Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Xingquan ZhuSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works465 from public data
- Graph WaveNet for Deep Spatial-Temporal Graph Modeling3,648
This paper proposes a novel graph neural network architecture, Graph WaveNet, for spatial-temporal graph modeling by developing a novel adaptive dependency matrix and learn it through node embedding, which can precisely capture the hidden spatial dependency in the data.
- A Survey on Knowledge Graphs: Representation, Acquisition, and Applications3,000
A comprehensive review of the knowledge graph covering overall research topics about: 1) knowledge graph representation learning; 2) knowledge acquisition and completion; 3) temporal knowledge graph; and 4) knowledge-aware applications and summarize recent breakthroughs and perspective directions to facilitate future research.
- Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks2,376
This paper proposes a general graph neural network framework designed specifically for multivariate time series data that outperforms the state-of-the-art baseline methods on 3 of 4 benchmark datasets and achieves on-par performance with other approaches on two traffic datasets which provide extra structural information.
- Unifying Large Language Models and Knowledge Graphs: A Roadmap1,833
A forward-looking roadmap for the unification of LLMs and KGs, in which LLMs and KGs play equal roles and work in a mutually beneficial way to enhance both LLMs and KGs for bidirectional reasoning driven by both data and knowledge.
- Adversarially Regularized Graph Autoencoder for Graph Embedding902
A novel adversarial graph embedding framework for graph data that encodes the topological structure and node content in a graph to a compact representation, on which a decoder is trained to reconstruct the graph structure.
- DiSAN: Directional Self-Attention Network for RNN/CNN-Free Language Understanding794
A novel attention mechanism in which the attention between elements from input sequence(s) is directional and multi-dimensional (i.e., feature-wise) and a light-weight neural net is proposed, based solely on the proposed attention without any RNN/CNN structure, which outperforms complicated RNN models on both prediction quality and time efficiency.
- Deep Learning for Time Series Anomaly Detection: A Survey791
This survey provides a structured and comprehensive overview of state-of-the-art deep learning for time series anomaly detection and provides a taxonomy based on anomaly detection strategies and deep learning models.
- Graph Self-Supervised Learning: A Survey786
A unified framework is constructed that mathematically formalizes the paradigm of graph SSL and describes the applications of graphSSL across various research fields and summarize the commonly used datasets, evaluation benchmark, performance comparison and open-source codes ofgraph SSL.
- Attributed Graph Clustering: A Deep Attentional Embedding Approach669
This paper proposes a goal-directed deep learning approach to graph clustering, Deep Attentional Embedded Graph Clustering (DAEGC), which focuses on attributed graphs to sufficiently explore the two sides of information in graphs.
- Reasoning on Graphs: Faithful and Interpretable Large Language Model Reasoning582
This paper proposes a novel method called reasoning on graphs (RoG) that synergizes LLMs with KGs to enable faithful and interpretable reasoning and presents a planning-retrieval-reasoning framework, where RoG first generates relation paths grounded by KGs as faithful plans.
- A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection578
This survey, for the first time, brings together a vast array of knowledge on GNN-based time series research, highlighting foundations, practical applications, and opportunities of graph neural networks for time series analysis.
- Anomaly Detection on Attributed Networks via Contrastive Self-Supervised Learning537
This article presents a novel Contrastive self-supervised Learning framework for Anomaly detection on attributed networks (CoLA for abbreviation), which fully exploits the local information from network data by sampling a novel type of contrastive instance pair, which can capture the relationship between each node and its neighboring substructure in an unsupervised way.
- MGAE: Marginalized Graph Autoencoder for Graph Clustering537
A marginalized graph convolutional network is proposed to corrupt network node content, allowing node content to interact with network features, and marginalizes the corrupted features in a graph autoencoder context to learn graph feature representations.
- Adversarially Regularized Graph Autoencoder.504
A novel adversarial graph embedding framework for graph data that encodes the topological structure and node content in a graph to a compact representation, on which a decoder is trained to reconstruct the graph structure.
- Foundation Models for Time Series Analysis: A Tutorial and Survey497
This survey serves to consolidate the latest advancements in FMs pertinent to time series analysis, accentuating their theoretical underpinnings, recent strides in development, and avenues for future exploration.
- Tri-party deep network representation439
TriDNR is a tri-party deep network representation model, using information from three parties: node structure, node content, and node labels (if available) to jointly learn optimal node representation, which results in up to 79% classification accuracy gain, compared to state-of-the-art methods.
- A Survey of Community Detection Approaches: From Statistical Modeling to Deep Learning396
A unified architecture of network community-finding methods to characterize the state-of-the-art of the field of community detection is developed and a new taxonomy that divides the existing methods into two categories, namely probabilistic graphical model and deep learning is introduced.
- Learning Graph Embedding With Adversarial Training Methods375
This article presents a novel adversarially regularized framework for graph embedding, employing the graph convolutional network as an encoder, that embeds the topological information and node content into a vector representation, from which a graph decoder is further built to reconstruct the input graph.
- Graph Neural Networks for Graphs With Heterophily: A Survey343
A systematic taxonomy that governs existing heterophilic GNN models for heterophilic graphs is proposed, aiming to facilitate the development of more effective GNNs across a spectrum of practical applications and learning tasks in the graph research community.
- Self-Supervised Learning for Time Series Analysis: Taxonomy, Progress, and Prospects304
This article comprehensively review existing surveys related to SSL and time series, and provides a new taxonomy of existing time series SSL methods by summarizing them from three perspectives: generative- based, contrastive-based, and adversarial-based.
- Towards Unsupervised Deep Graph Structure Learning299
A novel StrUcture Bootstrapping contrastive LearnIng fraMEwork (SUBLIME for abbreviation) with the aid of self-supervised contrastive learning is proposed, where the learned graph topology is optimized by data itself without any external guidance.
- 261
- Unsupervised Domain Adaptive Graph Convolutional Networks241
A novel approach, unsupervised domain adaptive graph convolutional networks (UDA-GCN), for domain adaptation learning for graphs, which jointly exploits local and global consistency for feature aggregation and facilitates knowledge transfer between graphs.
- Hyperspectral Image Classification With Context-Aware Dynamic Graph Convolutional Network231
A new HSI classification method based on the recently proposed Graph Convolutional Network (GCN), as it can flexibly encode the relations among arbitrarily structured non-Euclidean data, which yields significant improvement in the classification performance when compared with some state of theart approaches.
- 225
- Supervised Learning for Suicidal Ideation Detection in Online User Content214
An approach to understand suicidal ideation through online user-generated content with the goal of early detection via supervised learning byalysing users’ language preferences and topic descriptions reveals rich knowledge that can be used as an early warning system for detecting suicidal tendencies.
- 213
- Large Models for Time Series and Spatio-Temporal Data: A Survey and Outlook212
This survey presents a comprehensive, up-to-date review of large models tailored or adapted for time series and spatio-temporal data along four dimensions: data types, model categories, model scopes, and application areas/tasks.
- Multivariate Time Series Forecasting With Dynamic Graph Neural ODEs206
A continuous model to forecast multivariate time series forecasting is proposed, designed and solved to complement missing graph topologies and unify both spatial and temporal message passing, allowing deeper graph propagation and fine-grained temporal information aggregation to characterize stable and precise latent spatial-temporal dynamics.
- ANEMONE200
A novel graph anomaly detection framework, namely ANEMONE, to simultaneously identify the anomalies in multiple graph scales and employs a statistical anomaly estimator to evaluate the abnormality of each node according to the degree of agreement from multiple perspectives.
- Continual Learning for Large Language Models: A Survey198
This paper surveys recent works on continual learning for LLMs and catalogs continue learning techniques in a novel multi-staged categorization scheme, involving continual pretraining, instruction tuning, and alignment.
- 196
- An Empirical Survey on Long Document Summarization: Datasets, Models, and Metrics196
This survey provides a comprehensive overview of the research on long document summarization and a systematic evaluation across the three principal components of its research setting: benchmark datasets, summarization models, and evaluation metrics.
- Multi-Scale Contrastive Siamese Networks for Self-Supervised Graph Representation Learning194
A novel self-supervised approach to learn node representations by enhancing Siamese self-distillation with multi-scale contrastive learning that not only achieves new state-of-the-art results but also surpasses some semi- supervised counterparts by large margins.
- Trustworthy Graph Neural Networks: Aspects, Methods, and Trends192
A comprehensive roadmap to build trustworthy GNNs from the view of the various computing technologies involved is proposed and basic concepts and comprehensively summarize existing efforts for trustworthy GNNs are introduced, including robustness, explainability, privacy, fairness, accountability, and environmental well-being.
- Learning Private Neural Language Modeling with Attentive Aggregation180
This work proposes a novel model aggregation with an attention mechanism considering the contribution of client models to the global model, together with an optimization technique during server aggregation that minimizes the weighted distance between the server model and client models.
- Triformer: Triangular, Variable-Specific Attentions for Long Sequence Multivariate Time Series Forecasting--Full Version174
Strong empirical evidence on four datasets from multiple domains justifies the design choices, and it demonstrates that Triformer outperforms state-of-the-art methods w.r.t. both accuracy and efficiency.
- Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning170
This paper presents a novel GCN-based SSL algorithm which aims to enrich the supervision signals by utilizing both data similarities and graph structure, and designs a semi-supervised contrastive loss related to input features.
- Emerging trends in federated learning: from model fusion to federated X learning159
Various learning algorithms to improve the vanilla federated averaging algorithm are explored and model fusion methods such as adaptive aggregation, regularization, clustered methods, and Bayesian methods are reviewed.
- CARLA: Self-supervised contrastive representation learning for time series anomaly detection155
A novel end-to-end self-supervised ContrAstive Representation Learning approach for time series Anomaly detection (CARLA), which leverages existing generic knowledge about time series anomalies and injects various types of anomalies as negative samples.
- Rethinking and Scaling Up Graph Contrastive Learning: An Extremely Efficient Approach with Group Discrimination155
This work revisits GCL and introduces a new learning paradigm for self-supervised graph representation learning, namely, Group Discrimination (GD), and proposes a novel GD-based method called Graph Group discrimination (GGD), which outperforms its GCL counterparts in both accuracy and efficiency.
