In the evolving e-commerce field, recommendation systems crucially shape user experience and engagement. The rise of Consumer-to-Consumer (C2C) recommendation systems, noted for their flexibility and ease of access for customer vendors, marks a significant trend. However, the academic focus remains largely on Business-to-Consumer (B2C) models, leaving a gap filled by the limited C2C recommendation datasets that lack in item attributes, user diversity, and scale. The intricacy of C2C recommendation systems is further accentuated by the dual roles users assume as both sellers and buyers, introducing a spectrum of less uniform and varied inputs. Addressing this, we introduce MerRec, the first large-scale dataset specifically for C2C recommendations, sourced from the Mercari e-commerce platform, covering millions of users and products over 6 months in 2023. MerRec not only includes standard features such as user_id, item_id, and session_id, but also unique elements like timestamped action types, product taxonomy, and textual product attributes, offering a comprehensive dataset for research. This dataset, extensively evaluated across three recommendation tasks, establishes a new benchmark for the development of advanced recommendation algorithms in real-world scenarios, bridging the gap between academia and industry and propelling the study of C2C recommendations. Experiment code (https://github.com/mercari/mercari-ml-merrec-pub-us) and dataset (https://huggingface.co/datasets/mercari-us/merrec) are released.
论文检索
输入标题、作者或关键词，从 844 篇学术成果中精准定位
In this work, we address the challenge of reconstructing the complete 12-lead ECG signal from its incomplete parts. We focus on two main scenarios: (i) reconstructing missing signal segments within an ECG lead and (ii) recovering entire leads from signal in another unique lead. Two emerging clinical applications emphasize the relevance of our work. The first is the increasing need to digitize paper-stored ECGs for utilization in AI-based applications, often limited to digital 12 lead 10s ECGs. The second is the widespread use of wearable devices that record ECGs but typically capture only one or a few leads. In both cases, a non-negligible amount of information is lost or not recorded. Our approach aims to recover this missing signal. We propose ECGrecover, a U-Net neural network model trained on a novel composite objective function to address the reconstruction problem. This function incorporates both spatial and temporal features of the ECG by combining the distance in amplitude and sycnhronization through time between the reconstructed and the real digital signals. We used real-life ECG datasets and through comprehensive assessments compared ECGrecover with three state-of-the-art methods based on generative adversarial networks (EKGAN, Pix2Pix) as well as the CopyPaste strategy. The results demonstrated that ECGrecover consistently outperformed state-of-the-art methods in standard distortion metrics as well as in preserving critical ECG characteristics, particularly the P, QRS, and T wave coordinates.
A Deep Subgrouping Framework for Precision Drug Repurposing via Emulating Clinical Trials on Real-world Patient Data
PDF ↗Drug repurposing identifies new therapeutic uses for existing drugs, reducing the time and costs compared to traditional de novo drug discovery. Most existing drug repurposing studies using real-world patient data often treat the entire population as homogeneous, ignoring the heterogeneity of treatment responses across patient subgroups. This approach may overlook promising drugs that benefit specific subgroups but lack notable treatment effects across the entire population, potentially limiting the number of repurposable candidates identified. To address this, we introduce STEDR, a novel drug repurposing framework that integrates subgroup analysis with treatment effect estimation. Our approach first identifies repurposing candidates by emulating multiple clinical trials on real-world patient data and then characterizes patient subgroups by learning subgroup-specific treatment effects. We deploy STEDR to Alzheimer's Disease (AD), a condition with few approved drugs and known heterogeneity in treatment responses. We emulate trials for over one thousand medications on a large-scale real-world database covering over 8 million patients, identifying 14 drug candidates with beneficial effects to AD in characterized subgroups. Experiments demonstrate STEDR superior capability in identifying repurposing candidates compared to existing approaches. Additionally, our method can characterize clinically relevant patient subgroups associated with important AD-related risk factors, paving the way for precision drug repurposing.
