We develop accurate and efficient solutions for large-scale retrieval tasks where novel (zero-shot) items can arrive continuously at a rapid pace. Conventional Siamese-style approaches embed both queries and items through a small encoder and retrieve the items lying closest to the query. While this approach allows efficient addition and retrieval of novel items, the small encoder lacks sufficient capacity for the necessary world knowledge in complex retrieval tasks. The extreme classification approaches have addressed this by learning a separate classifier for each item observed in the training set which significantly increases the representation capacity of the model. Such classifiers outperform Siamese approaches on observed items, but cannot be trained for novel items due to data and latency constraints. To bridge these gaps, this paper develops: (1) A new algorithmic framework, EMMETT, which efficiently synthesizes classifiers on-the-fly for novel items, by relying on the readily available classifiers for observed items; (2) A new algorithm, IRENE, which is a simple and effective instance of EMMETT that is specifically suited for large-scale deployments, and (3) A new theoretical framework for analyzing the generalization performance in large-scale zero-shot retrieval which guides our algorithm and training related design decisions. Comprehensive experiments are conducted on a wide range of retrieval tasks which demonstrate that IRENE improves the zero-shot retrieval accuracy by up to 15% points in Recall@10 when added on top of leading encoders. Additionally, on an online A/B test in a large-scale ad retrieval task in a major search engine, IRENE improved the ad click-through rate by 4.2%. Lastly, we validate our design choices through extensive ablative experiments. The source code for IRENE is available at https://aka.ms/irene.
论文检索
输入标题、作者或关键词，从 642 篇学术成果中精准定位
Improving Multi-modal Recommender Systems by Denoising and Aligning Multi-modal Content and User Feedback
PDF ↗Multi-modal recommender systems (MRSs) are pivotal in diverse online web platforms and have garnered considerable attention in recent years. However, previous studies overlook the challenges of (1)noisy multi-modal content, (2) noisy user feedback, and (3) aligning multi-modal content and user feedback. To tackle these challenges, we propose Denoising and Aligning Multi-modal Recommender System (DA-MRS). To mitigate noise in multi-modal content, DA-MRS first constructs item-item graphs determined by consistent content similarity across modalities. To denoise user feedback, DA-MRS associates the probability of observed feedback with multi-modal content and devises a denoised BPR loss. Furthermore, DA-MRS implements Alignment guided by User preference to enhance task-specific item representation and Alignment guided by graded Item relations to provide finer-grained alignment. Extensive experiments verify that DA-MRS is a plug-and-play framework and achieves significant and consistent improvements across various datasets, backbone models, and noisy scenarios.
With the widespread adoption of electronic health records (EHR) data, deep learning techniques have been broadly utilized for various health prediction tasks. Nevertheless, the labeled data scarcity issue restricts the prediction power of these deep models. To enhance the generalization capability of deep learning models when faced with such situations, a common trend is to train generative adversarial networks (GANs) or diffusion models for data augmentation. However, due to limitations in sample size and potential label imbalance issues, these methods are prone to mode collapse problems. This results in the generation of new samples that fail to preserve the subtype structure within EHR data, thereby limiting their practicality in health prediction tasks that generally require detailed patient phenotyping. Aiming at the above problems, we propose a Prototype-based Mixup method, dubbed ProtoMix, which combines prior knowledge of intrinsic data features from subtype centroids (i.e., prototypes) to guide the synthesis of new samples. Specifically, ProtoMix employs a prototype-guided mixup training task to shift the decision boundary away from the subtypes. Then, ProtoMix optimizes the sampling weights in different areas of the data manifold via a prototype-guided mixup sampling strategy. Throughout the training process, ProtoMix dynamically expands the training distribution using an adaptive mixing coefficient computation method. Experimental evaluations on three real-world datasets demonstrate the efficacy of ProtoMix.
