World Wide Web最新文献

Deep neural networks (DNN) exhibit powerful feature extraction capabilities, making them highly advantageous in numerous tasks. DNN-based techniques have become widely adopted in the field of speaker recognition. However, imperceptible adversarial perturbations can severely disrupt the decisions made by DNNs. In addition, researchers identified universal adversarial perturbations that can efficiently and significantly attack deep neural networks. In this paper, we propose an algorithm for conducting effective universal adversarial attacks by investigating the dominant features in the speaker recognition task. Through experiments in various scenarios, we find that our perturbations are not only more effective and undetectable but also exhibit a certain degree of transferablity across different datasets and models.

Humanity needs to deliberate effectively at scale about highly complex and contentious problems. Current online deliberation tools—such as email, chatrooms, and forums—are however plagued by levels of discussion toxicity that deeply undercut the willingness and ability of the participants to engage in thoughtful, meaningful, deliberations. This has led many organizations to either shut down their forums or invest in expensive, frequently unreliable, and ethically fraught moderation of people's contributions in their forums. This paper includes a comprehensive review on online toxicity, and describes how a structured deliberation process can substantially reduce toxicity compared to current approaches. The key underlying insight is that unstructured conversations create, especially at scale, an “attention wars” dynamic wherein people are often incented to resort to extremified language in order to get visibility for their postings. A structured deliberation process wherein people collaboratively create a compact organized collection of answers and arguments removes this underlying incentive, and results, in our evaluation, in a 50% reduction of high-toxicity posts.

Graph Neural Networks (GNNs) have achieved remarkable success in diverse real-world applications. Traditional GNNs are designed based on homophily, which leads to poor performance under heterophily scenarios. Most current solutions deal with heterophily mainly by modeling the heterophily edges as data noises or high-frequency signals, treating all heterophilic edges as being of the same semantic. Consequently, they ignore the rich semantic information of these edges in heterophily graphs. To overcome this critic problem, we propose a novel GNN model based on relation vector translation named as Variational Relation Vector Graph Neural Network (VR-GNN). VR-GNN models relation generation and graph aggregation into an end-to-end model based on a variational inference framework. To be specific, the encoder utilizes the structure, feature and label to generate a fine-grained relation vector for each edge, which aims to infer its implicit semantic information. The decoder incorporates the generated relation vectors into the message-passing framework for deriving better node representations. We conduct extensive experiments on eight real-world datasets with different homophily-heterophily properties to verify model effectiveness. Extensive experimental results show that VR-GNN gains consistent and significant improvements against existing strong GNN methods under heterophily and competitive performance under homophily.

In tackling Tuberculosis (TB), a critical global health challenge, the integration of Foundation Models (FMs) into diagnostic processes represents a significant advance. FMs, with their extensive pre-training on diverse datasets, hold the promise of transforming TB diagnosis by leveraging their deep understanding and analytical capabilities. However, the application of these models in healthcare is complicated by the need to protect patient privacy, particularly when dealing with sensitive TB data from various medical centers. Our novel approach, FedARC, addresses this issue through personalized federated learning (PFL), enabling the use of private data without direct access. FedARC innovatively navigates data heterogeneity and privacy concerns by employing adaptive regularization and model-contrastive learning. This method not only aligns each center’s objective function with the global loss’s stationary point but also enhances model generalization across disparate data sources. Comprehensive evaluations on five publicly available chest X-ray image datasets demonstrate that foundation models profoundly influence outcomes, with our proposed method significantly surpassing contemporary methodologies in various scenarios.

Recommending medications with electronic health records (EHRs) is a challenging task for data-driven clinical decision support systems. Most existing models learnt representations for medical concepts based on EHRs and make recommendations with the learnt representations. However, most medications appear in EHR datasets for limited times (the frequency distribution of medications follows power law distribution), resulting in insufficient learning of their representations of the medications. Medical ontologies are the hierarchical classification systems for medical terms where similar terms will be in the same class on a certain level. In this paper, we propose OntoMedRec, the logically-pretrained and model-agnostic medical Ontology Encoders for Medication Recommendation that addresses data sparsity problem with medical ontologies. We conduct comprehensive experiments on real-world EHR datasets to evaluate the effectiveness of OntoMedRec by integrating it into various existing downstream medication recommendation models. The result shows the integration of OntoMedRec improves the performance of various models in both the entire EHR datasets and the admissions with few-shot medications. We provide the GitHub repository for the source code. (https://github.com/WaicongTam/OntoMedRec)

Group recommendation aims to suggest desired items for a group of users. Existing methods can achieve inspiring results in predicting the group preferences in data-rich groups. However, they could be ineffective in supporting cold-start groups due to their sparsity interactions, which prevents the model from understanding their intent. Although cold-start groups can be alleviated by meta-learning, we cannot apply it by using the same initialization for all groups due to their varying preferences. To tackle this problem, this paper proposes a memory-augmented meta-optimized model for group recommendation, namely GroupMO. Specifically, we adopt a clustering method to assemble the groups with similar profiles into the same cluster and design a representative group profile memory to guide the preliminary initialization of group embedding network for each group by utilizing those clusters. Besides, we also design a group shared preference memory to guide the prediction network initialization at a more refined granularity level for different groups, so that the shared knowledge can be better transferred to groups with similar preferences. Moreover, we incorporate those two memories to optimize the meta-learning process. Finally, extensive experiments on two real-world datasets demonstrate the superiority of our model.