Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks245 from public data
- Adversarial Attack and Defense on Graph Data: A Survey409
This review is intended to provide an overall landscape of more than 100 papers on adversarial attack and defense strategies for graph data, and establish a unified formulation encompassing most graph adversarial learning models.
- Heterogeneous Network Representation Learning: A Unified Framework With Survey and Benchmark324
This work aims to provide a unified framework to deeply summarize and evaluate existing research on heterogeneous network embedding (HNE), and provides a generic paradigm for the systematic categorization and analysis over the merits of various existing HNE algorithms.
- Bridging Collaborative Filtering and Semi-Supervised Learning316
This work proposes to devise a general and principled SSL (semi-supervised learning) framework, to alleviate data scarcity via smoothing among neighboring users and POIs, and treat various context by regularizing user preference based on context graphs.
- Domain Specialization as the Key to Make Large Language Models Disruptive: A Comprehensive Survey311
A systematic taxonomy is proposed that categorizes the LLM domain-specialization techniques based on the accessibility to LLMs and summarizes the framework for all the subcategories as well as their relations and differences to each other.
- BrainGB: A Benchmark for Brain Network Analysis With Graph Neural Networks244
This work presents BrainGB, a benchmark for brain network analysis with GNNs, a standardizes the process by summarizing brain network construction pipelines for both functional and structural neuroimaging modalities and modularizing the implementation of GNN designs.
- FBNETGEN: Task-aware GNN-based fMRI Analysis via Functional Brain Network Generation161
FBNETGEN is developed, a task-aware and interpretable fMRI analysis framework via deep brain network generation which learns to transform raw time-series features into task-oriented brain networks and provides unique interpretations by highlighting prediction-related brain regions.
- GAD-NR: Graph Anomaly Detection via Neighborhood Reconstruction154
The proposed \proj is a new variant of GAE that incorporates neighborhood reconstruction for graph anomaly detection, and excels at detecting all three types of anomalies across the datasets, demonstrating its comprehensive anomaly detection capabilities.
- GuardAgent: Safeguard LLM Agents by a Guard Agent via Knowledge-Enabled Reasoning146
This paper proposes GuardAgent, the first LLM agent as a guardrail to other LLM agents, and proposes two novel benchmarks: an EICU-AC benchmark for assessing privacy-related access control for healthcare agents and a Mind2Web-SC benchmark for safety evaluation for web agents.
- EHRAgent: Code Empowers Large Language Models for Few-shot Complex Tabular Reasoning on Electronic Health Records127
This work proposes EHRAgent, a large language model (LLM) agent empowered with accumulative domain knowledge and robust coding capability that enables autonomous code generation and execution to facilitate clinicians in directly interacting with EHRs using natural language.
- On Positional and Structural Node Features for Graph Neural Networks on Non-attributed Graphs116
This paper points out the two types of artificial node features, i.e., positional and structural node features and provides insights on why each of them is more appropriate for certain tasks, thereby leading to a practical guideline on the choices between different artificial nodes features for GNNs on non-attributed graphs.
- Conditional Structure Generation through Graph Variational Generative Adversarial Nets114
This work forms the novel problem of conditional structure generation, and proposes a novel unified model of graph variational generative adversarial nets (CondGen) to handle the intrinsic challenges of flexible context-structure conditioning and permutation-invariant generation.
- 103
- I Know You'll Be Back93
This paper develops ClusChurn, a systematic two-step framework for interpretable new user clustering and churn prediction, based on the intuition that properuser clustering can help understand and predict user churn.
- Understanding Structural Vulnerability in Graph Convolutional Networks78
This work theoretically and empirically demonstrate that structural adversarial examples can be attributed to the non-robust aggregation scheme (i.e., the weighted mean) of GCNs, and takes advantage of the breakdown point which can quantitatively measure the robustness of aggregation schemes.
- 4SDrug: Symptom-based Set-to-set Small and Safe Drug Recommendation77
A novel framework of Symptom-based Set-to-set Small and Safe drug recommendation (4SDrug) is proposed, which outperforms all baselines in most effectiveness measures, while yielding the smallest sets of recommended drugs and 26.83% DDI rate reduction from the ground-truth data.
- When Do GNNs Work: Understanding and Improving Neighborhood Aggregation75
This paper proposes novel metrics that quantitatively measure these two circumstances and integrate them into an Adaptive-layer module and shows that allowing for node-specific aggregation degrees have significant advantage over current GNNs.
- 74
- A review on knowledge graphs for healthcare: Resources, applications, and promises73
HKGs have emerged as a powerful tool for structuring medical knowledge, with broad applications across biomedical research, clinical decision-making, and public health, highlighting the potential of combining knowledge graphs with advanced machine learning models for healthcare transformation.
- 69
- SAIS: Supervising and Augmenting Intermediate Steps for Document-Level Relation Extraction66
This work proposes to explicitly teach the model to capture relevant contexts and entity types by supervising and augmenting intermediate steps (SAIS) for RE by assessing model uncertainty, which boosts the performance via evidence-based data augmentation and ensemble inference while reducing the computational cost.
- Structure-Enhanced Heterogeneous Graph Contrastive Learning63
A novel method to generate multiple semantic views for HGs and advocate the explicit use of structure embedding, which enriches the model with local structural patterns of the underlying HGs, so as to better mine true and hard negatives for GCL.
- 60
- Heterogeneous Network Representation Learning: Survey, Benchmark, Evaluation, and Beyond60
This work aims to provide a unified framework to deeply summarize and evaluate existing research on heterogeneous network embedding (HNE), and provides a generic paradigm for the systematic categorization and analysis over the merits of various existing HNE algorithms.
- Joint Embedding of Structural and Functional Brain Networks with Graph Neural Networks for Mental Illness Diagnosis57
This work develops a novel multiview GNN model which takes advantage of the message passing scheme by propagating messages based on degree statistics and brain region connectivities and employs contrastive learning for multimodal fusion.
- 56
- Secure Deep Graph Generation with Link Differential Privacy54
This paper leverage the differential privacy (DP) framework to formulate and enforce rigorous privacy constraints on deep graph generation models, with a focus on edge-DP to guarantee individual link privacy.
