data mining. As expected, a mastery of animal movement patterns and trends will improve our understanding of the interactions between and the changes of the animal world and the ecosystem and therefore help ensure the sustainability of our ecosystem.

Previous studies on supporting free-form keyword queries over RDBMSs provide users with linked structures (e.g., a set of joined tuples) that are relevant to a given keyword query. Most of them focus on ranking individual tuples from one table or joins of multiple tables containing a set of keywords. In this paper, we study the problem of keyword search in a data cube with text-rich dimension(s) (so-called text cube). The text cube is built on a multidimensional text database, where each row is associated with some text data (a document) and other structural dimensions (attributes). A cell in the text cube aggregates a set of documents with matching attribute values in a subset of dimensions. We define a keyword-based query language and an IR-style relevance model for scoring/ranking cells in the text cube. Given a keyword query, our goal is to find the top-k most relevant cells. We propose four approaches: inverted-index one-scan, document sorted-scan, bottom-up dynamic programming, and search-space ordering. The search-space ordering algorithm explores only a small portion of the text cube for finding the top-k answers, and enables early termination. Extensive experimental studies are conducted to verify the effectiveness and efficiency of the proposed approaches.

On-Line Analytical Processing (OLAP) has shown great success in many industry applications, including sales, marketing, management, financial data analysis, etc. In this paper, we propose Visual Cube and multi-dimensional OLAP of image collections, such as web images indexed in search engines (e.g., Google and Bing), product images (e.g. Amazon) and photos shared on social networks (e.g., Facebook and Flickr). It provides online responses to user requests with summarized statistics of image information and handles rich semantics related to image visual features. A clustering structure measure is proposed to help users freely navigate and explore images. Efficient algorithms are developed to construct Visual Cube. In addition, we introduce the new issue of Cell Overlapping in data cube and present efficient solutions for Visual Cube computation and OLAP operations. Extensive experiments are conducted and the results show good performance of our algorithms.

Previous studies on supporting keyword queries in RDBMSs provide users with a ranked list of relevant linked structures (e.g. joined tuples) or individual tuples. In this paper, we aim to support keyword search in a data cube with text-rich dimension(s) (so-called text cube). Each document is associated with structural dimensions. A cell in the text cube aggregates a set of documents with matching dimension values on a subset of dimensions. Given a keyword query, our goal is to find the top-k most relevant cells in the text cube. We propose a relevance scoring model and efficient ranking algorithms. Experiments are conducted to verify their efficiency.

Periodicity is a frequently happening phenomenon for moving objects. Finding periodic behaviors is essential to understanding object movements. However, periodic behaviors could be complicated, involving multiple interleaving periods, partial time span, and spatiotemporal noises and outliers.

Recent improvements in positioning technology make massive moving object data widely available. One important analysis is to find the moving objects that travel together. Existing methods put a strong constraint in defining moving object cluster, that they require the moving objects to stick together for consecutive timestamps. Our key observation is that the moving objects in a cluster may actually diverge temporarily and congregate at certain timestamps. Motivated by this, we propose the concept of swarm which captures the moving objects that move within arbitrary shape of clusters for certain timestamps that are possibly non-consecutive. The goal of our paper is to find all discriminative swarms, namely closed swarm . While the search space for closed swarms is prohibitively huge, we design a method, ObjectGrowth, to efficiently retrieve the answer. In ObjectGrowth, two effective pruning strategies are proposed to greatly reduce the search space and a novel closure checking rule is developed to report closed swarms on-the-fly. Empirical studies on the real data as well as large synthetic data demonstrate the effectiveness and efficiency of our methods.

There is a huge wealth of sequence data available, for example, customer purchase histories, program execution traces, DNA, and protein sequences. Analyzing this wealth of data to mine important knowledge is certainly a worthwhile goal. In this paper, as a step forward to analyzing patterns in sequences, we introduce the problem of mining closed repetitive gapped subsequences and propose efficient solutions. Given a database of sequences where each sequence is an ordered list of events, the pattern we would like to mine is called repetitive gapped subsequence, which is a subsequence (possibly with gaps between two successive events within it) of some sequences in the database. We introduce the concept of repetitive support to measure how frequently a pattern repeats in the database. Different from the sequential pattern mining problem, repetitive support captures not only repetitions of a pattern in different sequences but also the repetitions within a sequence. Given a user-specified support threshold min_sup, we study finding the set of all patterns with repetitive support no less than min_sup. To obtain a compact yet complete result set and improve the efficiency, we also study finding closed patterns. Efficient mining algorithms to find the complete set of desired patterns are proposed based on the idea of instance growth. Our performance study on various datasets shows the efficiency of our approach. A case study is also performed to show the utility of our approach.