- Towards Spatio-Temporal Aware Traffic Time Series Forecasting--Full Version154
This work proposes a framework that aims at turning spatio-temporal agnostic models to spatio -temporal aware models, and encode time series from different locations into stochastic variables, from which they generate location-specific and time-varying model parameters to better capture the spatio/temporal dynamics.
- Spatio-Temporal Joint Graph Convolutional Networks for Traffic Forecasting144
This work proposes a novel approach called Spatio-Temporal Joint Graph Convolutional Networks (STJGCN) for accurate traffic forecasting on road networks over multiple future time steps that encompasses the construction of both pre-defined and adaptive spatio-temporal joint graphs between any two time steps.
- A Survey on Diffusion Models for Time Series and Spatio-Temporal Data143
By separating applications for time series and spatio-temporal data, this study offers a structured perspective on model category, task type, data modality, and practical application domain.
- A Relation-Specific Attention Network for Joint Entity and Relation Extraction140
A relation-specific attention network (RSAN) that utilizes relation-aware attention mechanism to construct specific sentence representations for each relation, and then performs sequence labeling to extract its corresponding head and tail entities.
- 140
- A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment139
This paper introduces the concept of "full-stack" safety to systematically consider safety issues throughout the entire process of LLM training, deployment, and eventual commercialization, and represents the first safety survey to encompass the entire lifecycle of LLMs.
- 139
- A survey on neural-symbolic learning systems133
This research aims to propel this emerging field forward, offering researchers a comprehensive and holistic overview of the advancements in neural-symbolic learning systems from four distinct perspectives: challenges, methods, applications, and future directions.
- Large language models for scientific discovery in molecular property prediction130
LLM4SD, a framework designed to harness LLMs for driving scientific discovery in molecular property prediction by synthesizing knowledge from literature and inferring knowledge from scientific data, is introduced.
- Time series feature learning with labeled and unlabeled data128
A new Semi-Supervised Shapelets Learning model is presented to efficiently learn shapelets by using both labeled and unlabeled time series data in an integrated model that considers the least squares regression, the power of the pseudo-labels, shapelets regularization, and spectral analysis.
- 120
- Task-adaptive Neural Process for User Cold-Start Recommendation120
A novel meta-learning recommender called task-adaptive neural process (TaNP), a new member of the neural process family, where making recommendations for each user is associated with a corresponding stochastic process that enables the model to learn the relevance of different tasks and customize the global knowledge to the task-related decoder parameters for estimating user preferences.
- Physicochemical graph neural network for learning protein–ligand interaction fingerprints from sequence data117
This work introduces PSICHIC (PhySIcoCHemICal graph neural network), a framework uniquely incorporating physicochemical constraints to decode interaction fingerprints directly from sequence data alone, achieving state-of-the-art accuracy and interpretability in deciphering mechanisms underlying protein-ligand interactions.
- 116
- Structure-free Graph Condensation: From Large-scale Graphs to Condensed Graph-free Data115
This paper advocates a new Structure-Free Graph Condensation paradigm, named SFGC, to distill a large-scale graph into a small-scalegraph node set without explicit graph structures, i.e., graph-free data, whose topology is reduced to an identity matrix.
- Reasoning Like Human: Hierarchical Reinforcement Learning for Knowledge Graph Reasoning113
A novel Hierarchical Reinforcement Learning framework to learn chains of reasoning from a Knowledge Graph automatically, inspired by the hierarchical structure through which human handle cognitionally ambiguous cases is proposed.
- 113
- Correlation-Aware Spatial–Temporal Graph Learning for Multivariate Time-Series Anomaly Detection105
This work proposes a novel method, correlation-aware spatial–temporal graph learning (termed CST-GL), for time-series anomaly detection that can detect and diagnose anomalies effectively in general settings as well as enable early detection across different time delays.
- 104
- 104
- Transfer learning in agriculture: a review103
This paper provides a comprehensive examination of TL-based agricultural applications, covering pre-trained models, dataset sources, input image types, implementation platforms, and TL approaches, and identifies the challenges faced when applying TL in agriculture.
- Self-adaptive attribute weighting for Naive Bayes classification103
The proposed method, namely AISWNB, uses immunity theory in Artificial Immune Systems to search optimal attribute weight values, where self-adjusted weight values will alleviate the conditional independence assumption and help calculate the conditional probability in an accurate way.
- TimeFilter: Patch-Specific Spatial-Temporal Graph Filtration for Time Series Forecasting101
TimeFilter is proposed, a GNN-based framework for adaptive and fine-grained dependency modeling that refines the learned spatial-temporal dependencies by filtering out irrelevant correlations while preserving the most critical ones in a patch-specific manner.
- One-Shot Neural Architecture Search via Novelty Driven Sampling101
A new approach is presented, Efficient Novelty-driven Neural Architecture Search, to sample the most abnormal architecture to train the supernet, and only the weights of a single architecture sampled by the novelty search are optimized in each step to reduce the memory demand greatly.
- 100
- GOOD-D: On Unsupervised Graph Out-Of-Distribution Detection96
A new graph contrastive learning framework GOOD-D is developed, aiming at detecting OOD graphs solely based on unlabeled ID data, and a comprehensive benchmark is built to compare the proposed approach with different state-of-the-art methods.
- Deep learning data augmentation for Raman spectroscopy cancer tissue classification96
Experiments on real-world RS data demonstrate that (1) data augmentation can help improve skin cancer tissue classification accuracy, and (2) generative adversarial network can be used to generate reliable synthetic Raman spectroscopic data.
- Active Discriminative Network Representation Learning96
This paper proposes a novel method, termed as ANRMAB, to learn the active discriminative network representations with a multi-armed bandit mechanism in active learning setting, and designs three active learning query strategies based on the networking data and the learned network representations.
- An Explainable Deep Fusion Network for Affect Recognition Using Physiological Signals94
A deep learning model is proposed to process multimodal-multisensory bio-signals for affect recognition that supports batch training for different sampling rate signals at the same time, and the results show significant improvement compared to the state of the art.
- Adapting Membership Inference Attacks to GNN for Graph Classification: Approaches and Implications92
This paper takes the first step in MIA against GNNs for graph-level classification by inferring whether a graph sample has been used for training a GNN model, and confirms that, unlike the node-level classifier, MIAs on graph- level classification tasks are more co-related with the overfitting level ofGNNs rather than the statistic property of their training graphs.
- 92
- Overcoming Multi-Model Forgetting in One-Shot NAS With Diversity Maximization89
This paper proposes a Novelty Search based Architecture Selection (\textbf{NSAS}) loss function and demonstrates that the posterior probability could be calculated without the strict assumption when maximizing the diversity of the selected constraints.
- Multi-Graph Fusion Networks for Urban Region Embedding88
This paper proposes multi-graph fusion networks (MGFN) to enable the cross domain prediction tasks, and designs the multi-level cross-attention mechanism to learn the comprehensive embeddings from multiple mobility patterns based on intra-pattern and inter-pattern messages.
- Model Extraction Attacks on Graph Neural Networks87
This paper comprehensively investigate and develop model extraction attacks against GNN models and systematically formalise the threat modelling in the context of GNN model extraction and classify the adversarial threats into seven categories by considering different background knowledge of the attacker.
- Graph-constrained Reasoning: Faithful Reasoning on Knowledge Graphs with Large Language Models85
Graph-constrained reasoning (GCR) is introduced, a novel framework that bridges structured knowledge in KGs with unstructured reasoning in LLMs and achieves state-of-the-art performance and exhibits strong zero-shot generalizability to unseen KGs without additional training.
- 85
- 85
- 85
- 85
- 83
- Advances in Processing, Mining, and Learning Complex Data: From Foundations to Real‐World Applications82
This special issue contributes to the fundamental research in processing, mining, and learning complex data, focusing on the analysis of complex data sources.
- 82
- 75
- 74
- 73
- Robust Physical-World Attacks on Face Recognition71
A novel robust physical attack framework is proposed, dubbed PadvFace, to model these challenging variations of stickers, faces, and environmental conditions and an efficient Curriculum Adversarial Attack algorithm is proposed that gradually adapts adversarial stickers to environmental variations from easy to complex.
- Learning Strong Graph Neural Networks with Weak Information70
D2PT is proposed, a dual-channel GNN framework that performs long-range information propagation not only on the input graph with incomplete structure, but also on a global graph that encodes global semantic similarities.
- 70
- Contrastive Clustering Learning for Multi-Behavior Recommendation68
A novel multi-behavior recommendation model based on contrastive clustering learning (MBRCC) is proposed, and experimental results indicate that the MBRCC remarkably outperforms numerous existing recommendation models.
- 68
- Deep neighbor-aware embedding for node clustering in attributed graphs67
A clustering-directed deep learning approach, Deep Neighbor-aware Embedded Node Clustering ( DNENC for short) for clustering graph data, which encodes the topological structure and node content in a graph into a compact representation via a neighbor-aware graph autoencoder, which progressively absorbs information from neighbors via a convolutional or attentional encoder.
- 66
- 66
- LLM-based Agents Suffer from Hallucinations: A Survey of Taxonomy, Methods, and Directions65
This work presents the first comprehensive survey of hallucinations in LLM-based agents and proposes a new taxonomy that identifies different types of agent hallucinations occurring at different stages, and conducts an in-depth examination of eighteen triggering causes underlying the emergence of agent hallucinations.
- 64
- Cross-modal Clinical Graph Transformer for Ophthalmic Report Generation64
An automatic information extraction scheme based on natural language processing to obtain clinical entities and relations directly from in-domain training reports to endow models with the capability of incorporating expert knowledge.
- 64
- TxAllo: Dynamic Transaction Allocation in Sharded Blockchain Systems63
A deterministic and fast allocation scheme TxAllo is proposed to dynamically infer the allocation of accounts and their associated transactions and it directly optimizes the system throughput, considering both the number of cross-shard transactions and the workload balance among shards.
- 63
- Large Language Models in Drug Discovery and Development: From Disease Mechanisms to Clinical Trials62
This paper aims to provide a comprehensive overview for researchers and practitioners in computational biology, pharmacology, and AI4Science by offering insights into the potential transformative impact of LLMs on drug discovery and development.
- Large Language Models for Scientific Synthesis, Inference and Explanation62
This work presents a method for using general-purpose large language models to make inferences from scientific datasets of the form usually associated with special-purpose machine learning algorithms, and shows that the large language model can augment this knowledge by synthesizing from the scientific literature.