- Zero-Shot Scene Graph Relation Prediction Through Commonsense Knowledge Integration54
This work proposes CommOnsense-integrAted sCenegrapHrElation pRediction (COACHER), a framework to integrate commonsense knowledge for SGG, especially for zero-shot relation prediction, and develops novel graph mining pipelines to model the neighborhoods and paths around entities in an external Commonsense knowledge graph.
- 52
- Graph-Aware Language Model Pre-Training on a Large Graph Corpus Can Help Multiple Graph Applications50
A framework of graph-aware language model pre-training (GaLM) on a large graph corpus, which incorporates large language models and graph neural networks, and a variety of fine-tuning methods on downstream applications is proposed.
- BrainNet49
A novel model (BrainNet) is proposed that jointly learns the dynamic diffusion graphs and models the brain wave diffusion patterns and outperforms several latest state-of-the-art baselines derived from time-series analysis.
- Relation Learning on Social Networks with Multi-Modal Graph Edge Variational Autoencoders49
This work aims to develop a principled and principled frame- work that can pro le user relations as edge semantics in social networks by integrating multi-modal signals in the presence of noisy and incomplete data.
- MultiSage48
A contextualized GCN engine is presented by modeling the multipartite networks of target nodes and their intermediatecontext nodes that specify the contexts of their interactions to achieve interaction contextualization by treating neighboring target nodes based on intermediate context nodes.
- Graph Entropy Guided Node Embedding Dimension Selection for Graph Neural Networks43
This paper revisits NEDS from the perspective of minimum entropy principle and proposes a novel Minimum Graph Entropy (MinGE) algorithm, which considers both feature entropy and structure entropy on graphs, which are carefully designed according to the characteristics of the rich information in them.
- Meta-Graph Based HIN Spectral Embedding: Methods, Analyses, and Insights43
This work proposes an embedding compression method of autoencoder with l2,1-loss, which finds the most informative meta-graphs and embeddings in an end-to-end unsupervised manner and suggests a unified workflow to close the gap between the authors' meta- graph assessment and combination methods.
- Unveiling Implicit Deceptive Patterns in Multi-Modal Fake News via Neuro-Symbolic Reasoning42
A novel Neuro-Symbolic Latent Model is proposed, called NSLM, that not only derives accurate judgments on the veracity of news but also uncovers the implicit deceptive patterns as explanations that reveal how fake news is fabricated.
- MetaCare++: Meta-Learning with Hierarchical Subtyping for Cold-Start Diagnosis Prediction in Healthcare Data42
A novel Meta-learning framework for cold-start diagnosis prediction in healthCare data (MetaCare) is proposed, and a hierarchical patient subtyping mechanism is designed to bridge the modeling of both infrequent patients and rare diseases.
- LLMs-based Few-Shot Disease Predictions using EHR: A Novel Approach Combining Predictive Agent Reasoning and Critical Agent Instruction40
This study investigates the feasibility of applying Large Language Models to convert structured patient visit data into natural language narratives and proposes a novel approach that utilizes LLM agents with different roles: a predictor agent that makes predictions and generates reasoning processes and a critic agent that analyzes incorrect predictions and provides guidance for improving the reasoning of the predictor agent.
- Contrastive Intra- and Inter-Modality Generation for Enhancing Incomplete Multimedia Recommendation40
A novel framework of Contrastive Intra- and Inter-Modality Generation (CI2MG) for enhancing incomplete multimedia recommendation is proposed and an enhanced multimedia recommendation module is designed to alleviate the influences driven by task-irrelevant noise.
- Time-Series Event Prediction with Evolutionary State Graph39
This paper presents evolutionary state graph, a dynamic graph structure designed to systematically represent the evolving relations (edges) among states (nodes) along time, and proposes a novel graph neural network model, Evolutionary State Graph Network (EvoNet), to encode the evolutionary stategraph for accurate and interpretable time-series event prediction.
- Beyond Efficiency: A Systematic Survey of Resource-Efficient Large Language Models38
This survey aims to systematically address resource efficiency challenges in Large Language Models by reviewing a broad spectrum of techniques designed to enhance the resource efficiency of LLMs, and introduces a nuanced categorization of resource efficiency techniques by their specific resource types.
- DPAR: Decoupled Graph Neural Networks with Node-Level Differential Privacy38
This paper proposes a D ecoupled GNN with Differentially Differentiated PageRank (DPAR) for training GNNs with an enhanced privacy-utility tradeoff and empirically demonstrates better utilities of DPAR with the same level of node DP compared with state-of-the-art methods.
- Data-Efficient Brain Connectome Analysis via Multi-Task Meta-Learning38
This paper proposes to meta-train the model on datasets of large sample sizes and transfer the knowledge to small datasets, and also explores brain-network-oriented designs, including atlas transformation and adaptive task reweighing.
- When to Pre-Train Graph Neural Networks? From Data Generation Perspective!36
This paper introduces a generic framework W2PGNN to answer the crucial question of when to pre-train (.e., in what situations could the authors take advantage of graph pre-training) before performing effortful pre- training or fine-tuning, and provides a theoretically sound solution for the first application and extensive empirical justifications for the latter two applications.
- Federated Node Classification over Graphs with Latent Link-type Heterogeneity36
A framework FedLit is proposed that can dynamically detect the latent link-types during FL via an EM-based clustering algorithm and differentiate the message-passing through different types of links via multiple convolution channels and demonstrates both superior performance and rational behaviors of the proposed techniques.
- Learning and Updating Node Embedding on Dynamic Heterogeneous Information Network36
A method named Dynamic Heterogeneous Information Network Embedding (DyHINE) is proposed, which can update embeddings when the network evolves and is compared with state-of-the-art methods on the task of temporal link prediction.
- Lightweight Visual Question Answering using Scene Graphs36
A new model called Conditional Enhanced Graph ATtention network (CE-GAT) is designed to encode pairs of visual and semantic scene graphs with both node and edge features, which is seamlessly integrated with a textual question encoder to generate answers through question-graph conditioning.
- Preliminary analysis of the impact of lab results on large language model generated differential diagnoses35
Lab tests, including liver function, metabolic/toxicology panels, and serology, were generally interpreted correctly by LLMs for DDx, and incorporating lab data enhances accuracy by up to 30% across models.