Traffic Signal Control (TSC) is a crucial component in urban traffic management, aiming to optimize road network efficiency and reduce congestion. Traditional TSC methods, primarily based on transportation engineering and reinforcement learning (RL), often struggle with generalization abilities across varied traffic scenarios and lack interpretability. This paper presents LLMLight, a novel framework employing Large Language Models (LLMs) as decision-making agents for TSC. Specifically, the framework begins by instructing the LLM with a knowledgeable prompt detailing real-time traffic conditions. Leveraging the advanced generalization capabilities of LLMs, LLMLight engages a reasoning and decision-making process akin to human intuition for effective traffic control. Moreover, we build LightGPT, a specialized backbone LLM tailored for TSC tasks. By learning nuanced traffic patterns and control strategies, LightGPT enhances the LLMLight framework cost-effectively. Extensive experiments conducted on ten real-world and synthetic datasets, along with evaluations by fifteen human experts, demonstrate the exceptional effectiveness, generalization ability, and interpretability of LLMLight with LightGPT, outperforming nine baseline methods and ten advanced LLMs. Our project is available at https://github.com/usail-hkust/LLMTSCS.
Telegram is a widely adopted instant messaging platform. It has become worldwide popular because of its emphasis on privacy and its social network features such as channels-virtual rooms in which only the admins can post and broadcast messages to all the subscribers. Channels are used to deliver live updates (e.g., weather alerts) and content to a large audience (e.g., COVID-19 announcements) but unfortunately also to disseminate radical ideologies and coordinate attacks such as the Capitol Hill riot. This paper introduces the TGDataset, the most extensive publicly available collection of Telegram channels, comprising 120,979 channels and over 400 million messages. We outline the data collection process and provide a comprehensive overview of the data set. Using language detection, we identify the predominant languages within the dataset. We then focus on English channels, employing topic modeling to analyze the subjects they cover. Finally, we discuss some use cases in which our dataset can be instrumental in understanding the Telegram ecosystem and studying the diffusion of questionable news. Alongside the raw dataset, we release the scripts used in our analysis, as well as a list of channels associated with a novel conspiracy theory known as Sabmyk.
In the rapidly progressing field of generative models, the development of efficient and high-fidelity text-to-image diffusion systems represents a significant frontier. This study introduces YaART, a novel production-grade text-to-image cascaded diffusion model aligned to human preferences using Reinforcement Learning from Human Feedback (RLHF). During the development of YaART, we especially focus on the choices of the model and training dataset sizes, the aspects that were not systematically investigated for text-to-image cascaded diffusion models before. In particular, we comprehensively analyze how these choices affect both the efficiency of the training process and the quality of the generated images, which are highly important in practice. Furthermore, we demonstrate that models trained on smaller datasets of higher-quality images can successfully compete with those trained on larger datasets, establishing a more efficient scenario of diffusion models training. From the quality perspective, YaART is consistently preferred by users over many existing state-of-the-art models. The proposed system is integrated in five commercial products. YaART is also distributed for B2C clients via mobile and web applications and for B2B clients via API.
Recent advances in generative artificial intelligence have revolutionized information retrieval and content generation, opening up new opportunities for the e-commerce industry. Alignment learning between small models and parallel corpora cannot meet current needs. The success of ChatGPT demonstrates that large models need to first establish a fundamental understanding, and then utilize high-quality corpora for generation. Having a large model foundation is indispensable. In this paper, we establish a fundamental 10B multimodal model foundation for multimodal generation tasks and propose a scene-based alignment learning approach called conditional sample supervised fine-tuning for downstream generation tasks. Meanwhile, diffusion models are known to be vulnerable to outliers in training data. To address this, we utilize an alternative diffusion loss function that preserves the high quality of generated data like the original squared L2 loss while being robust to outliers.In practical test sets, the multimodal foundation fully demonstrates its alignment and comprehension abilities for graphic and textual content. Additionally, conditional fine-tuning and the design of the loss function significantly enhance the quality of generated content. The quality rate of images has increased by 34.3 percentage points, and prompt control has improved by 19.8 percentage points. The application of our framework in Baidu Search Ads has led to significant revenue growth. For instance, ads with generated image creatives have achieved a 29% higher click-through rate (CTR), resulting in a daily consumption of 3 million yuan.