- DAGCN: Dual Attention Graph Convolutional Networks62
DAGCN automatically learns the importance of neighbors at different hops using a novel attention graph convolution layer, and then employs a second attention component, a self-attention pooling layer, to generalize the graph representation from the various aspects of a matrix graph embedding.
- Graph spatiotemporal process for multivariate time series anomaly detection with missing values61
The experimental results show that the GST-Pro method can effectively detect anomalies in time series data and outperforms state-of-the-art methods, regardless of whether there are missing values present in the data.
- Dual Space Graph Contrastive Learning61
A novel graph contrastive learning method, namely Dual Space Graph Contrastive (DSGC) Learning is proposed, to conduct graph Contrastive learning among views generated in different spaces including the hyperbolic space and the Euclidean space to bridge the spaces and leverage advantages from both sides.
- Normalizing Flow-based Neural Process for Few-Shot Knowledge Graph Completion60
A normalizing flow-based neural process for few-shot knowledge graph completion (NP-FKGC), which unify normalizing flows and neural processes to model a complex distribution of KG completion functions and introduces an attentive relation path-based graph neural network to capture path information in KGs.
- Discrete Network Embedding60
A novel discrete network embedding (DNE) for more compact representations is proposed, in particular, DNE learns short binary codes to represent each node, using the Hamming similarity between two binary embeddings to well approximate the ground-truth similarity.
- Finding the Missing-half: Graph Complementary Learning for Homophily-prone and Heterophily-prone Graphs56
Graph cOmplementAry Learning is introduced, namely GOAL, which consists of two components: graph complementation and complemented graph convolution, which finds the missing-half structural information for a given graph to complement it.
- 56
- 56
- A Comprehensive Survey on Distributed Training of Graph Neural Networks55
A comprehensive survey of distributed GNN training is provided by investigating various optimization techniques used in distributed GNN training by investigating various optimization techniques proposed by recent work.
- Beyond Low-Pass Filtering: Graph Convolutional Networks With Automatic Filtering55
This paper proposes Automatic Graph Convolutional Networks (AutoGCN) to capture the full spectrum of graph signals and automatically update the bandwidth of graph convolutional filters, based on graph spectral theory.
- 55
- Classification of Lung Nodules Based on Deep Residual Networks and Migration Learning54
The experimental results demonstrate the effectiveness of the proposed deep residual network method in lung nodule classification for CT images, which improved the accuracy and the false positive rate compared with the conventional support vector machine (SVM)-based CAD system.
- Bayesian personalized ranking based on multiple-layer neighborhoods53
A new Bayesian Personalized Ranking algorithm based on multiple-layer neighborhoods (BPRN) is proposed, which distinctly alleviates the problems of data sparsity and cold-start users.
- SODE: Self-Adaptive One-Dependence Estimators for classification53
This paper proposes a self- Adaptive SPODEs, namely SODE, which uses immunity theory in artificial immune systems to automatically and self-adaptively select the weight for each single SPODE.
- Large language models for drug discovery and development52
This paper aims to provide a comprehensive overview for researchers and practitioners in computational biology, pharmacology, and AI4Science by offering insights into the potential transformative impact of LLMs on drug discovery and development.
- Foundation Models for Spatio-Temporal Data Science: A Tutorial and Survey52
A comprehensive review of Spatio-Temporal Foundation Models (STFMs) is provided, categorizing existing methodologies and identifying key research directions to advance ST general intelligence.
- CurvDrop: A Ricci Curvature Based Approach to Prevent Graph Neural Networks from Over-Smoothing and Over-Squashing52
A new Curvature-based topology-aware Dropout sampling technique named CurvDrop is proposed, in which the Discrete RicciCurvature is integrated into graph neural networks to enable more expressive graph models.
- 52
- 52
- 51
- 51
- Auto-HeG: Automated Graph Neural Network on Heterophilic Graphs50
This paper proposes a novel automated graph neural network on heterophilic graphs, namely Auto-HeG, to automatically buildheterophilic GNN models with expressive learning abilities, and incorporates heterophily into all stages of automatic heterophobic graph learning, including search space design, supernet training, and architecture selection.
- Dual Intent Enhanced Graph Neural Network for Session-based New Item Recommendation50
A dual-intent network to learn user intent from an attention mechanism and the distribution of historical data respectively is designed, which can simulate users’ decision-making process in interacting with a new item.
- 48
- TraverseNet: Unifying Space and Time in Message Passing for Traffic Forecasting48
This article proposes TraverseNet, a novel spatial–temporal graph neural network, viewing space and time as an inseparable whole, to mine spatial–temporal graphs while exploiting the evolving spatial–temporal dependencies for each node via message traverse mechanisms.
- Reinforcement Learning Based Meta-Path Discovery in Large-Scale Heterogeneous Information Networks48
This work presents a novel framework, Meta-path Discovery with Reinforcement Learning (MPDRL), to identify informative meta-paths from complex and large-scale HINs and proposes a novel multi-hop reasoning strategy in a reinforcement learning framework which aims to infer the next promising relation that links a source entity to a target entity.
- 48
- GADTI: Graph Autoencoder Approach for DTI Prediction From Heterogeneous Network46
A graph autoencoder approach for DTI prediction (GADTI) was proposed to discover potential interactions between drugs and targets using a heterogeneous network, which integrates diverse drug-related and target-related datasets.
- A Survey on Temporal Knowledge Graph Completion: Taxonomy, Progress, and Prospects45
A comprehensive review of TKGC methods and their details is provided, which focuses on continuous TKGs and predicts future events, and then classifies all extrapolation methods based on the algorithms they utilize.
- Graph Sequential Neural ODE Process for Link Prediction on Dynamic and Sparse Graphs45
GSNOP combines the advantage of the neural process and neural ordinary differential equation that models the link prediction on dynamic graphs as a dynamic-changing stochastic process and introduces the uncertainty into the predictions, making it generalize to more situations instead of overfitting to the sparse data.
- Predicting Human Mobility via Graph Convolutional Dual-attentive Networks45
A novel framework named GCDAN, which consists of two modules: spatio-temporal embedding and trajectory encoder-decoder, which employs a bidirectional diffusion graph convolution to preserve the spatial dependency in the location embedding.
- Graph Stochastic Neural Networks for Semi-supervised Learning45
A learnable graph neural network coupled with a high-dimensional latent variable to model the distribution of the classification function, and further adopt the amortised variational inference to approximate the intractable joint posterior for missing labels and the latent variable is introduced.
- Towards Neural Scaling Laws for Time Series Foundation Models44
This work examines two common TSFM architectures, encoder-only and decoder-only Transformers, and investigates their scaling behavior on both ID and OOD data, and compares the scaling properties across different architectures, showing that model architecture plays a significant role in scaling.
- Explore semantic topics and author communities for citation recommendation in bipartite bibliographic network44
A novel probabilistic topic model to automatically recommend citations for researchers is proposed, which considers not only text content similarity between papers but also community relevance among authors for effective citation recommendation.
- Heterogeneous Graph Attention Network for Small and Medium-Sized Enterprises Bankruptcy Prediction43
A heterogeneous-attention-network-based model (HAT) is proposed to facilitate SMEs bankruptcy prediction using publicly-accessible data to address graph heterogeneity, and extensive experiments in a real-world dataset demonstrate the attractiveness of the model compared with baselines.
- PREM: A Simple Yet Effective Approach for Node-Level Graph Anomaly Detection42
This work introduces a simple method termed PREprocessing and Matching, which eliminates the necessity for message-passing propagation during training, and employs a simple contrastive loss, leading to considerable reductions in training time and memory usage.
- Domain-Adversarial Graph Neural Networks for Text Classification42
An end-to-end, domain-adversarial graph neural networks (DAGNN), for cross-domain text classification, to model documents as graphs and use a domain- adversarial training principle to lean features from each graph (as well as learning the separation of domains) for effective text classification.
- 41
- 41
- Finding the best not the most: regularized loss minimization subgraph selection for graph classification41
A regularized loss minimization-driven feature selection method for graph classification which integrates subgraph selection and model learning into a unified framework to find discriminative subgraphs with guaranteed minimum loss w.r.t. the objective function is proposed.
- Exploring Relational Semantics for Inductive Knowledge Graph Completion40
A novel model called CFAG is proposed, which utilizes two granularity levels of relational semantics in a coarse-grained aggregator and a fine- grained generative adversarial net, for inductive KGC.
- Learning Graph Neural Networks with Positive and Unlabeled Nodes40
A novel GNN framework, long-short distance aggregation networks, is proposed, based on the adjacency matrix, to overcome limitations of existing GNN models, which only aggregate information from short distances in each round, and fail to capture long-distance relationship in graphs.
- Differentiable Neural Architecture Search in Equivalent Space with Exploration Enhancement39
A variational graph autoencoder is utilizes to injectively transform the discrete architecture space into an equivalently continuous latent space, to resolve the incongruence.
- 39
- Graph Learning under Distribution Shifts: A Comprehensive Survey on Domain Adaptation, Out-of-distribution, and Continual Learning38
This survey provides a comprehensive review and summary of the latest approaches, strategies, and insights that address distribution shifts within the context of graph learning and categorizes existing graph learning methods into several essential scenarios, including graph domain adaptation learning, graph out-of-distribution learning, and graph continual learning.
- Multi-Graph Learning with Positive and Unlabeled Bags38
A new multi-graph learning task with only positive and unlabeled bags, where labels are only available for bags but not for individual graphs inside the bag, is formulated and a puMGL learning framework is proposed which relies on two iteratively combined processes for multigraph learning.
- 37
- 37
- Integrating Graphs With Large Language Models: Methods and Prospects37
This article discusses and proposes open questions for integrating LLMs with graph-structured data for the future direction of the field and underscores the pivotal role of graphs in advancing LLMs.
- Optimizing OOD Detection in Molecular Graphs: A Novel Approach with Diffusion Models36
This work proposes to detect OOD molecules by adopting an auxiliary diffusion model-based framework, which compares similarities between input molecules and reconstructed graphs, which achieves more than 8% of average improvement in terms of detection AUC and AUPR accompanied by the reduced cost of testing time and memory consumption.