- Similarity Modeling on Heterogeneous Networks via Automatic Path Discovery35
This work combines continuous reinforcement learning and deep content embedding into a novel semi-supervised joint learning framework to automatically discover useful paths for pairs of nodes under both structural and content information in content-rich heterogeneous networks.
- Graph ODEs and Beyond: A Comprehensive Survey on Integrating Differential Equations with Graph Neural Networks34
This survey aims to provide a comprehensive overview of the burgeoning research at the intersection of GNNs and DEs, categorize existing methods, discuss their underlying principles, and highlight their applications across domains such as molecular modeling, traffic prediction, and epidemic spreading.
- 34
- Bi-directional Joint Inference for User Links and Attributes on Large Social Graphs34
Inspired by the well-known homophily phenomenon which found the two properties strongly interleaving, it is proposed to jointly learn them by leveraging their data redundancy and mutual reinforcement by iteratively addressing smoothness on the graph in two directions.
- Graph-Oriented Instruction Tuning of Large Language Models for Generic Graph Mining31
A novel framework that seamlessly integrates the strengths of GNNs and LLMs into one foundation model for graph mining across tasks and datasets, and designs a graph-aware instruction tuning strategy to facilitate mutual enhancement across multiple tasks and datasets while preventing catastrophic forgetting of LLMs’ generative abilities.
- Learning Task-Aware Effective Brain Connectivity for fMRI Analysis with Graph Neural Networks (Extended Abstract)31
The proposed TBDS, an end-to-end framework based on Task-aware Brain connectivity DAG (short for Directed Acyclic Graph) Structure generation for fMRI analysis, adopts a DAG learning approach to transform the raw time-series into task-aware brain connectivities.
- Dynamic Brain Transformer with Multi-Level Attention for Functional Brain Network Analysis30
A novel methodology, Dynamic bRAin Transformer (DART), which combines static and dynamic brain networks for more effective and nuanced brain function analysis, and shows a promising direction in neuroimaging studies, contributing to the comprehensive understanding of brain organization and the role of neural circuits.
- Certifiable Robustness to Discrete Adversarial Perturbations for Factorization Machines30
A robust training procedure is presented whose core idea is to increase the number of instances that are certifiably robust of factorization machines with respect to the discrete perturbation on input features.
- PTGB: Pre-Train Graph Neural Networks for Brain Network Analysis29
PTGB is proposed, a GNN pre-training framework that captures intrinsic brain network structures, regardless of clinical outcomes, and is easily adaptable to various downstream tasks that facilitates knowledge transfer across datasets with different ROI systems.
- 29
- Collab-RAG: Boosting Retrieval-Augmented Generation for Complex Question Answering via White-Box and Black-Box LLM Collaboration28
Collab-RAG, a collaborative training framework that leverages mutual enhancement between a white-box small language model (SLM) and a blackbox large language model (LLM) for RAG, substantially outperforms existing black-box-only and SLM fine-tuning baselines by 1.8%-14.2% on average.
- Correlation-Aware Graph Convolutional Networks for Multi-Label Node Classification27
A Correlation-aware Graph Convolutional Network (CorGCN) for multi-label node classification is proposed that can learn a graph that contains rich label-correlated information for each label and employs a Correlation-Enhanced Graph Convolution to model the relationships between labels during message passing to further bolster the classification process.
- A Survey on Unifying Large Language Models and Knowledge Graphs for Biomedicine and Healthcare26
The goal in this survey is to systematically investigate and summarize recent studies on the unification of LLMs and KGs, towards fully utilizing the value of complex data, unleashing the power of generative AI, and expediting next-generation AI for biomedicine and healthcare applications.
- DietAI24 as a framework for comprehensive nutrition estimation using multimodal large language models26
DietAI24 demonstrates that integrating MLLMs with RAG and standardized nutrition databases can substantially improve dietary assessment accuracy while enabling comprehensive nutrient analysis, and offers a scalable solution for nutrition research and clinical applications.
- PromptLink: Leveraging Large Language Models for Cross-Source Biomedical Concept Linking26
PromptLink is a generic framework without reliance on additional prior knowledge, context, or training data, making it well-suited for concept linking across various types of data sources.
- Better with Less: A Data-Active Perspective on Pre-Training Graph Neural Networks26
The curse of big data phenomenon in graph pre-training is identified: more training data do not necessarily lead to better downstream performance, and a better-with-less framework is proposed: fewer, but carefully chosen data are fed into a GNN model to enhance pre- training.
- 26
- Towards a general-purpose foundation model for fMRI analysis25
NeuroSTORM (Neuroimaging Foundation Model with Spatial-Temporal Optimized Representation Modeling) is introduced that learns generalizable representations directly from 4D fMRI volumes and enables efficient transfer to diverse downstream applications and is suggested to become a standardized foundation model for reproducible and transferable fMRI analysis.
- Causal Discovery through Synergizing Large Language Model and Data-Driven Reasoning25
A novel causal modeling framework, LLM-CD, which integrates the metadata-based reasoning capabilities of large language models (LLMs) with the data-driven modeling abilities of TCDA for causal discovery, and enhances causal discovery through an iterative process.
- Weakly-Supervised Scientific Document Classification via Retrieval-Augmented Multi-Stage Training25
WanDeR is proposed, which leverages dense retrieval to perform matching in the embedding space to capture the semantics of label names and further design the label name expansion module to enrich its representations.
- 24
- How Can Graph Neural Networks Help Document Retrieval: A Case Study on CORD19 with Concept Map Generation24
An empirical study on a large-scale multi-discipline dataset CORD-19 shows that instead of the complex structure-oriented GNNs such as GINs and GATs, the proposed semantics-oriented graph functions achieve better and more stable performance based on the BM25 retrieved candidates.
- 23
- 23
- Neural Embedding Propagation on Heterogeneous Networks22
This work proposes neural embedding propagation (NEP), which leverages distributed embeddings to represent objects and dynamically composed modular networks to model their complex interactions, and comprehensively demonstrates the effectiveness, efficiency, and robustness of NEP compared with state-of-the-art network embedding and SSL algorithms.
