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
- Huan LiuSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works374 from public data
- Fake News Detection on Social Media3,450
This survey presents a comprehensive review of detecting fake news on social media, including fake news characterizations on psychology and social theories, existing algorithms from a data mining perspective, evaluation metrics and representative datasets, and future research directions for fake news detection on socialMedia.
- Graph Neural Networks for Social Recommendation2,435
This paper provides a principled approach to jointly capture interactions and opinions in the user-item graph and proposes the framework GraphRec, which coherently models two graphs and heterogeneous strengths for social recommendations.
- Feature Selection2,424
This survey revisits feature selection research from a data perspective and reviews representative feature selection algorithms for conventional data, structured data, heterogeneous data and streaming data, and categorizes them into four main groups: similarity- based, information-theoretical-based, sparse-learning-based and statistical-based.
- Graph Structure Learning for Robust Graph Neural Networks975
A general framework Pro-GNN is proposed, which can jointly learn a structural graph and a robust graph neural network model from the perturbed graph guided by these properties, and achieves significantly better performance compared with the state-of-the-art defense methods, even when the graph is heavily perturbed.
- A Survey on Dialogue Systems788
This article generally divides existing dialogue systems into task-oriented and nontask- oriented models, then detail how deep learning techniques help them with representative algorithms and finally discusses some appealing research directions that can bring the dialogue system research into a new frontier.
- Adversarial Attacks and Defenses in Images, Graphs and Text: A Review772
A systematic and comprehensive overview of the main threats of attacks and the success of corresponding countermeasures against adversarial examples, for three most popular data types, including images, graphs and text is reviewed.
- 757
- 739
- Traffic Flow Prediction via Spatial Temporal Graph Neural Network681
A novel spatial temporal graph neural network for traffic flow prediction, which can comprehensively capture spatial and temporal patterns and provides a sequential component to model the traffic flow dynamics which can exploit both local and global temporal dependencies.
- Recommender Systems in the Era of Large Language Models (LLMs)604
This survey comprehensively review LLM-empowered recommender systems from various perspectives including pre-training, fine-tuning, and prompting paradigms, and comprehensively discusses the promising future directions in this emerging field.
- 581
- 504
- Exploring the Potential of Large Language Models (LLMs)in Learning on Graphs479
This paper aims to explore the potential of LLMs in graph machine learning, especially the node classification task, and investigates two possible pipelines: LLMs-as-Enhancers and LLMs-as-Predictors.
- 462
- Deep reinforcement learning for page-wise recommendations454
A principled approach to jointly generate a set of complementary items and the corresponding strategy to display them in a 2-D page is proposed and a novel page-wise recommendation framework based on deep reinforcement learning, DeepPage, which can optimize a page of items with proper display based on real-time feedback from users is proposed.
- 424
- Large Language Models for Education: A survey and outlook423
A comprehensive technological overview for educators, researchers, and policy makers to harness the power of LLMs, revolutionize educational practices, and foster a more effective personalized learning environment is provided.
- Attributed Network Embedding for Learning in a Dynamic Environment396
DANE first provides an offline method for a consensus embedding and then leverages matrix perturbation theory to maintain the freshness of the end embedding results in an online manner, and performs extensive experiments to corroborate the effectiveness and efficiency of the proposed framework.
- Recommendations with Negative Feedback via Pairwise Deep Reinforcement Learning394
This paper model the sequential interactions between users and a recommender system as a Markov Decision Process (MDP) and leverage Reinforcement Learning (RL) to automatically learn the optimal strategies via recommending trial-and-error items and receiving reinforcements of these items from users' feedback.
- mTrust390
Experimental results on real-world data from Epinions and Ciao show that the work of discerning multi-faceted trust can be applied to improve the performance of tasks such as rating prediction, facet-sensitive ranking, and status theory.
- Graph Convolutional Networks with EigenPooling371
A pooling operator based on graph Fourier transform is introduced, which can utilize the node features and local structures during the pooling process and is combined with traditional GCN convolutional layers to form a graph neural network framework for graph classification.
- A Survey of Signed Network Mining in Social Media368
A review of mining signed networks in the context of social media and discuss some promising research directions and new frontiers of signed network mining.
- 345
- Node Similarity Preserving Graph Convolutional Networks336
This work proposes a feature similarity preserving aggregation which adaptively integrates graph structure and node features and employs self-supervised learning to explicitly capture the complex feature similarity and dissimilarity relations between nodes.
- Trustworthy AI: A Computational Perspective310
A comprehensive appraisal of trustworthy AI from a computational perspective to help readers understand the latest technologies for achieving trustworthy AI and focuses on six of the most crucial dimensions.
- Streaming Graph Neural Networks296
DyGNN, a Dynamic Graph Neural Network model, which can model the dynamic information as the graph evolving is proposed, which keeps updating node information by capturing the sequential information of edges (interactions), the time intervals between edges and information propagation coherently.