- A Probabilistic Graphical Model Based on Neural-symbolic Reasoning for Visual Relationship Detection35
- DACAD: Domain Adaptation Contrastive Learning for Anomaly Detection in Multivariate Time Series34
DACAD utilizes an anomaly injection mechanism that enhances generalization across unseen anomalous classes, improving adaptability and robustness and employs supervised contrastive loss for the source domain and self-supervised contrastive triplet loss for the target domain, ensuring comprehensive feature representation learning and domain-invariant feature extraction.
- Toward Graph Self-Supervised Learning With Contrastive Adjusted Zooming34
This work introduces a novel self-supervised GRL algorithm via graph contrastive adjusted zooming, namely, G-Zoom, to learn node representations by leveraging the proposedadjusted zooming scheme, and demonstrates that this proposed model outperforms the state-of-the-art methods consistently.
- Nonnegative Matrix Factorization in Dimensionality Reduction: A Survey33
A novel classification scheme for dimensionality reduction is proposed to enhance understanding of Nonnegative Matrix Factorization and highlight areas that need further exploration and development.
- 33
- 33
- 33
- 33
- Domain-adaptive message passing graph neural network32
A domain-adaptive message passing graph neural network (DM-GNN), which integrates graph neuralnetwork (GNN) with conditional adversarial domain adaptation, which is capable of learning informative representations for node classification that are also transferrable across networks.
- 32
- Projective Ranking: A Transferable Evasion Attack Method on Graph Neural Networks32
The idea is to learn a powerful attack strategy considering the long-term benefits of perturbations, then adjust it as little as possible to generate adversarial samples under different budgets, so the learned attack strategy has better attack performance.
- Projective Ranking-based GNN Evasion Attacks31
This work first formulate the perturbation space and proposes an evaluation framework and the projective ranking method, which aims to learn a powerful attack strategy then adapt it as little as possible to generate adversarial samples under dynamic budget settings.
- Gradformer: Graph Transformer with Exponential Decay30
Gradformer is presented, a method innovatively integrating GT with the intrinsic inductive bias by applying an exponential decay mask to the attention matrix, which enables Gradformer to retain its ability to capture information from distant nodes while focusing on the graph's local details.
- GNNEvaluator: Evaluating GNN Performance On Unseen Graphs Without Labels30
This paper proposes a two-stage GNN model evaluation framework, including (1) DiscGraph set construction and (2) GNNEvaluator training and inference, which exploits the outputs of GNNs related to latent node embeddings and node class predictions.
- Clustering social audiences in business information networks30
A novel machine-learning approach that co-clusters business information networks to discover and understand these audiences, called CBIN, which is based on an overlapping clustering paradigm with a hold-out strategy to discover the optimal number of clusters given the underlying data.
- Towards Data-centric Graph Machine Learning: Review and Outlook29
A systematic framework is introduced that encompasses all stages of the graph data lifecycle, including graph data collection, exploration, improvement, exploitation, and maintenance, and a thorough taxonomy of each stage is presented to answer three critical graph-centric questions.
- 29
- Going Deep: Graph Convolutional Ladder-Shape Networks29
The proposed graph convolutional ladder-shape networks (GCLN), a novel graph neural network architecture that transmits messages from shallow layers to deeper layers to overcome the over-smoothing problem and dramatically extend the scale of the neural networks with improved performance, is presented.
- 29
- Unifying Unsupervised Graph-Level Anomaly Detection and Out-of-Distribution Detection: A Benchmark28
A comprehensive evaluation framework that unifies GLAD and GLOD under the concept of generalized graph-level OOD detection is presented, a benchmark encompasses 35 datasets spanning four practical anomaly and OOD detection scenarios, facilitating the comparison of 18 representative GLAD/GLOD methods.
- An Inductive Reasoning Model based on Interpretable Logical Rules over temporal knowledge graph28
Experimental results demonstrate that ILR-IR outperforms state-of-the-art baselines, showcasing its superior performance in TKG extrapolation reasoning and demonstrates remarkable generalization capabilities, even when applied to related datasets that share a common relation vocabulary.
- Uncertainty-Aware Graph Neural Networks: A Multihop Evidence Fusion Approach27
A novel evidence-fusing graph neural network (EFGNN) is proposed to achieve trustworthy prediction, enhance node classification accuracy, and make explicit the risk of wrong predictions.
- Unleash LLMs Potential for Sequential Recommendation by Coordinating Dual Dynamic Index Mechanism27
This paper proposes the End-to-End Dual Dynamic (ED2) recommender, the first LLM-based sequential RS which adopts dual dynamic index mechanism, targeting resolving the above limitations simultaneously.
- Knowledge Graphs and Pretrained Language Models Enhanced Representation Learning for Conversational Recommender Systems27
The knowledge-enhanced entity representation learning (KERL) framework, which leverages both the KG and a pretrained language model (PLM) to improve the semantic understanding of entities for CRS, achieves state-of-the-art results in both recommendation and response generation tasks.
- CGMN: A Contrastive Graph Matching Network for Self-Supervised Graph Similarity Learning27
A contrastive graph matching network (CGMN) is proposed for self-supervised graph similarity learning in order to calculate the similarity between any two input graph objects and the experiment results show that the proposed new approach is superior to the state-of-the-art methods in graph similaritylearning downstream tasks.
- 27
- Multi-graph-view subgraph mining for graph classification27
This paper derives an evaluation criterion to estimate the discriminative power and redundancy of subgraph features across all views, with a branch-and-bound algorithm being proposed to prune subgraph search space.
- 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.
- 26
- Dynamic classifier ensemble for positive unlabeled text stream classification26
This paper addresses the problem of classifying positive and unlabeled text stream with various concept drift by constructing an appropriate validation set and designing a novel dynamic weighting scheme in the classification phase.
- DrugPilot: LLM-based Parameterized Reasoning Agent for Drug Discovery25
DrugPilot is introduced, a LLM-based agent system with a parameterized reasoning architecture designed for end-to-end scientific workflows in drug discovery and significantly outperforms state-of-the-art agents such as ReAct and LoT under the Berkeley function-calling benchmark.
- Graph Sparsification via Mixture of Graphs25
This paper introduces Mixture-of-Graphs (MoG), leveraging the concept of Mixture-of-Experts (MoE), to dynamically select tailored pruning solutions for each node, and incorporates multiple sparsifier experts, each characterized by unique sparsity levels and pruning criteria.
- IME: Integrating Multi-curvature Shared and Specific Embedding for Temporal Knowledge Graph Completion25
A novel Integrating Multi-curvature shared and specific Embedding (IME) model for TKGC tasks, which models TKGs into multi-curvature spaces, including hyperspherical, hyperbolic, and Euclidean spaces and incorporates two key properties, namely space-shared property and space-specific property.
- Contrastive Graph Similarity Networks25
A comprehensive series of experiments indicate that CGSim outperforms state-of-the-art baselines on six datasets and significantly reduces the computational cost, which demonstrates the CGSim model’s superiority over other baselines.
- A Hybrid User Experience Evaluation Method for Mobile Games25
In order to reveal distinct relations among UX factors, sufficient user characteristics and categories of game events, an improved RIPPER algorithm is proposed by this paper to obtain the relations.
- 24
- 24
- Detecting Suicidal Ideation with Data Protection in Online Communities24
This work proposes a novel data-protecting solution to learn a model to train a local data-preserving model for each user which only shares their own model’s parameters with the server rather than their personal information.
- Augmented Commonsense Knowledge for Remote Object Grounding23
This work proposes an augmented commonsense knowledge model (ACK), which consists of knowledge graph-aware cross-modal and concept aggregation modules to enhance visual representation and visual-textual data alignment by integrating visible objects, commonsense knowledge, and concept history, which includes object and knowledge temporal information.
- 23
- ATTIC is an integrated approach for predicting A-to-I RNA editing sites in three species23
A novel stacked-ensemble learning model, ATTIC (A-To-I ediTing predICtor), is developed to accurately identify A-to-I editing sites across three species, including Homo sapiens, Mus musculus and Drosophila melanogaster, and is utilized as a useful tool to accelerate the identification of A- to-I RNA editing events and help characterize their roles in post-transcriptional regulation.
- MAMDR: A Model Agnostic Learning Framework for Multi-Domain Recommendation23
This paper presents a scalable MDR platform served in Taobao that enables to provide services for thousands of domains without specialists involved and proposes a novel model agnostic learning framework, namely MAMDR, for the multi-domain recommendation.
- GPTBIAS: A Comprehensive Framework for Evaluating Bias in Large Language Models23
This work proposes a bias evaluation framework named GPTBIAS that leverages the high performance of LLMs (e.g., GPT-4 \cite{openai2023gpt4}) to assess bias in models and introduces prompts specifically designed for evaluating model bias.
- BaLeNAS: Differentiable Architecture Search via the Bayesian Learning Rule23
This paper formulates the neural architecture search as a distribution learning problem through relaxing the architecture weights into Gaussian distributions through leveraging the natural-gradient variational inference (NGVI) and demonstrates how the differentiable NAS benefits from Bayesian principles, enhancing exploration and improving stability.
- Knowledge graph representation and reasoning23
This work states that efficient representation learning and reasoning can be one of the paths towards the emulation of high-level cognition and human-level intelligence in AI.
- Prediction of circRNA‐miRNA Associations Based on Network Embedding23
A new computational method, NECMA, based on network embedding to predict potential associations between circRNAs and miRNAs, which has better prediction accuracy than other state-of-the-art methods.
- Adaptive knowledge subgraph ensemble for robust and trustworthy knowledge graph completion23
Experimental results show that the robustness of the ensemble framework outperforms exiting knowledge graph embedding approaches on manually injected noise as well as inherent noisy extracted KGs.
- Unraveling Privacy Risks of Individual Fairness in Graph Neural Networks22
This work contributes to the comprehensively developing trustworthy GNNs by simultaneously addressing both fairness and privacy aspects by carefully introducing fairness-aware loss reweighting based on in-fluence function and privacy-aware graph structure perturbation modules within a fine-tuning mechanism.
- 22
- Graph Convolutional Incomplete Multi-modal Hashing22
The proposed Graph Convolutional Incomplete Multi-modal Hashing (GCIMH) can handle missing modalities in both offline training and online query stages and is superior over the state-of-the-arts on both complete and incomplete multi- modal retrieval.