- Empowering Graph Neural Network-Based Computational Drug Repositioning with Large Language Model-Inferred Knowledge Representation21
The utility of LLM-inferred knowledge representation in drug repositioning with its generality and applicability in other biomedical relation prediction tasks is showcased.
- Fusion Modeling: Combining Clinical and Imaging Data to Advance Cardiac Care21
The current status of cardiovascular risk prediction and fusion modeling is outlined, and an overview of different artificial intelligence approaches to automatically extract data from images and electronic medical records for this purpose is given.
- 21
- 21
- 20
- Transformer-Based Hierarchical Clustering for Brain Network Analysis19
This work proposes a novel interpretable transformer-based model for joint hierarchical cluster identification and brain network classification and shows that with the help of hierarchical clustering, the model achieves increased accuracy and reduced runtime complexity while providing plausible insight into the functional organization of brain regions.
- A simple but tough-to-beat baseline for fMRI time-series classification18
The findings suggest that the sequential information hidden in the temporal order may be far less important for the accurate fMRI classification than the stand-alone pieces of information scattered across the frames of the time series.
- TACCO: Task-guided Co-clustering of Clinical Concepts and Patient Visits for Disease Subtyping based on EHR Data17
TACCO is introduced, a novel framework that jointly discovers clusters of clinical concepts and patient visits based on a hypergraph modeling of EHR data and develops a novel self-supervised co-clustering framework that can be guided by the risk prediction task of specific diseases.
- HiPrompt: Few-Shot Biomedical Knowledge Fusion via Hierarchy-Oriented Prompting17
HiPrompt is proposed, a supervision-efficient knowledge fusion framework that elicits the few-shot reasoning ability of large language models through hierarchy-oriented prompts and demonstrates the effectiveness of HiPrompt on the collected KG-Hi-BKF benchmark datasets.
- 17
- 17
- Multi-Facet Recommender Networks with Spherical Optimization17
This work proposes the novel framework of Multi-fAcet Recommender networks with Spherical optimization (MARS), and designs a cross-facet similarity measurement to project users and items into multiple metric spaces for fine-grained representation learning, and devise a spherical optimization strategy to enhance the effectiveness and robustness of the multi- facet recommendation framework.
- Deep Generation of Heterogeneous Networks17
This work proposes a novel framework for heterogeneous graph generation (HGEN) that jointly captures the semantic, structural, and global distributions of heterogeneous graphs and develops a novel heterogeneous walk generator that hierarchically generates meta-paths and their path instances.
- Place Deduplication with Embeddings17
This work takes the anonymous place graph from Facebook as an example to systematically study the problem of place deduplication, and arrives at a systematic two-step data-driven pipeline based on place embedding with multiple novel techniques that works significantly better than the state-of-the-art.
- Rethinking functional brain connectome analysis: do graph deep learning models Help16
This study re-examines graph deep learning versus classical machine learning models based on four large-scale neuroimaging studies and proposes a hybrid model combining a linear model with a graph attention network through dual pathways, achieving robust predictions and enhanced interpretability by revealing both localized and global neural connectivity patterns.
- Towards a general-purpose foundation model for functional MRI analysis15
The Neuroimaging Foundation Model with Spatial-Temporal Optimized and Representation Modelling (NeuroSTORM), which learns generalizable representations directly from four-dimensional fMRI volumes and enables efficient transfer to diverse downstream applications, is introduced.
- Scaling Agentic Reinforcement Learning for Tool-Integrated Reasoning in VLMs15
VQA benchmarks show that VISTA-R1-8B outperforms state-of-the-art baselines with similar sizes by 9.51%-18.72%, demonstrating VISTA-Gym as an effective training ground to unlock the tool-integrated reasoning capabilities for VLMs.
- 15
- BrainODE: Dynamic Brain Signal Analysis via Graph-Aided Neural Ordinary Differential Equations15
A novel model called BrainODE is proposed to achieve continuous modeling of dynamic brain signals using Ordinary Differential Equations (ODE) and effectively reconstructs brain signals at any time point, mitigating the aforementioned three data challenges of brain signals altogether.
- 15
- Spatiotemporal Activity Modeling Under Data Scarcity: A Graph-Regularized Cross-Modal Embedding Approach15
BranchNet, a spatiotemporal activity model that transfers knowledge from external sources for alleviating data scarcity, is proposed, which adopts a graph-regularized cross-modal embedding framework and is capable of effectively transferring knowledge from External sources to learn better spatiotmporal activity models and outperforming strong baseline methods.
- Subgraph Federated Learning for Local Generalization14
The proposed FedLoG model generates global synthetic data by condensing the reliable information from each class representation and its structural information across clients, which enhances the generalization capabilities of local models, enabling them to handle unseen data effectively.
- MedAgentGym: A Scalable Agentic Training Environment for Code-Centric Reasoning in Biomedical Data Science14
Med-Copilot achieves performance gains of +43 .
- R-Mixup: Riemannian Mixup for Biological Networks14
R-MIXUP is proposed, a Mixup-based data augmentation technique that suits the symmetric positive definite (SPD) property of adjacency matrices from biological networks with optimized training efficiency and demonstrates the effectiveness with five real-world biological network datasets.
- GCN for HIN via Implicit Utilization of Attention and Meta-Paths14
A novel neural network method via implicitly utilizing attention and meta-paths is proposed, which can relieve the severe overfitting brought by the current over-parameterized attention mechanisms on HIN.
- 13
- Piecing It All Together: Verifying Multi-Hop Multimodal Claims13
It is shown that MMCV is challenging even for the latest state-of-the-art multimodal large language models, especially as the number of reasoning hops increases, and a human performance benchmark is established on a subset of MMCV.
- 13
- 13
- Counterfactual and Factual Reasoning over Hypergraphs for Interpretable Clinical Predictions on EHR.13
- 13
- 12
- 12
- 12
- Integrating Planning into Single-Turn Long-Form Text Generation12
This paper generates intermediate steps via an auxiliary task that teaches the LLM to plan, reason and structure before generating the final text in a single auxiliary task that does not require multiple rounds of prompting or planning.
- MuG: A Multimodal Classification Benchmark on Game Data with Tabular, Textual, and Visual Fields12
This work proposes a multimodal classification benchmark MuG with eight datasets that allows researchers to evaluate and improve their models, and presents experimental results obtained by several state-of-the-art unimodal classifiers and multimodAL classifiers, which demonstrate the challenging and multi-modal-dependent properties of the benchmark.