- Retrieval-Augmented Generation with Graphs (GraphRAG)295
A comprehensive and up-to-date survey on GraphRAG is presented and a holistic GraphRAG framework is proposed by defining its key components, including query processor, retriever, organizer, generator, and data source.
- User Identity Linkage across Online Social Networks278
Key achievements of user identity linkage across online social networks including state-of- the-art algorithms, evaluation metrics, and representative datasets are reviewed.
- Signed Network Embedding in Social Media262
A deep learning framework SiNE for signed network embedding is proposed that optimizes an objective function guided by social theories that provide a fundamental understanding of signed social networks.
- A Comprehensive Survey on Trustworthy Graph Neural Networks: Privacy, Robustness, Fairness, and Explainability261
A comprehensive survey of GNNs in the computational aspects of privacy, robustness, fairness, and explainability is given and the taxonomy of related methods and general frameworks for the multiple categories of trustworthy GNNs are formulated.
- What Your Images Reveal255
A new framework Visual Content Enhanced POI recommendation (VPOI), which incorporates visual contents for POI recommendations, is proposed, which demonstrates the effectiveness of the proposed framework on real-world datasets.
- 230
- Adversarial Attacks and Defenses on Graphs225
A comprehensive overview of existing graph adversarial attacks and the countermeasures is provided, categorize existing attacks and defenses, and review the corresponding state-of-the-art methods.
- Multimodal Recommender Systems: A Survey218
This article gives a comprehensive survey of the MRS models, mainly from technical views, and introduces the existing MRS models according to four categories, i.e., Modality Encoder, Feature Interaction, Feature Enhancement, and Model Optimization.
- A Unified View on Graph Neural Networks as Graph Signal Denoising213
It is established mathematically that the aggregation processes in a group of representative GNN models including GCN, GAT, PPNP, and APPNP can be regarded as solving a graph denoising problem with a smoothness assumption.
- Multi-Source Multi-Class Fake News Detection192
This paper introduces approaches to combine information from multiple sources and to discriminate between different degrees of fakeness, and proposes a Multi-source Multi-class Fake news Detection framework MMFD, which combines automated feature extraction, multi-source fusion and automated degrees offakeness detection into a coherent and interpretable model.
- 179
- Empowering Molecule Discovery for Molecule-Caption Translation With Large Language Models: A ChatGPT Perspective168
This work proposes a novel LLM-based framework (MolReGPT) for molecule-caption translation, where an In-Context Few-Shot Molecule Learning paradigm is introduced to empower molecule discovery with LLMs like ChatGPT to perform their in-context learning capability without domain-specific pre-training and fine-tuning.
- Learning Word Representations for Sentiment Analysis167
It is proved that incorporating prior sentiment knowledge into the embedding process has the potential to learn better representations for sentiment analysis.
- Condensing Graphs via One-Step Gradient Matching164
A one-step gradient matching scheme, which performs gradient matching for only one single step without training the network weights, is proposed, which can generate synthetic graphs that lead to lower classification loss on real graphs.
- Does Gender Matter? Towards Fairness in Dialogue Systems161
A benchmark dataset is constructed and quantitative measures to understand fairness in dialogue models are proposed and it is demonstrated that popular dialogue models show significant prejudice towards different genders and races.
- Micro Behaviors156
The effects of micro behaviors on recommendations are uncovered and an interpretable Recommendation framework RIB is proposed, which models inherently the sequence of mIcro Behaviors and their effects.
- Ranking Relevance in Yahoo Search156
This paper introduces three key techniques for base relevance -- ranking functions, semantic matching features and query rewriting, and describes solutions for recency sensitive relevance and location sensitive relevance.
- Memory Injection Attacks on LLM Agents via Query-Only Interaction151
A novel Memory INJection Attack, MINJA, without assuming that the attacker can directly modify the memory bank of the agent, which enables any user to influence agent memory, highlighting the risk.
- Learning Hierarchical Discourse-level Structure for Fake News Detection148
HDSF learns and constructs a discourse-level structure for fake/real news articles in an automated and data-driven manner and identifies insightful structure-related properties, which can explain the discovered structures and boost the understating of fake news.
- Node Classification in Signed Social Networks147
A principled way to mathematically model positive and negative links simultaneously is provided and a novel framework NCSSN for node classification in signed social networks is proposed.
- Negative Link Prediction in Social Media147
This paper investigates the novel problem of negative link prediction with only positive links and content-centric interactions in social media, and proposes a principled framework NeLP, which can exploit positive linksand content-focused interactions to predict negative links.
- Unsupervised Streaming Feature Selection in Social Media145
This paper investigates how to exploit link information in streaming feature selection, resulting in a novel unsupervised streaming feature Selection framework USFS, and shows the effectiveness and efficiency of the proposed framework comparing with the state-of-the-art un supervised feature selection algorithms.