- 22
- 22
- GSSNN: Graph Smoothing Splines Neural Networks22
This paper develops an effective feature smoothing and enhancement module Scaled Smoothing Splines (S3) to learn graph embedding, and designs a novel scoring module, which exploits closeness, degree, as well as self-attention values, to select important node features as knots for smoothing splines.
- Long-short Distance Aggregation Networks for Positive Unlabeled Graph Learning22
This paper presents a novel long-short distance aggregation networks (\textttLSDAN) for positive unlabeled (PU) graph learning, to generate multiple graphs at different distances based on the adjacency matrix, and further develop a long- short distance attention model for these graphs.
- Measuring distance-based semantic similarity using meronymy and hyponymy relations22
A novel model to calculate the path distance between concepts is proposed, and a similarity measure is proposed which nonlinearly transforms the distance to semantic similarity, which indicates the effectiveness of the distance model.
- 21
- Explainable Hyperbolic Temporal Point Process for User-Item Interaction Sequence Generation21
An Explainable Hyperbolic Temporal Point Process for User-Item Interaction Sequence Generation (EHTPP) is proposed, which regards each user-item interaction as an event in hyperbolic space and employs a temporal point process framework to model the probability of event occurrence.
- 21
- Leveraging Information Bottleneck for Scientific Document Summarization21
This paper presents an unsupervised extractive approach to summarize scientific long documents based on the Information Bottleneck principle with two separate steps that can be flexibly extended to a multi-view framework by different signals.
- 21
- Low-Bit Quantization for Attributed Network Representation Learning21
A new Low-Bit Quantization for Attributed Network Representation Learning model (LQANR) that can learn compact node representations with low bitwidth values while preserving high representation accuracy is presented.
- Towards large-scale social networks with online diffusion provenance detection21
This work proposes an online regression model for real-time diffusion provenance identification of large networks that first uses offline collected network cascades to infer the edge transmission weights, and then uses an online l 1 non-convex regression model as the identification model.
- Noise-Resilient Unsupervised Graph Representation Learning via Multi-Hop Feature Quality Estimation20
A Gaussian model is introduced that utilizes a learnable "meta-representation" as a condition to estimate the expectation and variance of multi-hop propagated features via neural networks via neural networks, which captures the semantic and structural information underlying multiple propagated features but is naturally less susceptible to interference by noise.
- Divide and Denoise: Empowering Simple Models for Robust Semi-Supervised Node Classification against Label Noise20
A label-noise robust GNN that discards the coupled message-passing scheme and proposes a novel reliable graph pseudo labeling algorithm that can effectively leverage the knowledge of unlabeled nodes while mitigating the adverse effects of noisy pseudo labels.
- 20
- 20
- Unifying Graph Contrastive Learning with Flexible Contextual Scopes20
The architecture of UGCL can be considered as a general framework to unify existing GCL methods and builds flexible contextual representations with tunable contextual scopes by controlling the power of an adjacency matrix.
- PUMPNET: a deep learning approach to pump operation detection20
The proposed PUMPNET algorithm could identify pool pump operation status with high accuracy in the low-frequency sampling scenario for thousands of household, compared to traditional NILM algorithms which process high sampling rate data and can only apply to limited number of households.
- Network-based methods for gene function prediction20
A survey for the state of the art of advanced methods of network-based gene function prediction and discusses the potential challenges is provided.
- 20
- Heterogeneous Information Network Embedding based Personalized Query-Focused Astronomy Reference Paper Recommendation20
This paper studies the problem of personalized query-focused astronomy reference paper recommendation and proposes a heterogeneous information network embedding based recommendation approach, which simultaneously captures intra-relationship among homogeneous vertices, interrelationships among heterogeneous vertice and correlations between vertices and text contents.
- 20
- Classifier Ensemble for Uncertain Data Stream Classification20
This paper proposes two types of ensemble based algorithms, Static Classifier Ensemble (SCE) and Dynamic Classifier ensemble (DCE) for mining uncertain data streams and experimental results reveal that DCE algorithm outperforms SCE algorithm.
- Graph-Augmented Large Language Model Agents: Current Progress and Future Prospects19
Categorize existing GLA methods by their primary functions in LLM agent systems, including planning, memory, and tool usage, and then analyze how graphs and graph learning algorithms contribute to each, to highlight key future directions to advance this field.
- Towards Expressive Spectral-Temporal Graph Neural Networks for Time Series Forecasting19
A theoretical framework is established that unravels the expressive power of spectral-temporal GNNs and proposes a simple instantiation named Temporal Graph Gegenbauer Convolution (TGGC), which significantly outperforms most existing models with only linear components and shows better model efficiency.
- Online GNN Evaluation Under Test-time Graph Distribution Shifts19
An effective learning behavior discrepancy score is developed, dubbed LeBeD, to estimate the test-time generalization errors of well-trained GNN models, which enables the effective evaluation of the well-trained GNNs' ability to capture test node semantics and structural representations, making it an expressive metric for estimating the generalization error in online GNN evaluation.
- 19
- Sign is Not a Remedy: Multiset-to-Multiset Message Passing for Learning on Heterophilic Graphs19
This work identifies two limitations of SMP: undesirable representation update for multi-hop neighbors and vulnerability against oversmoothing issues and proposes a novel message passing function called Multiset to Multiset GNN(M2M-GNN).
- 19
- Classifying networked text data with positive and unlabeled examples19
A novel learning algorithm, namely puNet (positive and unlabeled learning algorithm for Networked text data), is proposed, for efficiently classifying networked text, even if training datasets contain only a small amount of positive examples and a large amount of unlabeling ones.
- 18
- 18
- 18
- Robust Graph Representation Learning for Local Corruption Recovery18
This work establishes a graph learning scheme that automatically detects (locally) corrupted feature attributes and recovers robust embedding for prediction tasks and can recover a robust graph representation from black-box poisoning and achieve excellent performance.
- 18
- 18
- Fast Heterogeneous Federated Learning with Hybrid Client Selection18
A novel clustering-based client selection scheme to accelerate the FL convergence by variance reduction and the tighter convergence guarantee of the proposed method thanks to the variance reduction is presented.
- 18
- 18
- Low-Rank and Sparse Matrix Factorization for Scientific Paper Recommendation in Heterogeneous Network18
The proposed method seamlessly combines low-rank and sparse matrix factorization method with fine-grained paper and author affinity matrixes that are extracted from heterogeneous scientific network and can effectively alleviate the sparsity and cold start problems that exist in traditional Matrix factorization based collaborative filtering methods.
- Balancing user preferences by social networks: A condition-guided social recommendation model for mitigating popularity bias17
A Condition-Guided Social Recommendation Model is proposed to mitigate the model's popularity bias by denoising the social network and adjusting the weights of user's social preferences to counteract the popularity bias present in the recommendation model.
- The Heterophilic Snowflake Hypothesis: Training and Empowering GNNs for Heterophilic Graphs17
The prevailing concept of "one node one receptive field" to the heterophilic graph is transferred, or the first time, and a proxy label predictor is constructed, enabling each node to possess a latent prediction distribution, which assists connected nodes in determining whether they should aggregate their associated neighbors.
- Large Language Model Watermark Stealing With Mixed Integer Programming17
A novel green list stealing attack against the state-of-the-art LLM watermark scheme and systematically examine its vulnerability to this attack, formalized as a mixed integer programming problem with constraints.
- 17
- Advancing mRNA subcellular localization prediction with graph neural network and RNA structure17
Allocator is proposed, a multi-view parallel deep learning framework that seamlessly integrates the RNA sequence-level and structure-level information, enhancing the prediction of mRNA multi-localization.
- Deep transformer-based heterogeneous spatiotemporal graph learning for geographical traffic forecasting17
This model incorporates a temporal transformer that captures long-term temporal patterns in traffic data without simple data fusion and introduces adaptive normalized graph structures within different graph layers, enabling the model to capture dynamic spatial dependencies and adapt to diverse traffic scenarios.
- G2Pxy: Generative Open-Set Node Classification on Graphs with Proxy Unknowns17
A novel generative open-set node classification method, i.e., G2Pxy, which follows a stricter inductive learning setting where no information about unknown classes is available during training and validation and achieves superior effectiveness for unknown class detection and known class classification.
- Learning multi-level weight-centric features for few-shot learning17
A multi-level weight-centric feature learning to give full play to feature extractor's dual roles in few-shot learning and significantly outperforms its counterparts in both standard and generalized settings and using different network backbones.
- Grounding Visual Concepts for Zero-Shot Event Detection and Event Captioning17
This work is the first time to define and solve the MEC task, which is a further step towards understanding video events, and achieves state-of-the-art performance on the TRECVID MEDTest dataset, as well as the newly proposed TREC VID-MEC dataset.
- GraphRAG-Induced Dual Knowledge Structure Graphs for Personalized Learning Path Recommendation16
A novel approach to learning path recommendation is proposed, GraphRAG-Induced Dual Knowledge Structure Graphs for Personalized Learning Path Recommendation (KnowLP), which enhances learning path recommendations by incorporating both prerequisite and similarity relations between knowledge concepts.
- 16
- 16
- Two Heads Are Better Than One: Boosting Graph Sparse Training via Semantic and Topological Awareness16
This work proposes a new research line and concept termed Graph Sparse Training (GST), which dynamically manipulates sparsity at the data level, and introduces the Equilibria Sparsification Principle to guide this process.
- 16
- 16
- Monash-Summ@LongSumm 20 SciSummPip: An Unsupervised Scientific Paper Summarization Pipeline16
SciSummPip, an unsupervised text summarization system for multi-document in News domain inspired by SummP Pip, that includes a transformer-based language model SciBERT for contextual sentence representation and content selection and a summary length constraint is applied to adapt to the scientific domain.
- Boosting for graph classification with universum16
A new universum graph classification framework which leverages additional “non-example” graphs to help improve the graph classification accuracy and proposes a mathematical programming algorithm, ugBoost, which integrates discriminative subgraph selection and margin maximization into a unified framework to fully exploit the universum.
- OFA-MAS: One-for-All Multi-Agent System Topology Design based on Mixture-of-Experts Graph Generative Models15
OFA-MAS is proposed, a one-for-all framework that generates adaptive collaboration graphs for any task described in natural language through a single universal model, and Experiments show that OFA-MAS significantly outperforms specialized one-for-one models, generating highly adaptive MAS topologies.