- 12
- 12
- 11
- Controllable Gradient Item Retrieval11
A weakly-supervised method is proposed that can learn a disentangled item representation from user-item interaction data and ground the semantic meaning of attributes to dimensions of the item representation.
- 11
- 11
- Unified Heterogeneous Hypergraph Construction for Incomplete Multimedia Recommendation10
The HIRE framework with Sparse constraint named HIRES is devised, which uniquely integrates optimal transport and a \(\ell_{2,1}\) -norm to refine the hypergraph structure, establishing it as a powerful tool for personalized multimedia recommendations.
- Uncertainty-Aware Pre-Trained Foundation Models for Patient Risk Prediction via Gaussian Process10
Gaussian Process-based foundation models are proposed to enable the generation of accurate predictions with instance-level uncertainty quantification, thus allowing healthcare professionals to make more informed and cautious decisions, and enhancing patient outcomes with less expert intervention.
- Deep Dag Learning of Effective Brain Connectivity for FMRI Analysis10
DABNet is proposed, a Deep DAG learning framework based on Brain Networks for fMRI analysis, which adopts a brain network generator module, which harnesses the D AG learning approach to transform the raw time-series into effective brain connectivities.
- Motif-guided heterogeneous graph deep generation10
A general, end-to-end framework— HGEN for generating novel heterogeneous graphs with a newly proposed heterogeneous walk generator and a network motif generator to better characterize the higher-order structural distribution is formulated.
- 10
- 10
- 10
- 10
- PPG-Distill: Efficient Photoplethysmography Signals Analysis via Foundation Model Distillation9
PPG-Distill is presented, a knowledge distillation framework that transfers both global and local knowledge through prediction-, feature-, and patch-level distillation, and incorporates morphology distillation to preserve local waveform patterns and rhythm distillation to capture inter-patch temporal structures.
- Privacy-Enhancing Paradigms within Federated Multi-Agent Systems9
This paper proposes Embedded Privacy-Enhancing Agents (EPEAgent), an innovative solution that integrates seamlessly into the Retrieval-Augmented Generation (RAG) phase and the context retrieval stage, and minimizes data flows, ensuring that only task-relevant, agent-specific information is shared.
- Gra-CRC-miRTar: The pre-trained nucleotide-to-graph neural networks to identify potential miRNA targets in colorectal cancer9
Gra-CRC-miRTar is proposed, a pre-trained nucleotide-to-graph neural network framework for identifying potential miRNA targets in CRC that achieves better performance than other deep learning algorithms and existing predictors and provides an accurate and efficient framework to identify potential miRNA targets in CRC.
- 9
- Representation Learning of Temporal Graphs with Structural Roles9
A novel Role-based Temporal Graph Convolution Network (RTGCN) that fully leverages the global structural role information in temporal graphs and can effectively capture the static global structural roles by using hypergraph convolution neural networks.
- Enhancing Personalized Healthcare via Capturing Disease Severity, Interaction, and Progression9
This work proposes to perform personalized diagnosis prediction based on EHR data via capturing disease severity, interaction, and progression via severity-driven embeddings at the disease level, and enables personalized disease representations via severity-driven embeddings at the disease level.
- Scalable Graph Condensation with Evolving Capabilities8
This work introduces a novel framework for continual graph condensation, enabling efficient updates to the distilled graph that handle data streams without requiring costly retraining, and achieves better performance than most state-of-the-art graph condensation methods while delivering an around 1000× speedup on large datasets.
- BoxCare: A Box Embedding Model for Disease Representation and Diagnosis Prediction in Healthcare Data8
This work proposes to project disease embeddings and ICD-9 code embeddings into boxes, where a box is an axis-aligned hyperrectangle with a geometric region and two boxes can clearly "include" or "exclude" each other.
- 8
- Are Synthetic Time-series Data Really not as Good as Real Data?8
This paper has developed a method based on synthetic data that enables model training without the need for real data, surpassing the performance of models trained with real data and training a universal feature extractor based on the authors' synthetic data that is applicable to all time-series data.
- 8
- KoMen: Domain Knowledge Guided Interaction Recommendation for Emerging Scenarios8
KoMen first learns a set of global model parameters and then quickly adapts the parameters for an emerging scenario based on its similarities with the existing ones, which reduces model variance resulting from data scarcity while still achieving the expressiveness to handle diverse scenarios.
- Dynamic Network Anomaly Modeling of Cell-Phone Call Detail Records for Infectious Disease Surveillance8
Using cell-phone call records (CDR) linked with diagnostic information from Iceland during the H1N1v influenza outbreak, it is shown that GraphDNA outperforms state-of-the-art baselines on individual Date- of-Diagnosis (DoD) prediction, while tracking the epidemic signal in the overall population.
- Secure Network Release with Link Privacy8
This paper leverage the differential privacy (DP) framework, to formulate and enforce rigorous privacy constraints on deep graph generation models, with a focus on edge-DP to guarantee individual link privacy.
- ClusMFL: A Cluster-Enhanced Framework for Modality-Incomplete Multimodal Federated Learning in Brain Imaging Analysis7
ClusMFL is proposed, a novel MFL framework that leverages feature clustering for cross-institutional brain imaging analysis under modality incompleteness and employs a modality-aware aggregation strategy, further enhancing the model’s performance in scenarios with severe modality incompleteness.
- 7
- Shift-Robust Node Classification via Graph Adversarial Clustering7
This work introduces an unsupervised cluster GNN on target graph to group the similar nodes by graph homophily and demonstrates the superior accuracy of SRNC on generalizing to test graph with data shift.
- 7
- Integrating Large Language Models and Knowledge Graphs for Next-level AGI6
This tutorial will empower researchers and practitioners to unlock the next level of AGI by integrating the strengths of LLMs and KGs into cohesive, intelligent systems.
- 6
- 6
- 6
- MedAlign: Enhancing Combinatorial Medication Recommendation with Multi-modality Alignment6
This work proposes MedAlign, a novel combinatorial Medication recommendation framework with multi-modality Alignment, which aligns distinct modality distributions of medications within a unified latent space, generating consistent medication representations.