- Graph Trend Filtering Networks for Recommendation144
A principled graph trend collaborative filtering method is introduced and the Graph Trend Filtering Networks for recommendations (GTN) that can capture the adaptive reliability of the interactions is proposed.
- DEAR: Deep Reinforcement Learning for Online Advertising Impression in Recommender Systems143
This paper develops an RL-based framework that can continuously update its advertising strategies and maximize reward in the long run and designs a novel Deep Q-network architecture that can determine three internally related tasks jointly.
- Multi-dimensional Graph Convolutional Networks139
This paper proposes a multi-dimensional convolutional neural network model mGCN aiming to capture rich information in learning node-level representations formulti-dimensional graphs and demonstrates the effectiveness of the proposed framework on real-world multi- dimensional graphs.
- Modeling Temporal-Spatial Correlations for Crime Prediction128
This paper validate the existence of temporal-spatial correlations in crime and develop a principled approach to model these correlations into the coherent framework TCP for crime prediction, and experimental results on real-world data demonstrate the effectiveness of the proposed framework.
- Recommendations in Signed Social Networks118
This work provides a principled and mathematical approach to exploit signed social networks for recommendation, and proposes a model, RecSSN, to leverage positive and negative links insigned social networks.
- Streaming Recommender Systems117
A principled framework termed sRec is provided, which provides explicit continuous-time random process models of the creation of users and topics, and of the evolution of their interests, and which permits computationally efficient instantaneous on-line inference.
- Label-free Node Classification on Graphs with Large Language Models (LLMS)116
This work introduces a label-free node classification on graphs with LLMs pipeline, LLM-GNN, which amalgamates the strengths of both GNNs and LLMs while mitigating their limitations while leveraging the confidence scores derived from LLMs to advanced node selection.
- CellPLM: Pre-training of Cell Language Model Beyond Single Cells116
The proposed CellPLM is the first single-cell pre-trained transformer that encodes cell-cell relations and it consistently outperforms existingPre-trained and non-pre-trained models in diverse downstream tasks, with 100x times higher inference speed compared to existing pre- trained models.
- AI for Education (AI4EDU): Advancing Personalized Education with LLM and Adaptive Learning114
This workshop aims to bring together researchers and practitioners from academia and industry to explore cutting-edge AI technologies for personalized education, especially the potential of LLMs and adaptive learning technologies.
- Attributed Signed Network Embedding114
A novel framework SNEA is proposed, which exploits the network structure and user attributes simultaneously for network representation learning and demonstrates the effectiveness of SNEA on link prediction and node clustering with real-world datasets.
- Graph Representation Learning: Foundations, Methods, Applications and Systems113
This tutorial systematically review the foundations, techniques, applications and advances in graph representation learning, and first introduces the foundations on graph theory and graph Fourier analysis.
- Empowering Graph Representation Learning with Test-Time Graph Transformation109
This work proposes a graph transformation framework named GTrans which adapts and refines graph data at test time to achieve better performance and provides theoretical analysis on the design of the framework and discusses why adapting graph data works better than adapting the model.
- Deep social collaborative filtering109
DSCF is proposed, a Deep Social Collaborative Filtering framework, which can exploit the social relations with various aspects for recommender systems and shows the effectiveness of the proposed framework on two-real world datasets.
- Recommendation with Social Dimensions109
This paper employs social dimensions to simultaneously capture heterogeneity of social relations and weak dependency connections, and provides principled ways to model social dimensions, and proposes a recommendation framework SoDimRec which incorporates heterogeneity ofsocial relations andWeak dependency connections based on social dimensions.
- Adversarial Attacks and Defenses on Graphs: A Review and Empirical Study108
A comprehensive overview of existing graph adversarial attacks and the countermeasures is provided, categorize existing attacks and defenses, and review the corresponding state of the art methods.
- "Deep reinforcement learning for search, recommendation, and online advertising: a survey" by Xiangyu Zhao, Long Xia, Jiliang Tang, and Dawei Yin with Martin Vesely as coordinator106
An overview of deep reinforcement learning for search, recommendation, and online advertising from methodologies to applications, review representative algorithms, and discuss some appealing research directions are given.
- RAG vs. GraphRAG: A Systematic Evaluation and Key Insights105
A comprehensive benchmark study comparing RAG and GraphRAG on established text-based tasks, including question answering and query-based summarization is presented and a unified evaluation protocol is introduced that standardizes data preprocessing, retrieval configurations, and generation settings, enabling fair and reproducible comparisons.
- Linked Document Embedding for Classification103
This paper proposes a linked document embedding framework LDE, which combines link and label information with content information to learn document representations for classification, and experiments demonstrate the effectiveness of the proposed framework.