- 15
- Securing Graph Neural Networks in MLaaS: A Comprehensive Realization of Query-based Integrity Verification15
This research introduces a groundbreaking approach to protect GNN models in MLaaS from model-centric attacks, and proposes a query-based verification technique, fortified with innovative node fingerprint generation algorithms.
- Towards complex dynamic physics system simulation with graph neural ordinary equations15
GNSTODE is proposed - Graph Networks with Spatial-Temporal neural Ordinary Differential Equations (GNSTODE) that characterizes the varying spatial and temporal dependencies in particle systems using a united end-to-end framework and can serve as an effective tool for particle simulation in real-world applications.
- 15
- OpenWGL: open-world graph learning for unseen class node classification15
This paper proposes an uncertain node representation learning principle to use multiple versions of node feature representation to test a classifier’s response on a node, through which it can differentiate whether a node belongs to the unseen class.
- 15
- FreeGAD: A Training-Free yet Effective Approach for Graph Anomaly Detection14
FreeGAD is proposed, a novel training-free yet effective GAD method that leverages an affinity-gated residual encoder to generate anomaly-aware representations and achieves superior anomaly detection performance, efficiency, and scalability on multiple benchmark datasets from diverse domains, without any training or iterative optimization.
- Domain-Adaptive Graph Attention-Supervised Network for Cross-Network Edge Classification14
This work proposes a novel domain-adaptive graph attention-supervised network (DGASN) to effectively tackle the CNHHEC problem and applies direct supervision on graph attention learning based on the observed edge labels from the source network, thus lowering the negative effects of heterophilous edges while enlarging the positive effects of homophILous edges during neighborhood aggregation.
- Compatible Transformer for Irregularly Sampled Multivariate Time Series14
The proposed CoFormer is a transformer-based encoder to achieve comprehensive temporal-interaction feature learning for each individual sample in irregular multivariate time series, and can analyze irregularly sampled multivariate time series for many downstream tasks, including classification and prediction.
- Gated relational stacked denoising autoencoder with localized author embedding for global citation recommendation14
A novel neural network based model, called gated relational probabilistic stacked denoising autoencoder with localized author (GRSLA) embedding, for global citation recommendation task, comprised of two modules with different neural network architecture.
- 14
- 13
- DyVGRNN: DYnamic mixture Variational Graph Recurrent Neural Networks13
A novel integrated variational framework called DYnamic mixture Variational Graph Recurrent Neural Networks (DyVGRNN), which consists of extra latent random variables in structural and temporal modelling, which greatly outperforms state-of-the-art dynamic graph representation learning methods in terms of link prediction and clustering.
- Digerati – A multipath parallel hybrid deep learning framework for the identification of mycobacterial PE/PPE proteins13
The first deep learning-based approach, termed Digerati, for the rapid and accurate identification of PE and PPE family proteins, built upon a multipath parallel hybrid deep learning framework, which equips multi-layer convolutional neural networks with bidirectional, long short-term memory to effectively learn the higher-order feature representations of PE/PPE proteins.
- 13
- 13
- Multi-graph-view learning for complicated object classification13
A multi-graph-view bag learning algorithm (MGVBL), which aims to explore subgraph features from multiple graphviews for learning by enabling a joint regularization across multiple graph-views, and enforcing labeling constraints at the bag and graph levels, is able to discover most effective sub graph features across all graph- views for learning.
- Transforming Urban Dynamics: Harnessing Large Language Models for Smarter Mobility12
This article provides an overview of recent efforts and aims to shed light on the challenges and future opportunities to facilitate the adaptation of large language models for smarter mobility systems.
- 12
- 12
- 12
- Online unsupervised cross-view discrete hashing for large-scale retrieval12
A new online unsupervised cross-view discrete hashing method that considers similarity preservation and quantization loss is proposed, dubbed OUCDH that yields superior performance than existing state-of-the-art online cross-View hashing methods.
- 12
- 12
- From Few-Shot to Zero-Shot: Towards Generalist Graph Anomaly Detection11
This work proposes a novel generalist GAD paradigm that aims to develop a unified model capable of detecting anomalies on multiple unseen datasets without retraining/fine-tuning or customization and introduces ARC, a few-shot generalist GAD method with three key designs.
- Robust Graph Based Social Recommendation Through Contrastive Multi-View Learning11
This work introduces a novel framework named Robust Graph Based Social Recommendation through Contrastive Multi-view Learning (RGCML), which leverages denoised social relations and global intents as dual auxiliary information sources to provide comprehensive characterization of users.
- Few-Shot Learning on Graphs: From Meta-Learning to LLM-empowered Pre-Training and Beyond11
This tutorial offers an in-depth exploration of recent advancements in few-shot learning for graphs, providing a comparative analysis of state-of-the-art methods and identifying future research directions.
- Effective Instruction Parsing Plugin for Complex Logical Query Answering on Knowledge Graphs11
An effective Query Instruction Parsing Plugin (QIPP) that leverages the context awareness of Pre-trained Language Models (PLMs) to capture latent query patterns from code-like query instructions and a query-guided instruction decoder to adapt query patterns to KGQE models is proposed.
- 11
- 11
- 11
- Exploring Features for Complicated Objects: Cross-View Feature Selection for Multi-Instance Learning11
A new cross-view feature selection problem is formulated that aims to identify the most representative features across all feature views for MIL, and the solution to the objective function will ensure that the identified top-m features are the most informative ones across allfeature views.
- Continuous top-k query for graph streams11
A novel algorithm is proposed, Hoe-PGPL, to identify top-k correlated graphs from data stream, by using a sliding window which covers a number of consecutive batches of stream data records, and uses two level candidate checking to accurately estimate the correlation of the emerging candidate patterns, without rechecking the historical stream data.
- CGStream11
This paper proposes a novel algorithm, CGStream, to identify correlated graphs from data stream, by using a sliding window which covers a number of consecutive batches of stream data records, and derives an upper correlation bound and a heuristic rule to prune the candidate size which helps reduce the computation cost at each outlook.
- Multi-Modal Multi-Kernel Graph Learning for Autism Prediction and Biomarker Discovery10
Graph learning-based multi-modal integration and classification is one of the most challenging tasks for disease prediction and discriminative brain regions associated with autism are identified by the model, providing guidance for the study of autism pathology.
- 10
- Revisiting Edge Perturbation for Graph Neural Network in Graph Data Augmentation and Attack10
This paper proposes a unified formulation and establishes a quantizable boundary between two categories of edge perturbation methods and devise Edge Priority Detector (EPD) to generate a novel priority metric, bridging these methods up in the workflow.
- Out-of-Distribution Detection on Graphs: A Survey10
A rigorous definition of GOOD detection is provided and existing methods are systematically categorized into four types: enhancement-based, reconstruction-based, information propagation-based, and classification-based approaches.
- ConTIG: Continuous representation learning on temporal interaction graphs10
A two-module framework named ConTIG is proposed, a novel representation learning method on TIG that captures the continuous dynamic evolution of node embedding trajectories and introduces a self-attention mechanism to predict future node embeddings by aggregating historical temporal interaction information.
- Community Preserving Social Recommendation with Cyclic Transfer Learning10
This work proposes a community-preserving recommendation framework with cyclic transfer learning, incorporating heterogeneous community influence into the rating domain, and analyzes the characteristics of the community domain and its inter-influence on the ratingdomain, and construct link constraints and preference constraints in the communitydomain.
- Fire Burns, Sword Cuts: Commonsense Inductive Bias for Exploration in Text-based Games10
CommExpl, an exploration technique that injects external commonsense knowledge, via a pretrained language model (LM), into the agent during training when the agent is the most uncertain about its next action, is proposed.
- 10
- Multi-document summarization based on sentence cluster using non-negative matrix factorization10
The proposed framework employs NMTF to cluster sentences using inter-type relationships among documents, sentences and terms, and incorporate the intra-type information through manifold regularization.
- Raising the Bar in Graph OOD Generalization: Invariant Learning Beyond Explicit Environment Modeling9
A novel method, derived from the theoretical framework of GIL, that achieves state-of-the-art performance, significantly outperforming existing methods across graph data from various domains and with different distribution shifts.
- CONC: Complex-noise-resistant Open-set Node Classification with Adaptive Noise Detection9
A trustworthiness learner is adopted to learn the trustworthiness rates of the feature and label for each node while a decoder and an open-set classifier are trained to reconstruct the structure of a node and to predict its category simultaneously with the guidance of node trustworthiness.
- Cost-effective Data Labelling for Graph Neural Networks9
An unsupervised, scalable and flexible AL method that incurs low memory footprints and time cost, is flexible to the choice of underlying GNNs, and operates without requiring GNN-model-specific knowledge or labels of selected nodes is proposed.
- Maximizing Malicious Influence in Node Injection Attack9
This paper proposes a global node injection attack framework, MaxiMal, to maximize malicious information under a strict black-box setting and introduces a susceptible-reverse influence sampling strategy to select neighbor nodes that are able to spread malicious information widely.
- 9
- Geometric Relational Embeddings: A Survey9
This work surveys methods that underly geometric relational embeddings and categorizes them based on (i) the embedding geometries that are used to represent the data; and (ii) the relational reasoning tasks that they aim to improve.
- Cyclic label propagation for graph semi-supervised learning9
This work introduces a novel framework for graph semi-supervised learning termed as Cyclic Label Propagation (CycProp for abbreviation), which integrates GNNs into the process of label propagation in a cyclic and mutually reinforcing manner to exploit the advantages of both GNN and LPA.
- Label Embedding with Partial Heterogeneous Contexts9
A general Partial Heterogeneous Context Label Embedding (PHCLE) framework, which overcomes the partial context problem and can nicely incorporate more contexts, which both cannot be tackled with existing multi-context label embedding methods.
- Query-oriented citation recommendation based on network correlation9
It is argued that combining different kinds of paper’s scholarly roles can enhance citation recommendation performance, and a network correlation based query-oriented citation recommendation approach is proposed.
- 8
- 8
- 8
- RHGNN: imposing relational inductive bias for heterogeneous graph neural network8
A novel heterogeneous graph neural network called RHGNN is proposed, which overcomes the reliance on meta-paths by modeling the heterogeneous graph as a whole instead of splitting it, and jointly obtains personalized feature representations for nodes and edges.