- Open Visual Knowledge Extraction via Relation-Oriented Multimodality Model Prompting6
This work presents OpenVik, a first exploration to a new paradigm of open visual knowledge extraction which consists of an open relational region detector to detect regions potentially containing relational knowledge and a visual knowledge generator that generates format-free knowledge by prompting the large multimodality model with the detected region of interest.
- FedBrain: Federated Training of Graph Neural Networks for Connectome-based Brain Imaging Analysis6
FedBrain is proposed, a GNN-based personalized FL framework that takes into account the unique properties of brain network data, and presents a federated atlas mapping mechanism to overcome the feature and structure heterogeneity of brain networks arising from different ROI atlas systems.
- Graph Neural Network Modeling of Web Search Activity for Real-time Pandemic Forecasting6
This study proposes a novel self-supervised message-passing neural network framework for modeling local and cross-location dynamics in pandemic forecasting and demonstrates that the SMPNN model outperforms other models by achieving up to a 6.9% improvement in prediction accuracy and lower prediction errors during the early stages of disease outbreaks.
- 6
- 6
- 6
- Relationship Profiling over Social Networks: Reverse Smoothness from Similarity to Closeness6
A novel reverse smoothness principle is proposed by observing that the similarity-closeness duality of homophily is consistent with the well-known smoothness assumption in graph-based semi-supervised learning-- only the direction of inference is reversed.
- Abstract WP175: Predicting Post-Stroke Cognitive Impairment (PSCI) Using Multiple Machine Learning Approaches5
Hypergraph model approaches outperform traditional machine learning methods in utilizing EHRs for predicting PSCI after a stroke, and was associated with higher ACC, AUC, AUPR, and Macro-F1 than other models.
- MedAssist: LLM-Empowered Medical Assistant for Assisting the Scrutinization and Comprehension of Electronic Health Records5
MedAssist, a large language model (LLM)-empowered medical assistant designed to support the scrutinization and comprehension of electronic health records (EHRs) and advances patient care in realistic clinical scenarios.
- How to use Graph Data in the Wild to Help Graph Anomaly Detection?5
This work proposes a novel framework Wild-GAD, built upon a unified database, UniWildGraph, which comprises a large and diverse collection of graph data with broad domain coverage, ample data volume, and a unified feature space, and develops selection criteria based on representativity and diversity to identify the most suitable external data for each anomaly detection task.
- Evaluation of General Large Language Models in Contextually Assessing Semantic Concepts Extracted from Adult Critical Care Electronic Health Record Notes5
A comprehensive qualitative performance evaluation framework for LLMs is developed and operationalized that validates LLMs' capabilities in processing complex medical data but also establishes a benchmark for future LLM evaluations across specialized domains.
- 5
- 5
- 5
- Weakly Supervised Concept Map Generation Through Task-Guided Graph Translation5
GT-D2G (Graph Translation-based Document To Graph), an automatic concept map generation framework that leverages generalized NLP pipelines to derive semantic-rich initial graphs, and translates them into more concise structures under the weak supervision of downstream task labels is presented.
- 5
- Query-Specific Knowledge Summarization with Entity Evolutionary Networks5
To facilitate such a novel insightful search system, the proposed SetEvolve is proposed, which is a unified framework based on nonparanomal graphical models for evolutionary network construction from large text corpora.
- BioMedJImpact: A Comprehensive Dataset and LLM Pipeline for AI Engagement and Scientific Impact Analysis of Biomedical Journals4
This work analyzes how collaboration intensity and AI engagement rate jointly influence scientific impact across two temporal subsets (2016-2019, 2020-2023) and identifies two main patterns emerge: journals with larger author teams tend to have higher citation impact, while AI engagement rate is positively associated with Impact Factor only in the 2019 subset.
- Federated learning for cross-institution brain network analysis4
FedBrain is proposed, a GNN-based personalized FL framework that takes into account the unique properties of brain network data, and presents a federated atlas mapping mechanism to overcome the feature and structure heterogeneity of brain networks arising from different ROI atlas systems.
- User-Guided Clustering in Heterogeneous Information Networks via Motif-Based Comprehensive Transcription4
This work comprehensively transcribes the higher-order interaction signals to a series of tensors via motifs and proposes the MoCHIN model based on joint non-negative tensor factorization, which outperforms all baselines in three evaluation tasks under three different metrics.
- 4
- EHRBench: An Automated and Reliable EHR-based Benchmark for Clinical Decision Making with LLMs3
EHRBench is introduced, an automated and reliable EHR-grounded benchmark for evaluating LLM-based clinical decision-making at scale and shows consistent capability trends across settings, further validating the reliability of EHRBench and highlighting actionable gaps toward clinically reliable LLM systems.
- An Electrocardiogram Multi-Task Benchmark with Comprehensive Evaluations and Insightful Findings3
The experimental results show that general time-series / ECG foundation models achieve a top performance rate of 80%, indicating their effectiveness in ECG analysis, and highlights the limitations and potential of foundation models in advancing physiological waveform analysis.
- 3
- Towards Automatic Evaluation and Selection of PHI De-identification Models via Multi-Agent Collaboration3
By combining independent evaluation agents with LLM majority voting, TEAM-PHI offers a practical, secure, and cost-effective solution for automatic evaluation and best-model selection in PHI de-identification, even when ground-truth labels are limited.
- Abstract TMP102: Title: Prediction of Post-Stroke AF in ESUS Patients is Enhanced by Combining Expert-Derived Predictors and Embedding of Full Diagnostic Codes using Pre-Trained Hypergraph Neural Networks3
The results show that the learned 32-dim embedding vectors improves the prediction of post-ESUS AF and the embedding technique can significantly enhance predictive performance by integrating comprehensive medical information, maximizing the use of available data for improved outcomes.
- 3
- 3
- FedAA: Using Non-sensitive Modalities to Improve Federated Learning while Preserving Image Privacy3
Experiments show that FedAA significantly outperforms recent non-IID federated learning algorithms while preserving image privac, by sharing information from non-sensitive modalities.
- 3
- Graph Federated Learning with Hidden Representation Sharing3
This work first formulate the Graph Federated Learning (GFL) problem that unifies LoG and FL in multi-client systems and then proposes sharing hidden representation instead of the raw data of neighbors to protect data privacy as a solution.