- 102
- 98
- Automated Embedding Size Search in Deep Recommender Systems98
This paper proposes to dynamically search the embedding sizes for different users and items and introduces a novel embedding size adjustment policy network (ESAPN).
- A Path-constrained Framework for Discriminating Substitutable and Complementary Products in E-commerce98
A path-constrained framework (PMSC) for discriminating substitutes and complements is proposed, which first learns its embedding representations in a general semantic space, then incorporates each embedding with path- Constraints to further boost the discriminative ability of the model.
- 98
- Multi-Dimensional Network Embedding with Hierarchical Structure97
This paper provides an approach to capture independent information from each dimension and dependent information across dimensions and proposes a framework MINES, which performs Multi-dImension Network Embedding with hierarchical Structure.
- Attacking Black-box Recommendations via Copying Cross-domain User Profiles95
The proposed framework CopyAttack is a reinforcement learning based black-box attacking method that harnesses real users from a source domain by copying their profiles into the target domain with the goal of promoting a subset of items.
- Hierarchical Variational Memory Network for Dialogue Generation95
A novel hierarchical variational memory network (HVMN) is proposed, by adding the hierarchical structure and the variationalMemory network into a neural encoder-decoder network that can capture both the high-level abstract variations and long-term memories during dialogue tracking, which enables the random access of relevant dialogue histories.
- 94
- Enriching short text representation in microblog for clustering89
This work proposes a novel framework which performs multi-language knowledge integration and feature reduction simultaneously through matrix factorization techniques and investigates potential factors that contribute to the improved performance.
- Graph Adversarial Attack via Rewiring88
This paper proposes a graph rewiring operation to perform the attack to affect the graph in a less noticeable way compared to existing operations such as adding/deleting edges and analyzes how its generated perturbation impacts the target model and the advantages of the rewires operations.
- 88
- Context-aware review helpfulness rating prediction85
This paper identifies various types of context information, model them mathematically, and propose a context-aware review helpfulness rating prediction framework CAP, which demonstrates the effectiveness of the proposed framework and the importance of context awareness in solving the review helpfuls rating prediction problem.
- Jointly Learning to Recommend and Advertise84
A novel two-level reinforcement learning framework to jointly optimize the recommending and advertising strategies, where the first level generates a list of recommendations to optimize user experience in the long run; then the second level inserts ads into the recommendation list that can balance the immediate advertising revenue from advertisers and the negative influence of ads on long-term user experience.
- Learning to Rewrite Queries82
A learning to rewrite framework that consists of a candidate generating phase and a candidate ranking phase that allows for the flexibility to reuse most of existing query rewriters and to explicitly optimize search relevance is proposed.
- Mitigating Gender Bias for Neural Dialogue Generation with Adversarial Learning79
This paper proposes a novel adversarial learning framework Debiased-Chat to train dialogue models free from gender bias while keeping their performance, and shows that this framework significantly reduces gender bias in dialogue models while maintaining the response quality.
- Multi-Scale One-Class Recurrent Neural Networks for Discrete Event Sequence Anomaly Detection78
OC4Seq is proposed, a multi-scale one-class recurrent neural network for detecting anomalies in discrete event sequences that integrates the anomaly detection objective with recurrent neural networks (RNNs) to embed the discreteevent sequences into latent spaces, where anomalies can be easily detected.
- Deep Adversarial Social Recommendation75
A novel deep adversarial social recommendation framework DASO is proposed that adopts a bidirectional mapping method to transfer users' information between social domain and item domain using adversarial learning.
- 71
- Copyright Protection in Generative AI: A Technical Perspective70
A comprehensive overview of copyright protection from a technical perspective is provided, examining from two distinct viewpoints: the copyrights pertaining to the source data held by the data owners and those of the generative models maintained by the model builders.
- Bringing Generative AI to Adaptive Learning in Education70
It is argued that this union of generative AI and adaptive learning will contribute significantly to the development of the next-stage learning format in education.
- AutoDim: Field-aware Embedding Dimension Searchin Recommender Systems70
An AutoML-based framework (AutoDim) is proposed, which can automatically select dimensions for different feature fields in a data-driven fashion and derive a hard and discrete embedding component architecture according to the maximal weights and retrain the whole recommender framework.
- 70
- Understanding and Discovering Deliberate Self-harm Content in Social Media70
The aim of this paper is to understand self-harm content and provide automatic approaches to its detection, and to propose frameworks that incorporate the findings to discover self- Harm content under both supervised and unsupervised settings.
- Crime in Urban Areas:69
An overview to key theories from criminology is given, crime analysis on urban data is summarized, state-of-the-art algorithms for various types of computational crime tasks are reviewed and some appealing research directions that can bring the urban crime research into a new frontier are discussed.