- 8
- 8
- 8
- 8
- 8
- 8
- GenIAS: Generator for Instantiating Anomalies in time Series7
This work proposes a synthetic anomaly generation method named Generator for Instantiating Anomalies in Time Series (GenIAS), which generates realistic and diverse anomalies via a novel learnable perturbation in the latent space of a variational autoencoder.
- 7
- A Cross-Field Fusion Strategy for Drug-Target Interaction Prediction7
This work proposes the siamese drug-target interaction SiamDTI prediction method, which utilizes a double channel network structure for cross-field supervised learning and achieves higher accuracy levels than other state-of-the-art (SOTA) methods on novel drugs and targets.
- 7
- 7
- Universal network representation for heterogeneous information networks7
A universal network representation approach (UNRA), that represents different types of nodes in heterogeneous information networks in a continuous and common vector space, and achieves outstanding performance, compared to six other state-of-the-art algorithms.
- 6
- Towards One-for-All Anomaly Detection for Tabular Data6
OFA-TAD is proposed, a generalist one-for-all (OFA) TAD framework that only requires one-time training on multiple source datasets and can generalize to unseen datasets from diverse domains on-the-fly.
- 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.
- DA-MoE: Addressing depth-sensitivity in graph-level analysis through mixture of experts6
The depth adaptive mixture of expert (DA-MoE) method is proposed, which incorporates two main improvements to GNN backbone: 1) DA-MoE employs different GNN layers, each considered an expert with its own parameters, effectively addressing the depth-sensitivity issue in graph data.
- 6
- 6
- GatorCLR: Personalized predictions of patient outcomes on electronic health records using self-supervised contrastive graph representation6
This work investigates and design self-supervised learning (SSL) paradigms to learn high-quality representations from longitudinal EHR data, aiming to effectively capture longitudinal relationships and patterns for improved patient outcome predictions.
- Unified Knowledge-Guided Molecular Graph Encoder with multimodal fusion and multi-task learning6
The Unified Knowledge-Guided Molecular Graph Encoder (UKGE) is presented, a groundbreaking framework that leverages heterogeneous graphs to unify the representation of diverse molecular modalities, enabling comprehensive and unified molecular representations.
- 6
- Towards Model Extraction Attacks in GAN-Based Image Translation via Domain Shift Mitigation6
Diverging from the traditional approach of bridging the distribution gap between attacker queries and victim training samples, this paper opts to mitigate the effect caused by the different distributions, known as the domain shift, and introduces a new regularization term that penalizes high-frequency noise.
- Client-side Gradient Inversion Against Federated Learning from Poisoning6
This work proposes Client-side poisoning Gradient Inversion (CGI), which is a novel attack method that can be launched from clients and shows the feasibility of a client-side adversary with limited knowledge being able to recover the training samples from the aggregated global model.
- 6
- 6
- Graph Ladder Networks for Network Classification6
This paper proposes an effective deep learning model, Graph Ladder Networks (GLN), for node classification in networks which unifies the representation learning and network classification into one single framework by exploiting both labeled and unlabeled nodes in a network.
- 5
- 5
- 5
- Not All Edges are Equally Robust: Evaluating the Robustness of Ranking-Based Federated Learning5
A novel local model poisoning attack against FRL is introduced, namely Vulnerable Edge Manipulation (VEM) attack, which focuses on identifying and perturbing the most vulnerable edges in each layer and leveraging an optimization-based approach to maximize the attack's impact.
- 5
- Temporal dynamics unleashed: Elevating variational graph attention5
Empirical tests underscore VarGATDyn’s dominance in dynamic link prediction across various datasets, highlighting its proficiency in capturing multimodal distributions and temporal dynamics.
- A Novel Neural-symbolic System under Statistical Relational Learning5
This work proposes a neural-symbolic framework based on statistical relational learning that effectively integrates deep learning models with symbolic reasoning in a mutually beneficial manner and introduces a quantitative strategy to evaluate the interpretability of the model's predictions and providing insights into the reasoning process.
- 5
- Toward the Automated Construction of Probabilistic Knowledge Graphs for the Maritime Domain5
The current implementation of Maritime DeepDive is reported on, together with preliminary results on extracting probabilistic events from maritime piracy incidents, and this pipeline was evaluated on a manually crafted gold standard, yielding promising results.
- Exploring Gradient-Guided Masked Language Model to Detect Textual Adversarial Attacks4
Gradient-guided MLMD is introduced, which leverages gradient information to identify and skip non-keywords during detection, significantly reducing resource consumption without compromising detection performance.
- Unsupervised Deep Graph Structure and Embedding Learning4
The proposed UPGNN learns graph structure and embedding without label supervision, and thus can be applied various downstream tasks and reduce computational complexity, providing a efficient alternative to UPGNN.
- Complex query answering over knowledge graphs foundation model using region embeddings on a lie group4
A Lie group is introduced as a compact embedding space for complex query embedding, enhancing ability to handle the intricacies of knowledge graphs the foundation model and complex query processing.
- 4
- VR-GNN: variational relation vector graph neural network for modeling homophily and heterophily4
This work proposes a novel GNN model based on relation vector translation named as Variational Relation Vector Graph Neural Network (VR-GNN), which gains consistent and significant improvements against existing strong GNN methods under heterophily and competitive performance under homophily.
- 4
- Guest Editorial: Graph-powered machine learning in future-generation computing systems4
A set of research topics such as knowledge graph reasoning, graph self-supervised learning, temporal graph modeling, and graph embedding techniques have emerged, and applications such as graph-based anomaly detection, social recommendation, social influence analytics are becoming important issues for the research community.
- 4
- 4
- 4
- IEEE Access Special Section Editorial: Advanced Data Mining Methods for Social Computing4
This IEEE ACCESS Special Section on advanced data mining methods for social computing aims at bringing together researchers to disseminate their findings in the quickly growing field of social computing.
- Locally Weighted Learning: How and When Does it Work in Bayesian Networks?4
Experiments and comparisons demonstrate that locally weighting technologies just slightly outperforms unweighted complex BNCs on ACC and AUC, which means that although locally weighting could significantly improve the performance of NB, it could not work well on BNCs with complex structures.
- 3
- Personalized Multi-Interest Modeling for Cross-Domain Recommendation to Cold-Start Users3
This work proposes a personalized multi-interest modeling framework for CDR to cold-start users, termed as NF-NPCDR, and introduces a stochastic adaptive decoder to incorporate both the personalized and common preference for cold-start users, adaptively modulating both preference for better recommendation.
- 3
- Mosaic: Client-driven Account Allocation Framework in Sharded Blockchains3
Pilotto significantly enhances computational efficiency, achieving a four-order-of-magnitude reduction in computation time, with the reduced input data size from 1.44 GB to an average of 228.66 bytes per account.
- Adversarial Contrastive Graph Masked AutoEncoder Against Graph Structure and Feature Dual Attacks3
The proposed Adversarial Contrastive Graph Masked AutoEncoder (ACGMAE) to defend against graph structure and feature dual attacks and demonstrates the effectiveness of the proposed ACGMAE especially under graph structure and feature dual attacks.
- Regressor-free Molecule Generation to Support Drug Response Prediction3
Regressor-free guidance molecule generation is introduced to ensure sampling within a more effective space and support DRP, and a common-sense numerical knowledge graph is designed that constrains the order of text representations.
- All Nodes are created Not Equal: Node-Specific Layer Aggregation and Filtration for GNN3
A Node-Specific Layer Aggregation and Filtration architecture, termed NoSAF, a framework capable of filtering and processing information from each individual nodes, aiming to provide a reliable information filter for each layer's nodes to sieve out information beneficial for the subsequent layer.
- 3
- 3
- 3
- Thrifty Neural Architecture Search for Medical Image Segmentation (Student Abstract)3
A novel approach, dubbed Thrifty NAS, to design computation and memory-efficient models for medical image segmentation automatically, with much fewer parameters while retaining competitive performance is presented.
- 3
- 2
- 2
- 2
- Beyond Pairwise Interactions: Equivariant Hypergraph Diffusion for Crystal Structure Prediction2
The Equivariant Hypergraph Diffusion Model (EH-Diff), a generative framework designed to exploit the symmetry-preserving properties of hypergraphs, is proposed, which provides an efficient and accurate method for predicting crystal structures, with rigorous theoretical guarantees on invariance preservation.
- Graph Stochastic Neural Process for Inductive Few-shot Knowledge Graph Completion2
This work casts I-FKGC as an inductive reasoning problem, and proposes a novel Graph Stochastic Neural Process (GS-NP) approach, which consists of two major modules, which consists of a neural process-based hypothesis extractor and a graph stochastic attention-based predictor.
- 2
- 2
- Domain-Level Disentanglement Framework Based on Information Enhancement for Cross-Domain Cold-Start Recommendation2
A generalized framework that Domain-level Disentanglement framework based on information enhancement for Cross-domain Cold-start Recommendation, which achieves deep utilization of domain-level information through independent extraction of domain knowledge and fusion using heuristic strategies.
- 2
- 2
- Knowledge-aware contrastive heterogeneous molecular graph learning2
KCHML conceptualizes molecules through three distinct graph views—molecular, elemental, and pharmacological—enhanced by heterogeneous molecular graphs and a dual message-passing mechanism, offering a comprehensive representation for property prediction, as well as for downstream tasks such as drug-drug interaction prediction.
- Similarity Preserving Transformer Cross-Modal Hashing for Video-Text Retrieval2
The proposed Similarity Preserving Transformer Cross-Modal Hashing (SPTCH), a new unsupervised deep cross-modal hashing method for video-text retrieval, outperforms state-of-the-arts in video-text retrieval.
- 2
- Decision-focused Graph Neural Networks for Combinatorial Optimization2
This work designs an end-to-end framework that utilizes GNNs to address CO problems with auxiliary support and introduces a decision-focused framework that utilizes GNNs to address CO problems with auxiliary support.
- 2
- Geometry Contrastive Learning on Heterogeneous Graphs2
Geometry Contrastive Learning (GCL) views a heterogeneous graph from Euclidean and hyperbolic perspective simultaneously, aiming to make a strong merger of the ability of modeling rich semantics and complex structures, which is expected to bring in more benefits for downstream tasks.