With the proliferation of mobile sensing techniques, huge amounts of time series data are generated and accumulated in various domains, fueling plenty of real-world applications. In this setting, time series anomaly detection is practically important. It endeavors to identify deviant samples from the normal sample distribution in time series. Existing approaches generally assume that all the time series is available at a central location. However, we are witnessing the decentralized collection of time series due to the deployment of various edge devices. To bridge the gap between the decentralized time series data and the centralized anomaly detection algorithms, we propose a Parameter-efficient Federated Anomaly Detection framework named PeFAD with the increasing privacy concerns. PeFAD for the first time employs the pre-trained language model (PLM) as the body of the client's local model, which can benefit from its cross-modality knowledge transfer capability. To reduce the communication overhead and local model adaptation cost, we propose a parameter-efficient federated training module such that clients only need to fine-tune small-scale parameters and transmit them to the server for update. PeFAD utilizes a novel anomaly-driven mask selection strategy to mitigate the impact of neglected anomalies during training. A knowledge distillation operation on a synthetic privacy-preserving dataset that is shared by all the clients is also proposed to address the data heterogeneity issue across clients. We conduct extensive evaluations on four real datasets, where PeFAD outperforms existing state-of-the-art baselines by up to 28.74%.
Multimodal electronic health record (EHR) data can offer a holistic assessment of a patient's health status, supporting various predictive healthcare tasks. Recently, several studies have embraced the multitask learning approach in the healthcare domain, exploiting the inherent correlations among clinical tasks to predict multiple outcomes simultaneously. However, existing methods necessitate samples to possess complete labels for all tasks, which places heavy demands on the data and restricts the flexibility of the model. Meanwhile, within a multitask framework with multimodal inputs, how to comprehensively consider the information disparity among modalities and among tasks still remains a challenging problem. To tackle these issues, a unified healthcare prediction model, also named by FlexCare, is proposed to flexibly accommodate incomplete multimodal inputs, promoting the adaption to multiple healthcare tasks. The proposed model breaks the conventional paradigm of parallel multitask prediction by decomposing it into a series of asynchronous single-task prediction. Specifically, a task-agnostic multimodal information extraction module is presented to capture decorrelated representations of diverse intra- and inter-modality patterns. Taking full account of the information disparities between different modalities and different tasks, we present a task-guided hierarchical multimodal fusion module that integrates the refined modality-level representations into an individual patient-level representation. Experimental results on multiple tasks from MIMIC-IV/MIMIC-CXR/MIMIC-NOTE datasets demonstrate the effectiveness of the proposed method. Additionally, further analysis underscores the feasibility and potential of employing such a multitask strategy in the healthcare domain. The source code is available at https://github.com/mhxu1998/FlexCare **REMOVE 2nd URL**://github.com/mhxu1998/FlexCare.
Active Voltage Control (AVC) on the Power Distribution Networks (PDNs) aims to stabilize the voltage levels to ensure efficient and reliable operation of power systems. With the increasing integration of distributed energy resources, recent efforts have explored employing multi-agent reinforcement learning (MARL) techniques to realize effective AVC. Existing methods mainly focus on the acquisition of short-term AVC strategies, i.e., only learning AVC within the short-term training trajectories of a singular diurnal cycle. However, due to the dynamic nature of load demands and renewable energy, the operation states of real-world PDNs may exhibit significant distribution shifts across varying timescales (e.g., daily and seasonal changes). This can render those short-term strategies suboptimal or even obsolete when performing continuous AVC over extended periods. In this paper, we propose a novel temporal prototype-aware learning method, abbreviated as TPA, to learn time-adaptive AVC under short-term training trajectories. At the heart of TPA are two complementary components, namely multi-scale dynamic encoder and temporal prototype-aware policy, that can be readily incorporated into various MARL methods. The former component integrates a stacked transformer network to learn underlying temporal dependencies at different timescales of the PDNs, while the latter implements a learnable prototype matching mechanism to construct a dedicated AVC policy that can dynamically adapt to the evolving operation states. Experimental results on the AVC benchmark with different PDN sizes demonstrate that the proposed TPA surpasses the state-of-the-art counterparts not only in terms of control performance but also by offering model transferability. Our code is available at https://github.com/Canyizl/TPA-for-AVC.