Survey data are common and invaluable in social science research for understanding population processes and supporting policymaking and planning. Depending on the nature and scale, survey data sharing comes with privacy risks, and data collectors and agencies are constrained by disclosure permissions, limiting usage across research groups and institutes. Previous methods for synthetic data generation and deidentification may not entirely prevent information disclosures, or they may sacrifice data quality and granularity. Using a large-scale national voter file at both national and state levels, this paper introduces an end-to-end pipeline to streamline synthetic data generation and evaluation for survey researchers. This study selects four generative approaches based on different statistical assumptions: the regression-based Synthpop, the generative deep learning-based CTGAN and TVAE, and the large language model-based REaLDTabFormer, and compares them to the baseline synthetic minority oversampling technique (SMOTE). We consider three key dimensions of evaluation (utility, fidelity, and privacy) to highlight the strengths and weaknesses of each approach, and systematically evaluate across various datasets and training sizes. The results reveal that Synthpop is optimized for general utility (i.e., fidelity), while TVAE excels in downstream applications (i.e., target-specific utility) but compromises on general utility and potentially risks data overfitting. REaLDTabFormer demonstrates a balanced performance in both general and target-specific utility, whereas CTGAN offers the best privacy protection. We recommend that future researchers select a generative method by considering the trade-offs between performance across various evaluation dimensions, training size, data type, and computational infrastructure.
Real-Time Bidding (RTB) is a trading mechanism that allocates advertising (ad) requests through online auctions. Participants in these auctions typically include an ad exchange (AdX) and several demand-side platforms (DSPs). When an RTB auction begins, the AdX first establishes the reserve price set by publishers as the starting bid, after which the DSPs bid to compete for potential ad impressions. The reserve price strategy is crucial to the ad revenue of publishers; however, due to the strategic and dynamic bidding behavior of DSPs, optimizing the reserve price presents a significant challenge. In this work, we report a novel adaptive reserve price strategy based on reinforcement learning (RL). In our scheme, value bucket identification is leveraged to estimate the intrinsic values of ad inventories. Following this estimation, specialized reward functions are utilized to generate informative reward signals for RL models. Furthermore, we study the issue of risk management on the publisher side and develop a risk-aware instantiation to model risk tendency, considering both empirical expert knowledge and real-time trading conditions. Extensive experiments using real-world datasets collected from operational environments have demonstrated the effectiveness of the proposed method.
Pre-training large Tabular Language Models (TaLMs) on tabular data has shown effectiveness for table understanding tasks. However, training proprietary large TaLMs on a company's private databases requires substantial computational resources. This paper presents an efficient multi-expert TaLM architecture and training method tailored for multi-domain databases and modest infrastructure. This architecture leverages a divide-and-conquer pretraining approach and a sparsely activated fine-tuning paradigm to reduce computation. Using this architecture, we pre-train and fine-tune a TaLM with 10 billion parameters on a banking database under simple computational infrastructures. We apply our TaLM to support various important banking applications, including risk assessment, information prediction, and profit assessment. Compared with previous baselines, our model achieves +29.3% in precision@0.6% on risk assessment and +16.5% in accuracy on information prediction, showing great effectiveness and profitability of our model. This model is successfully deployed in WeBank and now supports many real business scenarios.
Phishing scams on the Ethereum network have become a serious threat, especially with the influx of new users into the cryptocurrency market. Current detection methods are mainly focused on long-term consistent transaction patterns with smooth temporal dynamics. However, these methods often struggle to differentiate between phishing and non-phishing users, whose behaviours may appear deceptively similar. Additionally, they face challenges such as network sparsity and data leakage, leading to significant performance limitations. To address these issues, we introduce TEMPER, a novel sequential learning framework designed to jointly capture the subtle distinctions between long- and short-term user behaviours and their correlations to provide more comprehensive insights. TEMPER effectively generates distinguishable user embeddings, enabling the accurate identification of phishing users. Unlike previous approaches, TEMPER mitigates data leakage through a novel sequential transaction sampling algorithm and addresses network sparsity with short-term temporal learning. Through extensive experimentation on three real-world Ethereum datasets, TEMPER demonstrates its efficacy by achieving a 3-4% improvement in the F1-Score compared to existing baseline models, representing a significant advancement in Ethereum phishing user detection.
Today's top advertisers typically manage hundreds of campaigns simultaneously and consistently launch new ones throughout the year. A crucial challenge for marketing managers is determining the optimal allocation of limited budgets across various ad lines in each campaign to maximize cumulative returns, especially given the huge uncertainty in return outcomes. In this paper, we propose to formulate budget allocation as a multi-task combinatorial bandit problem and introduce a novel online budget allocation system. The proposed system: i) integrates a Bayesian hierarchical model to intelligently utilize the metadata of campaigns and ad lines and budget size, ensuring efficient information sharing; ii) provides the flexibility to incorporate diverse modeling techniques such as Linear Regression, Gaussian Processes, and Neural Networks, catering to diverse environmental complexities; and iii) employs the Thompson sampling technique to strike a balance between exploration and exploitation. Through extensive empirical evaluations with both synthetic data and real-world data from Amazon campaigns, our system demonstrates robustness and adaptability, effectively maximizing the overall cumulative returns. A Python implementation of the proposed procedure is available at https://anonymous.4open.science/r/MCMAB.