- 2
- 2
- 2
- 2
- 2
- Decentralized Learning with Average Difference Aggregation for Proactive Online Social Care.2
A decentralized learning framework to enable proactive social care equipped with artificial intelligence and data protection, and a novel model aggregation algorithm, namely the average difference descent aggregation (AvgDiffAgg for short).
- 2
- MolX: A Geometric Foundation Model for Protein-Ligand Modelling1
MolX is a Graph Transformer foundation model that jointly learns geometric and chemical representations of protein pockets and ligands from large-scale 3D structural data, providing a unified framework for predicting and interpreting complex small-molecule-protein interactions.
- Graph-to-Tree: Topological Decomposition for Self-Supervised Learning1
Topological Decomposition for Self-supervised Learning (TopDSL), a framework that injects multi-scale signals into graph representation learning with superior performance over state-of-the-art baselines on various graph learning benchmarks.
- CAFE: Cross-View Adaptive Fusion and Cluster Center Enhancement for Robust Multi-View Clustering1
This work designs a cross-view adaptive fusion module that incorporates dual weights at both the view level and the sample level, enabling effective coordination of consistent and complementary information across views, and proposes a dual-driven cluster center enhancement framework to refine cluster structures.
- negMIX: Negative Mixup for OOD Generalization in Open-Set Node Classification1
A novel negative Mixup method purposefully crafted for the open-set scenario with theoretical justification is devised, to enhance the model's generalization to OOD nodes and yield clearer ID/OOD boundary.
- 1
- 1
- Multi-view contrastive learning with Static attributes and Dynamic interests for Sequential Recommendation1
By leveraging contrastive learning, SDSR aligns representations across multiple data views—such as interaction graphs, knowledge graphs, and sequential data—creating a unified user-item model that bridges long-term preferences with short-term trends, and ensures consistency across various representations.
- 1
- Sharpness-aware Zeroth-order Optimization for Graph Transformers1
A Sharpness-aware Zeroth-order Optimizer (SZO) that combines Sharpness-Aware Minimization (SAM) technique facilitating convergence within a flatter neighborhood, and leverages parallel computing for efficient gradient estimation is proposed.
- $\text{M}^{2}$LLM: Multi-view Molecular Representation Learning with Large Language Models1
This work proposes M^2LLM, a multi-view framework that integrates three perspectives: the molecular structure view, the molecular task view, and the molecular rules view that achieves state-of-the-art performance on multiple benchmarks across classification and regression tasks.
- Unsupervised Federated Graph Learning1
The proposed FedPAM is superior over classical and SOTA compared methods, and adaptively integrates the global model parameters in a low-rank tensor space, which facilitates to fuse the high-order knowledge among clients.
- INSPIRE-GNN: Intelligent Sensor Placement to Improve Sparse Bicycling Network Prediction via Reinforcement Learning Boosted Graph Neural Networks1
INSPIRE-GNN, a novel Reinforcement Learning (RL)-boosted hybrid Graph Neural Network (GNN) framework designed to optimize sensor placement and improve link-level bicycling volume estimation in data-sparse environments, demonstrates significant improvements in volume estimation.
- 1
- 1
- OASIS: Harnessing Diffusion Adversarial Network for Ocean Salinity Imputation using Sparse Drifter Trajectories1
The OceAn Salinity Imputation System is introduced, a novel diffusion adversarial framework designed to address ocean salinity measurement challenges by employing a transformer-based global dependency capturing module to learn long-range spatio-temporal correlations from sparse trajectories and constructing a generative imputation model that conditions on easily observed tidal covariates to progressively refine imputed salinity fields.
- 1
- 1
- 1
- DropNaE: Alleviating irregularity for large-scale graph representation learning1
This paper proposes DropNaE to alleviate the irregularity in large-scale graphs by conditionally dropping nodes and edges before GNN training and proposes DropNaE containing two variants to transform the irregular degree distribution of the large-scale graph to a uniform one, based on the proposed metric.
- 1
- Rethinking Efficiency and Redundancy in Training Large-scale Graphs1
This paper proposes DropReef, a once-for-all method to detect and drop the redundancy in large-scale graphs once and for all, helping reduce the training time while ensuring no sacrifice in the model accuracy.
- Cyclic Transfer Learning for Recommender Systems with Heterogeneous Feedbacks1
A recommendation framework based on heterogeneous feedbacks and cyclic transfer learning (HCTL-Rec) and two specific recommendation algorithms which combine the novel framework with two auxiliary domains to better represent users’ ratings are presented.
- 1
- 1
- 1
- Towards Extracting Graph Neural Network Models via Prediction Queries (Student Abstract)1
This paper exploits the threat of model extraction attacks against GNN models by leveraging both the node queries and the graph structure to extract the GNNs and produces a surrogate model with more than 80% equivalent predictions as the target model.
- 1
- 1
- 1
- IEEE Access Special Section Editorial: Advanced Data Analytics for Large-Scale Complex Data Environments1
Novel data analytics methods are needed to discover meaningful knowledge in advanced applications from objects with large-scale characteristics, such as chemical pounds in bio-pharmacy, brain regions in brain networks and users in social networks.
- Direct Discriminative Bag Mapping for Multi-Instance Learning1
This paper proposes a direct discriminative mapping approach for multi-instance learning (MILDM), which identifies instances to directly distinguish bags in the new mapping space.
- Meta-Learning Driven Few-Shot Knowledge Transfer with Dual-Stage Adaptive Data Replay for Cross-Domain Recommendation–
A new meta-learning driven few-shot knowledge transfer method that leverages overlapping parts as anchors to guide the non-overlapping parts in autonomously exploring alignment rules through meta-learning, thus enabling few-shot knowledge transfer and flexible handling of different cross-domain scenarios.
- –
- UniEdit: A Graph-based MoE Alternative to Sequence Generation for Molecular Editing–
This paper proposes UniEdit, a Unified Graph-based Mixture-of-Experts (MoE) Molecular Editing model that offers a robust alternative to LLMs and incorporates a Mixture-of-Experts architecture that dynamically routes tasks to specialized components.
- Natural Language-Powered Functional Protein Sequence and Structure Co-Design with Multi-Modal Knowledge Fusion–
This work proposes a Protein sequence–structure–function Consistency Design model empowered by Natural Language function description, dubbed ProtcdNl, which expands the protein design space and ensures alignment with function-aware framework, achieving the design of novel, diverse proteins with high functional fidelity.
- Foundation Models for Time Series Analysis: Concepts, Methodologies, and Applications–
This tutorial offers insights into FM theory, implementation, and practical applications, including principles, pre-processing techniques, and modeling strategies, and best practices for integrating FMs into workflows and exploring applications in diverse fields, such as finance, healthcare, and transportation.
- –
- AbstainGNN: Teaching Graph Neural Networks to Abstain for Graph Classification–
This work proposes AbstainGNN, a novel and theory-driven framework for graph classification with abstention that enables GNNs to reject uncertain predictions instead of producing incorrect decisions, and develops an efficient two-stage training strategy consisting of predictive function warm-start and abstention function calibration.
- Graph Navier-Stokes Networks–
Graph Navier-Stokes Networks is introduced, a novel architecture that transcends conventional diffusion-based message passing by incorporating convection into graph structures and reveals, for the first time, a systematic inverse correlation between graph homophily and the strength of convection during message passing.
- –
- –
- Multi-Source Unsupervised Graph Domain Adaptation via Concise Propagation–Transformation Pipeline–
This paper proposes a novel approach, which leverages Concise Propagation–Transformation pipeline for multi-source unsupervised Graph Domain Adaptation, dubbed as CPT-GDA, to better capture complementary knowledge from multiple sources in an efficient manner.
- SRLST: a unified multimodal representation learning framework for spatial transcriptomics analysis–
Across distinct experimental datasets, SRLST consistently outperforms existing methods in delineating cortical organization, identifying small discontinuous tissue compartments, and capturing complex intratumor heterogeneity.
- –
- –
- 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.
- –
- Adaptive Tokenization: On the Hop-Overpriority Problem in Tokenized Graph Learning Models–
The Learnable Graph Token List (LGTL), a plug-and-play module to replace hand-designed token lists in TGLMs, is proposed, which adaptively adjusts the weights across hops and prioritizes informative nodes within hops through a graph attention gate module and a selection module, respectively.
- –
- –
- –
- Graph positive-unlabeled learning via Bootstrapping Label Disambiguation–
This paper introduces a learning method called Bootstrap Label Disambiguation (BLD), which progressively resolves label ambiguities during the training of binary classifiers and significantly outperforms existing approaches and in many cases even surpasses fully labeled classification models.
- TGLsta: Low-resource Textual Graph Learning with Semantic and Topological Awareness via LLMs–
A novel method of low-resource textual graph node classification based on large language models, i.e., Textual graph learning with semantic and topological awareness (TGLsta), which comprehensively explores the semantic information, near neighborhood information, and the topology information in textual graphs, where these components are the most important information source contained in textual graphs.
- –
- –
- –
- –
- –
- Inferring Neuron-level Brain Circuit Connection via Graph Neural Network Amidst Small Established Connections–
This paper proposes Neuronal Circuit Prediction Network (NCPNet), a simple and effective model to jointly learn node structural representation and neighborhood representation, constructing neuronal connection pair feature for inferring neuron-level connections in a brain circuit network.
- –
- –
- The 3rd International Workshop on Machine Learning on Graphs (MLoG)–
The 3rd International Workshop on Machine Learning on Graphs (MLoG), held in conjunction with the 16th ACM Conference on Web Search and Data Mining (WSDM), provides a venue to gather academia researchers and industry researchers/practitioners to present the recent progress on machine learning on graphs.
- –
- –
- –
- Complex graph stream mining–
A novel algorithm is proposed, CGStream, to identify correlated graphs from a data stream, by using a sliding window, which covers a number of consecutive batches of stream data records, and experimental results demonstrate that the proposed algorithm is several times, or even an order of magnitude, more efficient than the straightforward algorithms.
- Mining Top-k Minimal Redundancy Frequent Patterns over Uncertain Databases–
This paper presents a new algorithm, UTFP, to mine top-k minimal redundancy frequent patterns of length no less than minimum length \(min\_l\) without setting threshold, and proposes a set of strategies to prune and reduce search space.
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 2026-10-12. 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.