- 69
- Exploring implicit hierarchical structures for recommender systems68
This paper proposes a novel recommendation framework HSR to bridge the gap, which enables us to capture the implicit hierarchical structures of users and items simultaneously, and experimental results demonstrate the effectiveness of the proposed framework.
- AutoLoss: Automated Loss Function Search in Recommendations67
This work proposes an AutoLoss framework that can automatically and adaptively search for the appropriate loss function from a set of candidates, and develops a novel controller network, which can dynamically adjust the loss probabilities in a differentiable manner.
- Reconstruction-based Unsupervised Feature Selection: An Embedded Approach66
A novel reconstruction-based unsupervised feature selection framework REFS is proposed, which embeds the reconstruction function learning process into feature selection and demonstrates the effectiveness of the proposed framework on various types of real-world datasets.
- 65
- Transferable Unlearnable Examples64
A novel unlearnable strategy based on Classwise Separability Discriminant (CSD) is proposed, which aims to better transfer the Unlearnable effects to other training settings and datasets by enhancing the linear separability.
- Rethinking Large Language Model Architectures for Sequential Recommendations62
This work endeavors to streamline existing LLM-based recommendation models and proposes a simple yet highly effective model Lite-LLM4Rec, which circumvents the beam search decoding by using a straight item projection head for ranking scores generation and introduces a hierarchical LLM structure tailored to efficiently handle the extensive contextual information associated with items.
- Signed Network Modeling Based on Structural Balance Theory62
This paper provides a principled approach to capture important properties and principles of signed networks and proposes a novel signed network model guided by Structural Balance Theory.
- 61
- DiffusionShield: A Watermark for Data Copyright Protection against Generative Diffusion Models59
A novel watermark scheme, Diffusion Shield, is introduced against Generative Diffusion Models, which protects images from infringement by encoding the ownership message into an imperceptible watermark and injecting it into images.
- Feature Overcorrelation in Deep Graph Neural Networks59
A new perspective to look at the performance degradation of deep GNNs, i.e., feature overcorrelation, is proposed and a general framework DeCorr is proposed which can encourage GNN's to encode less redundant information.
- Trust in Social Media59
The explosion of user-generated content poses novel challenges for online users to find relevant information, or, to create and share information at any place at any time.
- 58
- Deep Knowledge Tracing with Side Information56
This paper exploits side relations to improve knowledge tracing and design a novel framework DTKS, and the experimental results validate the effectiveness of the proposed framework and demonstrate the importance of side information in knowledge tracing.
- Deep Adversarial Network Alignment55
This work presents a Deep Adversarial Network Alignment (DANA) framework that first uses deep adversarial learning to discover complex mappings for aligning the embedding distributions of the two networks, and performs an efficient nearest neighbor node alignment.
- 54
- Automatic Short Answer Grading via Multiway Attention Networks54
This paper proposes a generalized end-to-end ASAG learning framework which aims to autonomously extract linguistic information from both student and reference answers and accurately model the semantic relations between free-text student andreference answers in open-ended domain.
- Signed Link Analysis in Social Media Networks53
This paper adopts three social science theories, namely Emotional Information, Diffusion of Innovations and Individual Personality, to guide the task of link analysis in signed networks.
- Online Academic Course Performance Prediction Using Relational Graph Convolutional Neural Network.53
Deep Online Performance Evaluation (DOPE) is introduced, which models the student course relations in an online system as a knowledge graph, then utilizes an advanced graph neural network to extract course and student embeddings, harnesses a recurrent neural network to encode the system’s temporal student behavioral data, and ultimately predicts a student’s performance in a given course.
- Balance in Signed Bipartite Networks49
This work conducts the first comprehensive analysis and validation of balance theory using the smallest cycle in signed bipartite networks - signed butterflies (i.e., cycles of length 4 containing the two node types).
- 48
- Revisiting Link Prediction: A Data Perspective48
This paper unearth relationships among those factors where (i) global structural proximity only shows effectiveness when local structural proximity is deficient, and (ii) the incompatibility can be found between feature and structural proximity.
- Exploiting Emotional Information for Trust/Distrust Prediction48
A principled way to capture emotional information mathematically is provided and a novel trust/distrust prediction framework ETD is proposed, which demonstrates the effectiveness of the proposed framework and the importance of emotional information in trust/ Distrust prediction.
- ROSE: Role-based Signed Network Embedding47
The core idea is that rather than directly finding the similarities of two nodes from the complex paths connecting them, they can obtain their similarities through simple paths connecting their different roles.
- 47
- 46
- Unsupervised Sentiment Analysis with Signed Social Networks46
This paper incorporates explicit sentiment signals in textual terms and implicit sentiment signals from signed social networks into a coherent model SignedSenti for unsupervised sentiment analysis, and empirical experiments corroborate its effectiveness.
- DeepRobust: a Platform for Adversarial Attacks and Defenses45
DeepRobust is a useful tool to measure deep learning model robustness and to find the suitable countermeasures against adversarial attacks and is introduced with detailed instructions.