Online experiments in internet systems, also known as A/B tests, are used for a wide range of system tuning problems, such as optimizing recommender system ranking policies and learning adaptive streaming controllers. Decision-makers generally wish to optimize for long-term treatment effects of the system changes, which often requires running experiments for a long time as short-term measurements can be misleading due to non-stationarity in treatment effects over time. The sequential experimentation strategies---which typically involve several iterations---can be prohibitively long in such cases. We describe a novel approach that combines fast experiments (e.g., biased experiments run only for a few hours or days) and/or offline proxies (e.g., off-policy evaluation) with long-running, slow experiments to perform sequential, Bayesian optimization over large action spaces in a short amount of time.
Making ideal decisions as a product leader in a web-facing company is incredibly challenging. Beyond navigating the ambiguity of customer satisfaction and achieving business goals, leaders must also ensure their products and services remain relevant, desirable, and profitable. Data and experimentation are crucial for testing product hypotheses and informing decisions. Online controlled experiments, such as A/B testing, can provide highly reliable data to support decisions. However, these experiments can be time-consuming and costly, particularly when assessing impacts on key business metrics like retention or long-term value. Offline experimentation allows for rapid iteration and testing but often lacks the same level of confidence and clarity regarding business metrics impact. To address this, we introduce a novel, lightweight, and flexible approach called scenario analysis. This method aims to support product leaders' decisions by using user data and estimates of business metrics. While it cannot fully replace online experiments, it offers valuable insights into trade-offs involved in growth or consumption shifts, estimates trends in long-term outcomes like retention, and can generate hypotheses about relationships between metrics at scale. We implemented scenario analysis in a tool named ForTune. We conducted experiments with this tool using a publicly available dataset and reported the results of experiments carried out by Spotify, a large audio streaming service, using ForTune in production. In both cases, the tool reasonably predicted the outcomes of controlled experiments, provided that features were carefully chosen. We demonstrate how this method was used to make strategic decisions regarding the impact of prioritizing one type of content over another at Spotify.
In online advertising, advertisers commonly utilize auto-bidding services to bid for impression opportunities. A typical objective of the auto-bidder is to optimize the advertiser's cumulative value of winning impressions within specified budget constraints. However, such a problem is challenging due to the complex bidding environment faced by diverse advertisers. To address this challenge, we introduce ABPlanner, a few-shot adaptable budget planner designed to improve budget-constrained auto-bidding. ABPlanner is based on a hierarchical bidding framework that decomposes the bidding process into shorter, manageable stages. Within this framework, ABPlanner allocates the budget across all stages, allowing a low-level auto-bidder to bids based on the budget allocation plan. The adaptability of ABPlanner is achieved through a sequential decision-making approach, inspired by in-context reinforcement learning. For each advertiser, ABPlanner adjusts the budget allocation plan episode by episode, using data from previous episodes as prompt for current decisions. This enables ABPlanner to quickly adapt to different advertisers with few-shot data, providing a sample-efficient solution. Extensive simulation experiments and real-world A/B testing validate the effectiveness of ABPlanner, demonstrating its capability to enhance the cumulative value achieved by auto-bidders.