- 3
- 2
- 2
- Seeing Through the Brain: New Insights from Decoding Visual Stimuli with fMRI2
PRISM is proposed, a model that Projects fMRI sIgnals into a Structured text space as an interMediate representation for visual stimuli reconstruction that outperforms existing methods and highlights the importance of using structured text as the intermediate space to bridge fMRI signals and image reconstruction.
- 2
- 2
- Large Language Model Empowered Logical Relations Mining for Personalized Recommendation2
RelRec is proposed, a novel approach that leverages large language models (LLMs) to mine logical relations and satisfy users' logical requirements in personalized recommendation tasks and significantly outperforms existing methods.
- Future Matters for Present: Towards Effective Physical Simulation over Meshes2
Through pyramid-like alternative propagation between the foresight step and refinement step, the proposed framework FAIR can generate accurate long-term trajectories, achieving a significant error reduction compared with the best baseline on four benchmark datasets.
- 2
- HypMix: Hyperbolic Representation Learning for Graphs with Mixed Hierarchical and Non-hierarchical Structures2
This work proposes a new hyperbolic representation learning model that can handle complex hierarchical structures and also learn the representation of both hierarchical and non-hierarchic structures.
- Federated Node Classification over Distributed Ego-Networks with Secure Contrastive Embedding Sharing2
An FGL method is proposed for distributed ego-networks in which clients obtain complete neighborhood information of local nodes through sharing node embeddings with other clients, and a contrastive learning mechanism is proposed to bridge the gap between local and global node embeddings and stabilize the local training of graph neural network models.
- BrainSTEAM: A Practical Pipeline for Connectome-based fMRI Analysis towards Subject Classification2
This study proposes BrainSTEAM, an integrated framework featuring a spatio-temporal module that consists of an EdgeConv GNN model, an autoencoder network, and a Mixup strategy that aims to dynamically segment the time series signals of the ROI features for each subject into chunked sequences.
- 2
- 2
- Higher-Order Clustering in Heterogeneous Information Networks.2
The MoCHIN model is proposed which is applicable to arbitrary forms of HIN motifs, which is often necessary for the application scenario in HINs due to their rich and diverse semantics encapsulated in the heterogeneity and to overcome the curse of dimensionality.
- Awaken the Giant: Activating LLMs via Deep Model Guidance for Boundary-aware Medication Recommendation1
GiantMed leverages a deep model to identify boundary medications and directs the LLM to focus on these clinically ambiguous yet informative cases, and augment contextual knowledge of boundary medications by retrieving relevant historical EHRs and integrating Drug-Drug Interaction constraints.
- FedKDD 2025: The 2025 International Joint Workshop on Federated Learning for Data Mining and Graph Analytics1
This workshop will encourage ideas exchange on the trustworthiness, scalability, and robustness of distributed data mining and graph analytics and their emergent challenges and their emergent challenges.
- Beyond Prediction: AI as a Creative Partner in Organic Synthesis1
Artificial intelligence’s role in organic synthesis is still mostly viewed in two ways: helping automate retrosynthetic planning and speeding up reaction optimization.
- MANI-Pure: Magnitude-Adaptive Noise Injection for Adversarial Purification1
MANI-Pure is introduced, a magnitude-adaptive purification framework that leverages the magnitude spectrum of inputs to guide the purification process, effectively suppressing adversarial perturbations in fragile high-frequency, low-magnitude bands while preserving semantically critical low-frequency content.
- Type 2 Diabetes Subtyping via Phenotype and Genotype Co-Learning1
A cohort of 42,256 participants from the National Institutes of Health’s All of Us Research Program is selected, where the proposed hypergraph framework achieves an AUROC of 89.64% on predicting T2D when integrating phenotypic and genetic features.
- Capturing the Additional Cardiovascular Benefits of SGLT2 Inhibitors and GLP-1 Receptor Agonists Beyond the Control of Traditional Risk Factors in People With Diabetes1
This study revealed that SGLT2is could further reduce CHF risks beyond the control of traditional risk factors but may offer additional pathways to offset overall benefits from traditional risk factor control in stroke risks.
- Boosting E-commerce Content Diversity: A Graph-based RAG Approach with User Reviews1
This work proposes a novel framework that integrates graph-based knowledge into Retrieval-Augmented Generation (RAG) to enhance content generation, and leverages user reviews to construct an item-feature graph that significantly improves diversity while preserving fidelity.
- Medical Retrieval-Augmentation Generation Framework for Healthcare Prediction1
This work proposes MedGR, a novel framework for healthcare prediction that incorporates graph-based clinical text indexing with a dual-level medical retrieval architecture that synthesizes information from multiple sources into coherent and contextually enriched responses in an efficient manner.
- 1
- Measuring Spiritual Values and Bias of Large Language Models1
This work starts with verification of the hypothesis that different spiritual values indeed lead to different sensitivity to different hate target groups, and proposes to continue pre-training LLMs on spiritual texts, demonstrating the effectiveness of this approach in mitigating spiritual bias.
- FedKDD: International Joint Workshop on Federated Learning for Data Mining and Graph Analytics1
This workshop will encourage ideas exchange on the trustworthiness, scalability, and robustness of distributed data mining and graph analytics and their emergent challenges and their emergent challenges.
- Enhancing Progressive Diagnosis Prediction in Healthcare with Continuous Normalizing Flows1
This work proposes an enhanced progressive diagnostic prediction model (i.e., ProCNF), which integrates continuous normalizing flows (CNF) and neural ordinary differential equations (ODEs) to achieve more accurate approximations of patient health trajectories while capturing the continuity underlying disease progression.
- From Basic to Extra Features: Hypergraph Transformer Pretrain-then-Finetuning for Balanced Clinical Predictions on EHR1
This work proposes HTP–Star1, which leverages hypergraph structures with a pretrain-then-finetune framework for modeling EHR data, enabling seamless integration of additional features and demonstrates that HTP–Star consistently outperforms various baselines while striking a balance between patients with basic and extra features.
- 1
- LENS: label sparsity-tolerant adversarial learning on spatial deceptive reviews1
Extensive experiments show that LENS consistently outperformed the state-of-the-art semi-supervised fake review detection methods on few labels at all different labeling rates for real and fake reviews, respectively, in a label-starving setting.