In this paper, we describe oCast, an energy-optimal multicast routing protocol for wireless sensor networks. The general minimum-energy multicast problem is NP-hard. Intermittent connectivity that results from duty-cycling further complicates the problem. Nevertheless, we present both a centralized and distributed algorithm that are provably optimal when the number of destinations is small. This model is motivated by scenarios where sensors report to a small number of base stations or where data needs to be replicated on a small number of other nodes. We further propose an extended version of oCast, called Delay Bounded oCast (DB-oCast), which can discover optimal multicast trees under a predefined delay bound. Finally, we demonstrate the advantages of our schemes through both theoretical analysis and simulations.

The spatial and temporal databases have been studied widely and intensively over years. In this paper, we study how to answer queries of finding the best departure time that minimizes the total travel time from a place to another, over a road network, where the traffic conditions dynamically change from time to time. We study a generalized form of this problem, called the time-dependent shortest-path problem. A time-dependent graph GT is a graph that has an edge-delay function, wi, j(t), associated with each edge (vi, vj), to be stored in a database. The edge-delay function wi, j(t) specifies how much time it takes to travel from node vi to node vj, if it departs from vi at time t. A user-specified query is to ask the minimum-travel-time path, from a source node, vs, to a destination node, ve, over the time-dependent graph, GT, with the best departure time to be selected from a time interval T. We denote this user query as LTT(vs, ve, T) over GT. The challenge of this problem is the added complexity due to the time dependency in the time-dependent graph. That is, edge delays are not constants, and can vary from time to time. In this paper, we propose a novel algorithm to find the minimum-travel-time path with the best departure time for a LTT(vs, ve, T) query over a large graph GT. Our approach outperforms existing algorithms in terms of both time complexity in theory and efficiency in practice. We will discuss the design of our algorithm, together with its correctness and complexity. We conducted extensive experimental studies over large graphs and will report our findings.

Due to rapid growth of the Internet technology and new scientific/technological advances, the number of applications that model data as graphs increases, because graphs have high expressive power to model complicated structures. The dominance of graphs in real-world applications asks for new graph data management so that users can access graph data effectively and efficiently. In this paper, we study a graph pattern matching problem over a large data graph. The problem is to find all patterns in a large data graph that match a user-given graph pattern. We propose a new two-step R-join (reachability join) algorithm with filter step and fetch step based on a cluster-based join-index with graph codes. We consider the filter step as an R-semijoin, and propose a new optimization approach by interleaving R-joins with R-semijoins. We conducted extensive performance studies, and confirm the efficiency of our proposed new approaches.

Since Jim Gray introduced the concept of rdquodata cuberdquo in 1997, data cube, associated with online analytical processing (OLAP), has become a driving engine in data warehouse industry. Because the boom of Internet has given rise to an ever increasing amount of text data associated with other multidimensional information, it is natural to propose a data cube model that integrates the power of traditional OLAP and IR techniques for text. In this paper, we propose a text-cube model on multidimensional text database and study effective OLAP over such data. Two kinds of hierarchies are distinguishable inside: dimensional hierarchy and term hierarchy. By incorporating these hierarchies, we conduct systematic studies on efficient text-cube implementation, OLAP execution and query processing. Our performance study shows the high promise of our methods.

Sequential pattern mining is an interesting data mining problem with many real-world applications. This problem has been studied extensively in static databases. However, in recent years, emerging applications have introduced a new form of data called data stream. In a data stream, new elements are generated continuously. This poses additional constraints on the methods used for mining such data: memory usage is restricted, the infinitely flowing original dataset cannot be scanned multiple times, and current results should be available on demand.This paper introduces two effective methods for mining sequential patterns from data streams: the SS-BE method and the SS-MB method. The proposed methods break the stream into batches and only process each batch once. The two methods use different pruning strategies that restrict the memory usage but can still guarantee that all true sequential patterns are output at the end of any batch. Both algorithms scale linearly in execution time as the number of sequences grows, making them effective methods for sequential pattern mining in data streams. The experimental results also show that our methods are very accurate in that only a small fraction of the patterns that are output are false positives. Even for these false positives, SS-BE guarantees that their true support is above a pre-defined threshold.

It is widely realized that the integration of database and information retrieval techniques will provide users with a wide range of high quality services. In this paper, we study processing an l-keyword query, p1, p2, ···, pl, against a relational database which can be modeled as a weighted graph, G(V, E). Here V is a set of nodes (tuples) and E is a set of edges representing foreign key references between tuples. Let Vi V be a set of nodes that contain the keyword pi. We study finding top-k minimum cost connected trees that contain at least one node in every subset Vi, and denote our problem as GST-k. When k = 1, it is known as a minimum cost group Steiner tree problem which is NP-Complete. We observe that the number of keywords, l, is small, and propose a novel parameterized solution, with l as a parameter, to find the optimal GST-1, in time complexity O(3ln + 2l((l + log n)n + m)), where n and m are the numbers of nodes and edges in graph G. Our solution can handle graphs with a large number of nodes. Our GST-1 solution can be easily extended to support GST-k, which outperforms the existing GST-k solutions over both weighted undirected/directed graphs. We conducted extensive experimental studies, and report our finding.