- 43
- Model-Based Reinforcement Learning for Whole-Chain Recommendations.43
This paper proposes a multi-agent reinforcement learning based approach (DeepChain), which can capture the sequential correlation among different scenarios and jointly optimize multiple recommendation strategies.
- Recommender Systems with Heterogeneous Side Information42
A novel framework jointly captures flat and hierarchical side information with mathematical coherence is proposed and empirical results show that it is able to lead a significant performance gain over the state-of-the-art methods.
- 42
- CrossFire42
A cross-platform knowledge transfer framework, which can capture both 1) cross- Platform knowledge transfer, and 2) within-platform correlations among user-user relations and user-item interactions is proposed.
- Deep Learning in Single-cell Analysis40
A comprehensive survey on deep learning in single-cell analysis, including seven popular tasks spanning different stages of the single-cell analysis pipeline, including multimodal integration, imputation, clustering, spatial domain identification, cell-type deconvolution, cell segmentation, and cell-type annotation.
- Jointly Attacking Graph Neural Network and its Explanations40
A novel attack framework (GEAttack) for graphs is proposed, which can attack both a GNN model and its explanations by exploiting their vulnerabilities simultaneously, and is the very first effort to attack both GNNs and explanations on graph-structured data for the trustworthiness of Gnns.
- SA-GDA: Spectral Augmentation for Graph Domain Adaptation40
The Spectral Augmentation for Graph Domain Adaptation (SA-GDA) for graph node classification is presented, a dual graph convolutional network is developed to jointly exploits local and global consistency for feature aggregation, and a domain classifier with an adversarial learning submodule is utilized to facilitate knowledge transfer between different domain graphs.
- Multi-Type Urban Crime Prediction40
This paper verifies the existence of correlations among different types of crime from temporal and spatial perspectives, and proposes a coherent framework to mathematically model these correlations for crime prediction.
- LPFormer: An Adaptive Graph Transformer for Link Prediction39
LPFormer is proposed, which attempts to adaptively learn the pairwise encodings for each link via an attention module that learns the pairwise encoding that exists between nodes by modeling multiple factors integral to link prediction.
- 38
- Toward Degree Bias in Embedding-Based Knowledge Graph Completion38
This paper validate the existence of degree bias in embedding-based KGC and identify the key factor to degree bias, and introduces a novel data augmentation method, KG-Mixup, to generate synthetic triples to mitigate such bias.
- The good, the bad, and the ugly: uncovering novel research opportunities in social media mining38
In the endeavor of employing the good to tame the bad with the help of the ugly, this work deepen the understanding of ever growing and continuously evolving data and create innovative solutions with interdisciplinary and collaborative research of data science.
- A Survey on Small Language Models in the Era of Large Language Models: Architecture, Capabilities, and Trustworthiness37
There is a lack of tutorials on cutting-edge SLM technologies, prompting us to conduct one, and the trustworthiness of SLMs remains a critical yet underexplored research area.
- The Authors Matter: Understanding and Mitigating Implicit Bias in Deep Text Classification37
This paper demonstrates that implicit bias exists in different text classification tasks for different demographic groups, and proposes a framework Debiased-TC to train deep text classifiers to make predictions on the right features and consequently mitigate implicit bias.
- How Do Large Language Models Understand Graph Patterns? A Benchmark for Graph Pattern Comprehension36
The findings reveal that LLMs have preliminary abilities to understand graph patterns, with O1-mini outperforming in the majority of tasks and Formatting input data to align with the knowledge acquired during pretraining can enhance performance.
- 36
- 36
- CoSelect: Feature Selection with Instance Selection for Social Media Data36
A novel feature selection framework, CoSelect, is proposed, which can exploit link information by applying social correlation theories, incorporate instance selection with feature selection, and select relevant instances and features simultaneously.
- 35
- Epidemic Graph Convolutional Network34
This work first observes the similarities in the two sampling strategies to that of epidemic and diffusion network models, then harnesses this understanding to fuse together the benefits of sampling from both a local and global perspective while alleviating some of the inherent issues found in both through the use of a low-dimensional approximation for the path-based Katz similarity measure.
- Dynamic Graph Neural Networks33
DGNN is proposed, a new graph neural network model which can model the dynamic information as the graph evolving, which can keep updating node information by capturing the sequential information of edges, the time intervals between edges and information propagation coherently.
- Rating Distribution Calibration for Selection Bias Mitigation in Recommendations32
A novel self-supervised learning framework, i.e., Rating Distribution Calibration (RDC), is proposed, to tackle selection bias without introducing additional models and aims to correct the predicted rating distribution of biased users by taking advantage of that of their similar unbiased users.