- A Bird's-Eye Tutorial of Graph Attention Architectures1
This tutorial looks at GNN variants from the point of view of the attention function and iteratively builds the reader's understanding of different graph attention variants to benefit researchers dealing with graph-structured problems.
- 1
- Structure-Preserving Graph Kernel for Brain Network Classification1
Results show that relevant EEG signals are primarily encoded in the alpha and theta bands during the emotion regulation task, which is consistent with previous findings.
- 1
- 1
- The Second Workshop on SciSoc Agents & LLMs: Agentic AI for Scientific and Societal Advances–
This workshop aims to foster a comprehensive understanding of how LLMs and agent-based architectures jointly redefine traditional research methodologies, and explore novel agentic frameworks, multi-agent collaboration, and real-world deployments that leverage LLM-powered agents to accelerate scientific progress and generate meaningful societal impact.
- –
- –
- –
- –
- –
- Why Empirical Risk Minimization Performs Well for Open Set Domain Adaptation: A Theoretical Analysis From Causal View–
This work forms the novel concepts of the fully informative causal invariance model (FICIM) and the partially informative causal invariance model (PICIM) and derives an OSDA theoretical bound to prove that the ERM performs well when the source domain follows FICIM, while it performs poorly when the source domain follows PICIM.
- Cognitive Bifurcation: Dual-Progressive Causal Diffusion with Hippocampal Memory for Continual Graph Learning–
The Evolving Hippocampal Memory is introduced that re-activates this global knowledge to distill wisdom back into the fast adapter, ensuring robust adaptation to continuous distribution shifts.
- RelSciFM 2026: The workshop on Reliable Scientific Foundation Models: Design, Training, Grounding, and Verification–
The workshop aims to bring together AI researchers and domain scientists to distill shared challenges, surface best practices, and produce community artifacts such as benchmark specifications and evaluation checklists for trustworthy SciFMs.
- –
- A Benchmark Analysis of Graph and Non-Graph Methods for Caenorhabditis Elegans Neuron Classification–
A benchmark for Caenorhabditis elegans neuron classification is established, comparing four graph methods against four non-graph methods and highlighting that Spatial and Connection features are key predictors for Caenorhabditis elegans neuron classes.
- –
- –
- FedKDD/FedMAS 2026: The 2026 International Joint Workshop on Federated Learning for Multi-agent Systems and Data Mining–
This workshop seeks to bring together researchers and practitioners from academia and industry to explore the convergence of federated learning and multi-agent systems, with a focus on scalability, trustworthiness, adaptive coordination, and the broader societal impact of federated multi-agent intelligence.
- EviCare: Enhancing Diagnosis Prediction with Deep Model-Guided Evidence for In-Context Reasoning–
EviCare is an in-context reasoning framework that integrates deep model guidance into LLM-based diagnosis prediction, which consistently outperforms both LLM-only and deep model-only baselines by an average of 20.65% across precision and accuracy metrics.
- –
- Beyond Prediction: Toward Verifiable Physiological Waveform Reasoning: Advances, Challenges, and Future Directions–
This tutorial advocates verifiable physiological waveform reasoning, where conclusions are supported by explicit evidence objects, and highlights a dual-process blueprint and a Verification Ledger that enable evidence-grounded, replayable assessment beyond accuracy, including traceability, robustness, and calibrated abstention.
- KDD Health Day 2025: Harnessing AI Opportunities in Biomedicine and Healthcare–
An extended abstract discusses recent advancements, challenges, and future directions of AI ecosystems, focusing on integrating AI-ready data sets, interdisciplinary collaborations, and ethical AI practices.
- –
- MaSH: Maximal Separating Poincaré Hyperplanes for Hierarchical Imbalanced Learning–
A hyperbolic geometric inductive bias that enhances the generalization capability of hyperbolic classifiers, especially on the class-imbalanced settings, as well as performing experiments on imbalanced/long-tailed classification.
- –
- –
- –
- Build fair machine learning models to predict adverse outcomes for Heart failure patients with preserved ejection fraction (HFpEF) and with reduced ejection fraction (HFrEF)–
The prediction models for HFpEF and HFrEF offer an explainable and equity-enhancing approach for risk stratification in personalized HF clinical care and show improved prediction utility and support targeted interventions that are essential for patients' health outcomes and critical for clinical decision-making.
- –
- –
- –
- Attack as Defense: Run-time Backdoor Implantation for Image Content Protection–
This work innovatively prevent the abuse of image content modification by implanting the backdoor into image-editing models through the first framework for run-time backdoor implantation, which is both time- and resource- efficient.
- Evaluating the Impact of Lab Test Results on Large Language Models Generated Differential Diagnoses from Clinical Case Vignettes–
Lab tests, including liver function, metabolic/toxicology panels, and serology/immune tests, were generally interpreted correctly by LLMs for differential diagnosis, and lab results significantly improved accuracy.
- –
- The 1st International Workshop on Federated Learning with Graph Data (FedGraph)–
The nascent field of federated learning (FL), which aims to enable individual clients to jointly train their models while keeping their local data decentralized and completely private, is a promising paradigm for large-scale distributed and private training of GNNs.
- TAXOGAN: Hierarchical Network Representation Learning via Taxonomy Guided Generative Adversarial Networks (Extended Abstract)–
This work aims to improve network embedding by modeling the conditional node proximity in networks indicated by node labels residing in real taxonomies, which is too coarse by solely looking at the hierarchical topologies.
- Modeling Evolutionary State Graph for Time Series Prediction–
This paper presents evolutionary state graph, a dynamic graph structure designed to systematically represent the evolving relations (edges) among states (nodes) along time, and proposes a novel graph neural network model, Evolutionary State Graph Network (EvoNet), to encode the evolutionary stategraph for accurate and interpretable time-series event prediction.
- Flash memory performance on a highly scalable IOV system–
This work describes an IOV system built on the PCI-Express over Ethernet (PCIeOE) protocol, which combines these two ubiquitous, standardized technologies in a novel fashion and presents a thorough characterization of the latency and performance of the system using very high performance flash memory as an endpoint device.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.