Long-term time series forecasting (LTSF) aims to predict future values for a long time based on historical data. The period term is an essential component of the time series, which is complex yet important for LTSF. Although existing studies have achieved promising results, they still have limitations in modeling dynamic complicated periods. Most studies only focus on static periods with fixed time steps, while very few studies attempt to capture dynamic periods in the time domain. In this paper, we dissect the original time series in time and frequency domains and empirically find that changes in periods are more easily captured and quantified in the frequency domain. Based on this observation, we propose to explore dynamic period features using rotation in the frequency domain. To this end, we develop the frequency-based rotation network (FRNet), a novel LTSF method to effectively capture the features of the dynamic complicated periods. FRNet decomposes the original time series into period and trend components. Based on the complex-valued linear networks, it leverages a period frequency rotation module to predict the period component and a patch frequency rotation module to predict the trend component, respectively. Extensive experiments on seven real-world datasets consistently demonstrate the superiority of FRNet over various state-of-the-art methods. The source code is available at https://github.com/SiriZhang45/FRNet.
Vertical federated learning (VFL) is an emerging collaborative machine learning paradigm to facilitate the utilization of private features distributed across multiple parties. During the inference process of VFL, the involved parties need to upload their local embeddings to be aggregated for the final prediction. Despite its remarkable performances, the inference process of the current VFL system is vulnerable to the strategic behavior of involved parties, as they could easily change the uploaded local embeddings to exert direct influences on the prediction result. In a representative case study of federated recommendation, we find the allocation of display opportunities to be severely disrupted due to the parties' preferences in display content. In order to elicit the true local embeddings for VFL system, we propose a distribution-based penalty mechanism to detect and penalize the strategic behaviors in collaborative inference. As the key motivation of our design, we theoretically prove the power of constraining the distribution of uploaded embeddings in preventing the dishonest parties from achieving higher utility. Our mechanism leverages statistical two-sample tests to distinguish whether the distribution of uploaded embeddings is reasonable, and penalize the dishonest party through deactivating her uploaded embeddings. The resulted mechanism could be shown to admit truth-telling to converge to a Bayesian Nash equilibrium asymptotically under mild conditions. The experimental results further demonstrate the effectiveness of the proposed mechanism to reduce the dishonest utility increase of strategic behaviors and promote the truthful uploading of local embeddings in inferences.
Graph Neural Architecture Search (GNAS) has achieved superior performance on various graph-structured tasks. However, existing GNAS studies overlook the applications of GNAS in resource-constrained scenarios. This paper proposes to design a joint graph data and architecture mechanism, which identifies important sub-architectures via the valuable graph data. To search for optimal lightweight Graph Neural Networks (GNNs), we propose a Lightweight Graph Neural Architecture Search with Graph SparsIfication and Network Pruning (GASSIP) method. In particular, GASSIP comprises an operation-pruned architecture search module to enable efficient lightweight GNN search. Meanwhile, we design a novel curriculum graph data sparsification module with an architecture-aware edge-removing difficulty measurement to help select optimal sub-architectures. With the aid of two differentiable masks, we iteratively optimize these two modules to efficiently search for the optimal lightweight architecture. Extensive experiments on five benchmarks demonstrate the effectiveness of GASSIP. Particularly, our method achieves on-par or even higher node classification performance with half or fewer model parameters of searched GNNs and a sparser graph.
Make Your Home Safe: Time-aware Unsupervised User Behavior Anomaly Detection in Smart Homes via Loss-guided Mask
PDF ↗Smart homes, powered by the Internet of Things, offer great convenience but also pose security concerns due to abnormal behaviors, such as improper operations of users and potential attacks from malicious attackers. Several behavior modeling methods have been proposed to identify abnormal behaviors and mitigate potential risks. However, their performance often falls short because they do not effectively learn less frequent behaviors, consider temporal context, or account for the impact of noise in human behaviors. In this paper, we propose SmartGuard, an autoencoder-based unsupervised user behavior anomaly detection framework. First, we design a Loss-guided Dynamic Mask Strategy (LDMS) to encourage the model to learn less frequent behaviors, which are often overlooked during learning. Second, we propose a Three-level Time-aware Position Embedding (TTPE) to incorporate temporal information into positional embedding to detect temporal context anomaly. Third, we propose a Noise-aware Weighted Reconstruction Loss (NWRL) that assigns different weights for routine behaviors and noise behaviors to mitigate the interference of noise behaviors during inference. Comprehensive experiments demonstrate that SmartGuard consistently outperforms state-of-the-art baselines and also offers highly interpretable results.