- UserSim: User Simulation via Supervised GenerativeAdversarial Network32
A user simulator based on a Generative Adversarial Network (GAN) that captures the underlying distribution of users’ historical logs and generates realistic logs that can be considered as augmentations of real logs; while the discriminator not only distinguishes real and fake logs but also predictsUsers’ behaviors.
- FT-Shield: A Watermark Against Unauthorized Fine-tuning in Text-to-Image Diffusion Models31
The proposed FT-Shield addresses copyright protection challenges by designing new watermark generation and detection strategies and introduces an innovative algorithm for watermark generation that integrates a Mixture of Experts (MoE) approach for watermark detection.
- Position: Graph Foundation Models are Already Here31
A novel perspective for the GFM development is proposed by advocating for a “graph vocabulary”, in which the basic transferable units underlying graphs encode the invariance on graphs, which can potentially advance the future GFM design following the neural scaling laws.
- Decision Boundary of Deep Neural Networks: Challenges and Opportunities30
A novel approach is proposed to generate instances near decision boundary of pre-trained DNNs and then leverage these instances to characterize the behaviour of deep models and thus solve the challenges of investigating the decision boundaries of deep neural network classifiers.
- 30
- Twitter summarization with social-temporal context30
This paper provides a methodology to model temporal context globally and locally, and proposes a novel unsupervised summarization framework with social-temporal context for Twitter data, to assess the proposed framework and demonstrate the importance of social- Temporal context in Twitter summarization.
- A LLM-Powered Automatic Grading Framework with Human-Level Guidelines Optimization29
A unified multi-agent ASAG framework, GradeOpt, is proposed, which leverages large language models (LLMs) as graders for SAGs and incorporates two additional LLM-based agents - the reflector and the refiner - into the multi-agent system.
- Graph-level Representation Learning with Joint-Embedding Predictive Architectures28
This paper employs masked modeling and focuses on predicting the latent representations of masked subgraphs starting from the latent representation of a context subgraph, and devise an alternative prediction objective that consists of predicting the coordinates of the encoded subgraphs on the unit hyperbola in the 2D plane.
- An Adaptive Graph Pre-training Framework for Localized Collaborative Filtering28
An adaptive graph pre-training framework for localized collaborative filtering (ADAPT) is designed that does not require transferring user/item embeddings, and is able to capture both the common knowledge across different graphs and the uniqueness for each graph simultaneously.
- Graduate Employment Prediction with Bias28
A framework to predict students' employment status while considering biases, which solves the heterogeneity of student courses by embedding academic performance into a unified space and designs a bias-based regularization to capture the job market biases.
- Positive-Unlabeled Learning in Streaming Networks28
A principled probabilistic approach SPU is proposed to leverage the characteristics of the streaming PU inputs and provides real-time adaptations and predictions by identifying the potential negative signals concealed in unlabeled data.
- PDHG-Unrolled Learning-to-Optimize Method for Large-Scale Linear Programming27
The proposed PDHG-Net can recover PDHG algorithm, thus can approximate optimal solutions of LP instances with a polynomial number of neurons, and can significantly accelerate LP solving, achieving up to a 3$\times speedup compared to FOMs for large-scale LP problems.
- Masked Graph Autoencoder with Non-discrete Bandwidths27
This paper proposes a novel, informative, and effective topological masked graph autoencoder using bandwidth masking and a layer-wise bandwidth prediction objective, which outperforms representative baselines in both self-supervised link prediction and node classification on numerous datasets, solely with a structure-learning pretext.
- 26
- 26
- Understanding compromised accounts on Twitter26
The findings suggest that the accounts can be compromised by two different types of hackers and the content they post tend to follow several common themes, and there are other meta-information that can be exploited to help improve the detection of compromised accounts.
- Reasoning with Graphs: Structuring Implicit Knowledge to Enhance LLMs Reasoning25
Reasoning with Graphs (RwG) is proposed by first constructing explicit graphs from the context and then leveraging these graphs to enhance LLM reasoning performance on reasoning tasks, improving both logical reasoning and multi-hop question answering tasks.
- Identifying Rhetorical Questions in Social Media24
This is the first framework to model the motivations for posting rhetorical questions to identify them on social media platforms and evaluates the framework using questions posted on Twitter.
- Mixture of Link Predictors on Graphs23
It is revealed that different node pairs within the same dataset necessitate varied pairwise information for accurate prediction and models that only apply the same pairwise information uniformly could achieve suboptimal performance.
- 23
- A Roadmap for Incorporating Online Social Media in Educational Research23
This chapter presents a roadmap for how to incorporate online social media in education research projects and presents some learned lessons, tips, and tricks, as well as case studies from the Teachers in Social Media project.
- Paired Restricted Boltzmann Machine for Linked Data23
A paired Restricted Boltzmann Machine (pRBM) is proposed, which is able to leverage the attribute and link information of linked data for representation learning and experimental results on real-world datasets demonstrate the effectiveness of the proposed framework.