In this paper, we present a novel transformer architecture tailored for learning robust power system state representations, which strives to optimize power dispatch for the power flow adjustment across different transmission sections. Specifically, our proposed approach, named Powerformer, develops a dedicated section-adaptive attention mechanism, separating itself from the self-attention employed in conventional transformers. This mechanism effectively integrates power system states with transmission section information, which facilitates the development of robust state representations. Furthermore, by considering the graph topology of power system and the electrical attributes of bus nodes, we introduce two customized strategies to further enhance the expressiveness: graph neural network propagation and multi-factor attention mechanism. Extensive evaluations are conducted on three power system scenarios, including the IEEE 118-bus system, a realistic China 300-bus system, and a large-scale European system with 9241 buses, where Powerformer demonstrates its superior performance over several popular baseline methods. The code is available at: https://github.com/Cra2yDavid/Powerformer
Modern recommendation systems can be broadly divided into two key stages: the ranking stage, where the system predicts various user engagements (e.g., click-through rate, like rate, follow rate, watch time), and the value model stage, which aggregates these predictive scores through a function (e.g., a linear combination defined by a weight vector) to measure the value of each content by a single numerical score. Both stages play roughly equally important roles in real industrial systems; however, how to optimize the model weights for the second stage still lacks systematic study. This paper focuses on optimizing the second stage through auto-tuning technology. Although general auto-tuning systems and solutions - both from established production practices and open-source solutions - can address this problem, they typically require weeks or even months to identify a feasible solution. Such prolonged tuning processes are unacceptable in production environments for recommendation systems, as suboptimal value models can severely degrade user experience. An effective auto-tuning solution is required to identify a viable model within 2-3 days, rather than the extended timelines typically associated with existing approaches. In this paper, we introduce a practical auto-tuning system named HyperZero that addresses these time constraints while effectively solving the unique challenges inherent in modern recommendation systems. Moreover, this framework has the potential to be expanded to broader tuning tasks within recommendation systems.
In this paper, we analyze the feature-based knowledge distillation for recommendation from the frequency perspective. By defining knowledge as different frequency components of the features, we theoretically demonstrate that regular feature-based knowledge distillation is equivalent to equally minimizing losses on all knowledge and further analyze how this equal loss weight allocation method leads to important knowledge being overlooked. In light of this, we propose to emphasize important knowledge by redistributing knowledge weights. Furthermore, we propose FreqD, a lightweight knowledge reweighting method, to avoid the computational cost of calculating losses on each knowledge. Extensive experiments demonstrate that FreqD consistently and significantly outperforms state-of-the-art knowledge distillation methods for recommender systems. Our code is available at https://github.com/woriazzc/KDs.
Online advertising has become a core revenue driver for internet industry, with ad auctions playing a crucial role in ensuring platform revenue and advertiser incentives. Classical auction mechanisms, such as GSP, rely on the independent CTR assumption and fail to account for the interplay among the displayed items, also called as externalities in economics. Recent advancements in learning-based auctions enable the encoding of high-dimensional contextual features. However, existing methods are limited by the ''prediction-before-allocation'' design paradigm, which models set-level externalities within candidate ads and fails to consider the context of the final allocation, leading to suboptimal results. In this work, we introduce Contextual Generative Auction (CGA), a novel framework that incorporates permutation-level externalities in multi-slot ad auctions. Built on the structure of our theoretically derived optimal auction, CGA decouples the optimization of allocation and payment. We construct an autoregressive generative model for allocation, and reformulate incentive compatibility (IC) constraint into minimizing ex-post regret that supports gradient computation, enabling end-to-end learning of the optimal payment rule. Extensive offline and online experiments demonstrate that CGA significantly enhances platform revenue and CTR compared to existing methods, and effectively approximates the optimal auction with nearly maximal revenue and minimal regret.
The advent of the "pre-train, prompt'' paradigm has recently extended its generalization ability and data efficiency to graph representation learning, following its achievements in Natural Language Processing (NLP). Initial graph prompt tuning approaches tailored specialized prompting functions for Graph Neural Network (GNN) models pre-trained with specific strategies, such as edge prediction, thus limiting their applicability. In contrast, another pioneering line of research has explored universal prompting via adding prompts to the input graph's feature space, thereby removing the reliance on specific pre-training strategies. However, the necessity to add feature prompts to all nodes remains an open question. Motivated by findings from prompt tuning research in the NLP domain, which suggest that highly capable pre-trained models need less conditioning signal to achieve desired behaviors, we advocate for strategically incorporating necessary and lightweight feature prompts to certain graph nodes to enhance downstream task performance. This introduces a combinatorial optimization problem, requiring a policy to decide 1) which nodes to prompt and 2) what specific feature prompts to attach. We then address the problem by framing the prompt incorporation process as a sequential decision-making problem and propose our method, RELIEF, which employs Reinforcement Learning (RL) to optimize it. At each step, the RL agent selects a node (discrete action) and determines the prompt content (continuous action), aiming to maximize cumulative performance gain. Extensive experiments on graph and node-level tasks with various pre-training strategies in few-shot scenarios demonstrate that our RELIEF outperforms fine-tuning and other prompt-based approaches in classification performance and data efficiency. The code is available at https://github.com/JasonZhujp/RELIEF.