How to Avoid Jumping to Conclusions: Measuring the Robustness of Outstanding Facts in Knowledge Graphs
PDF ↗An outstanding fact (OF) is a striking claim by which some entities stand out from their peers on some attribute. OFs serve data journalism, fact checking, and recommendation. However, one could jump to conclusions by selecting truthful OFs while intentionally or inadvertently ignoring lateral contexts and data that render them less striking. This jumping conclusion bias from unstable OFs may disorient the public, including voters and consumers, raising concerns about fairness and transparency in political and business competition. It is thus ethically imperative for several stakeholders to measure the robustness of OFs with respect to lateral contexts and data. Unfortunately, a capacity for such inspection of OFs mined from knowledge graphs (KGs) is missing. In this paper, we propose a methodology that inspects the robustness of OFs in KGs by perturbation analysis. We define (1) entity perturbation, which detects outlying contexts by perturbing context entities in the OF; and (2) data perturbation, which considers plausible data that render an OF less striking. We compute the expected strikingness scores of OFs over perturbation relevance distributions and assess an OF as robust if its measured strikingness does not deviate significantly from the expected. We devise a suite of exact and sampling algorithms for perturbation analysis on large KGs. Extensive experiments reveal that our methodology accurately and efficiently detects frail OFs generated by existing mining approaches on KGs. We also show the effectiveness of our approaches through case and user studies.
Understanding urban regional characteristics is pivotal in driving critical insights for urban planning and management. We have witnessed the successful application of pre-trained Foundation Models (FMs) in generating universal representations for various downstream tasks. However, applying this principle to the geospatial domain remains challenging, primarily due to the difficulty of gathering extensive data for developing a dedicated urban foundation model. Though there have been some attempts to empower the existing FMs with urban data, most of them focus on single-modality FMs without considering the multi-modality nature of urban region understanding tasks. To address this gap, we introduce ReFound - a novel framework for Re-training a Foundation model for urban region understanding, harnessing the strengths of both language and visual FMs. In this framework, we first invent a Mixture-of-Geospatial-Expert (MoGE) Transformer, to effectively integrate the embedding of multi-source geospatial data. Building on this, ReFound is enhanced by jointly distilling knowledge from language, visual, and visual-language FMs respectively, thus augmenting its generalization capabilities. Meanwhile, we design a masked geospatial data modeling approach alongside a cross-modal spatial alignment mechanism, to enhance the spatial knowledge of ReFound derived from geospatial data. Extensive experiments conducted on six real-world datasets over three urban region understanding tasks demonstrate the superior performance of our framework.
Motif-Consistent Counterfactuals with Adversarial Refinement for Graph-level Anomaly Detection
PDF ↗Graph-level anomaly detection is significant in diverse domains. To improve detection performance, counterfactual graphs have been exploited to benefit the generalization capacity by learning causal relations. Most existing studies directly introduce perturbations (e.g., flipping edges) to generate counterfactual graphs, which are prone to alter the semantics of generated examples and make them off the data manifold, resulting in sub-optimal performance. To address these issues, we propose a novel approach, Motif-consistent Counterfactuals with Adversarial Refinement (MotifCAR), for graph-level anomaly detection. The model combines the motif of one graph, the core subgraph containing the identification (category) information, and the contextual subgraph (non-motif) of another graph to produce a raw counterfactual graph. However, the produced raw graph might be distorted and cannot satisfy the important counterfactual properties: Realism, Validity, Proximity and Sparsity. Towards that, we present a Generative Adversarial Network (GAN)-based graph optimizer to refine the raw counterfactual graphs. It adopts the discriminator to guide the generator to generate graphs close to realistic data, i.e., meet the property Realism. Further, we design the motif consistency to force the motif of the generated graphs to be consistent with the realistic graphs, meeting the property Validity. Also, we devise the contextual loss and connection loss to control the contextual subgraph and the newly added links to meet the properties Proximity and Sparsity. As a result, the model can generate high-quality counterfactual graphs. Experiments demonstrate the superiority of MotifCAR.
Performative Debias with Fair-exposure Optimization Driven by Strategic Agents in Recommender Systems
PDF ↗Data bias, e.g., popularity impairs the dynamics of two-sided markets within recommender systems. This overshadows the less visible but potentially intriguing long-tail items that could capture user interest. Despite the abundance of research surrounding this issue, it still poses challenges and remains a hot topic in academic circles. Along this line, in this paper, we developed a re-ranking approach in dynamic settings with fair-exposure optimization driven by strategic agents. Designed for the producer side, the execution of agents assumes content creators can modify item features based on strategic incentives to maximize their exposure. This iterative process entails an end-to-end optimization, employing differentiable ranking operators that simultaneously target accuracy and fairness. Joint objectives ensure the performance of recommendations while enhancing the visibility of tail items. We also leveraged the performativity nature of predictions to illustrate how strategic learning influences content creators to shift towards fairness efficiently, thereby incentivizing features of tail items. Through comprehensive experiments on both public and industrial datasets, we have substantiated the effectiveness and dominance of the proposed method especially on unveiling the potential of tail items.
In this paper, we study cascading failures in power grids through the lens of information diffusion models. Similar to the spread of rumors or influence in an online social network, it has been observed that failures (outages) in a power grid can spread contagiously, driven by viral spread mechanisms. We employ a stochastic diffusion model that is Markovian (memoryless) and local (the activation of one node, i.e., transmission line, can only be caused by its neighbors). Our model integrates viral diffusion principles with physics-based concepts, by correlating the diffusion weights (contagion probabilities between transmission lines) with the hyperparametric Information Cascades (IC) model. We show that this diffusion model can be learned from traces of cascading failures, enabling accurate modeling and prediction of failure propagation. This approach facilitates actionable information through well-understood and efficient graph analysis methods and graph diffusion simulations. Furthermore, by leveraging the hyperparametric model, we can predict diffusion and mitigate the risks of cascading failures even in unseen grid configurations, whereas existing methods falter due to a lack of training data. Extensive experiments based on a benchmark power grid and simulations therein show that our approach effectively captures the failure diffusion phenomena and guides decisions to strengthen the grid, reducing the risk of large-scale cascading failures. Additionally, we characterize our model's sample complexity, improving upon the existing bound.
FLea: Addressing Data Scarcity and Label Skew in Federated Learning via Privacy-preserving Feature Augmentation
PDF ↗Federated Learning (FL) enables model development by leveraging data distributed across numerous edge devices without transferring local data to a central server. However, existing FL methods still face challenges when dealing with scarce and label-skewed data across devices, resulting in local model overfitting and drift, consequently hindering the performance of the global model. In response to these challenges, we propose a pioneering framework called FLea, incorporating the following key components: i) A global feature buffer that stores activation-target pairs shared from multiple clients to support local training. This design mitigates local model drift caused by the absence of certain classes; ii) A feature augmentation approach based on local and global activation mix-ups for local training. This strategy enlarges the training samples, thereby reducing the risk of local overfitting; iii) An obfuscation method to minimize the correlation between intermediate activations and the source data, enhancing the privacy of shared features. To verify the superiority of FLea, we conduct extensive experiments using a wide range of data modalities, simulating different levels of local data scarcity and label skew. The results demonstrate that FLea consistently outperforms state-of-the-art FL counterparts (among 13 of the experimented 18 settings, the improvement is over 5%) while concurrently mitigating the privacy vulnerabilities associated with shared features. Code is available at https://github.com/XTxiatong/FLea.git.
Kemeny's constant for random walks on a graph is defined as the mean hitting time from one node to another selected randomly according to the stationary distribution. It has found numerous applications and attracted considerable research interest. However, exact computation of Kemeny's constant requires matrix inversion, which scales poorly for large networks with millions of nodes. Existing approximation algorithms either leverage properties exclusive to undirected graphs or involve inefficient simulation, leaving room for further optimization. To address these limitations for directed graphs, we propose two novel approximation algorithms for estimating Kemeny's constant on directed graphs with theoretical error guarantees. Extensive numerical experiments on real-world networks validate the superiority of our algorithms over baseline methods in terms of efficiency and accuracy.
Predicting information diffusion is a fundamental task in online social networks (OSNs). Recent studies mainly focus on the popularity prediction of specific content but ignore the correlation between multiple pieces of information. The topic is often used to correlate such information and can correspond to multi-source information. The popularity of a topic relies not only on information diffusion time but also on users' followership. Current solutions concentrate on hard time partition, lacking versatility. Meanwhile, the hop-based sampling adopted in state-of-the-art (SOTA) methods encounters redundant user followership. Moreover, many SOTA methods are not designed with good modularity and lack evaluation for each functional module and enlightening discussion. This paper presents a novel extensible framework, coined as HIF, for effective popularity prediction in OSNs with four original contributions. First, HIF adopts a soft partition of users and time intervals to better learn users' behavioral preferences over time. Second, HIF utilizes weighted sampling to optimize the construction of heterogeneous graphs and reduce redundancy. Furthermore, HIF supports multi-task collaborative optimization to improve its learning capability. Finally, as an extensible framework, HIF provides generic module slots to combine different submodules (e.g., RNNs, Transformer encoders). Experiments show that HIF significantly improves performance and interpretability compared to SOTAs.
The fraudulent insurance claim is critical for the insurance industry. Insurance companies or agency platforms aim to confidently estimate the fraud risk of claims by gathering data from various sources. Although more data sources can improve the estimation accuracy, they inevitably lead to increased costs. Therefore, a great challenge of fraud risk verification lies in well balancing these two aspects. To this end, this paper proposes a framework named cost-efficient fraud risk optimization with submodularity (CEROS) to optimize the process of fraud risk verification. CEROS efficiently allocates investigation resources across multiple information sources, balancing the trade-off between accuracy and cost. CEROS consists of two parts that we propose: a submodular set-wise classification model called SSCM to estimate the submodular objective function, and a primal-dual algorithm with segmentation point called PDA-SP to solve the objective function. Specifically, SSCM models the fraud probability associated with multiple information sources and ensures the properties of submodularity of fraud risk without making independence assumption. The submodularity in SSCM enables PDA-SP to significantly speed up dual optimization. Theoretically, we disclose that when PDA-SP optimizes this dual optimization problem, the process is monotonicity. Finally, the trade-off coefficients output by PDA-SP that balance accuracy and cost in fraud risk verification are applied to online insurance claim decision-making. We conduct experiments on offline trials and online A/B tests in two business areas at Alipay: healthcare insurance recommendation and claim verification. The extensive results indicate that, compared with other methods, CEROS achieves acceleration of 66.9% in convergence speed and meanwhile 18.8% in cost reduction. Currently, CEROS has been successfully deployed in Alipay.
The graph-based recommendation has achieved great success in recent years. However, most existing graph-based recommendations focus on capturing user preference based on positive edges/feedback, while ignoring negative edges/feedback (e.g., dislike, low rating) that widely exist in real-world recommender systems. How to utilize negative feedback in graph-based recommendations still remains underexplored. In this study, we first conducted a comprehensive experimental analysis and found that (1) existing graph neural networks are not well-suited for modeling negative feedback, which acts as a high-frequency signal in a user-item graph. (2) The graph-based recommendation suffers from the representation degeneration problem. Based on the two observations, we propose a novel model that models positive and negative feedback from a frequency filter perspective called Dual-frequency Graph Neural Network for Sign-aware Recommendation (DFGNN). Specifically, in DFGNN, the designed dual-frequency graph filter (DGF) captures both low-frequency and high-frequency signals that contain positive and negative feedback. Furthermore, the proposed signed graph regularization is applied to maintain the user/item embedding uniform in the embedding space to alleviate the representation degeneration problem. Additionally, we conduct extensive experiments on real-world datasets and demonstrate the effectiveness of the proposed model. Codes of our model will be released upon acceptance.