- Investigating Out-of-Distribution Generalization of GNNs: An Architecture Perspective22
This work provides the first comprehensive investigation of OOD generalization on graphs from an architecture perspective, by examining the common building blocks of modern GNNs by developing a novel GNN backbone model, DGat, designed to harness the robust properties of both graph self-attention mechanism and the decoupled architecture.
- Graph Neural Networks in Natural Language Processing22
This chapter introduces the various graph representations that are extensively used in NLP, and shows how different NLP tasks can be tackled from a graph perspective.
- 21
- Single-Cell Multimodal Prediction via Transformers21
This work proposes a scMoFormer framework which can readily incorporate external domain knowledge and model the interactions within each modality and cross modalities and achieves superior performance on various benchmark datasets.
- Imbalanced Adversarial Training with Reweighting21
The poor data separability is one key reason causing this strong tension between under-represented and well-represented classes and the Separable Reweighted Adversarial Training (SRAT) framework is proposed to facilitate adversarial training under imbalanced scenarios, by learning more separable features for different classes.
- Toward Annotator Group Bias in Crowdsourcing21
It is revealed that annotators within the same demographic group tend to show consistent group bias in annotation tasks and thus a novel probabilistic graphical framework GroupAnno is developed to capture annotator group bias with an extended Expectation Maximization (EM) algorithm.
- 21
- Graph Embedding21
The edge congestion EC (G, H) is the minimum, over all embeddings f: V (G) → V (H), of the maximum number of wires that cross any edge of H.
- Understanding and Identifying Advocates for Political Campaigns on Social Media21
This paper draws from social movement theories and design a quantitative framework to study the nuanced message strategies, propagation strategies, and community structure adopted by advocates for political campaigns in social media.
- 20
- Node-wise Filtering in Graph Neural Networks: A Mixture of Experts Approach20
This work theoretically demonstrate that a global filter optimized for one pattern can adversely affect performance on nodes with differing patterns, and introduces a novel GNN framework Node-MoE that utilizes a mixture of experts to adaptively select the appropriate filters for different nodes.
- 20
- Adversarial Attacks and Defenses20
This tutorial provides a comprehensive overview on the frontiers and advances of adversarial attacks and their countermeasures, and introduces DeepRobust, a Pytorch adversarial learning library which aims to build a comprehensive and easy-to-use platform to foster this research field.
- 20
- Toward Dual Roles of Users in Recommender Systems20
A principled way to exploit the rater role mathematically is provided and a novel recommender system DualRec is proposed, which captures both the reviewer role and the raters role of users simultaneously for recommendation.
- 19
- 19
- 19
- Multi-document summarization via group sparse learning19
A novel compressive sensing based multi-document summarization with group sparse learning (SGS) framework is proposed, which can maximally reconstruct the original documents via minimizing the approximation error and jointly select summary sentences with the learnt group structure information among sentences.
- Enhancing LLM-Based Short Answer Grading with Retrieval-Augmented Generation18
This work proposes an adaptive RAG framework for automated grading that dynamically retrieves and incorporates domain-specific knowledge based on the question and student answer context and combines semantic search and curated educational sources to retrieve valuable reference materials.
- Global-and-Local Aware Data Generation for the Class Imbalance Problem18
A novel framework GL-GAN is proposed, which leverages the SMOTE method to explore local distribution in a learned latent space and employs GAN to capture the global information, so that synthetic minority samples can be generated under even extremely imbalanced scenarios.
- 18
- Representation Learning from Limited Educational Data with Crowdsourced Labels18
A grouping based deep neural network to learn embeddings from a limited number of training samples and a Bayesian confidence estimator to capture the inconsistency among crowdsourced labels are designed.
- Understanding and Identifying Rhetorical Questions in Social Media18
This is the first framework, to the best of the knowledge, to model the possible motivations for posting rhetorical questions on social media platforms and develop a quantitative framework from these motivations to identify rhetorical questions in social media.
- Delving Deep into Personal Photo and Video Search18
This paper is the first to study personal media search using large-scale real-world search logs and proposes the deep query understanding model to learn a mapping from the personal queries to the concepts in the clicked photos.
- A Theoretical Understanding of Chain-of-Thought: Coherent Reasoning and Error-Aware Demonstration17
This work theoretically shows that, compared to Stepwise ICL, the transformer gains better error correction ability and more accurate predictions if the reasoning from earlier steps (Coherent CoT) is integrated, and proposes an improvement on CoT by incorporating both correct and incorrect reasoning paths in the demonstration.
- Content Knowledge Identification with Multi-agent Large Language Models (LLMs)17
This work proposes a Multi-Agent LLMs-based framework, LLMAgent-CK, to assess the user responses' coverage of identified CK learning goals without human annotations, and presents promising CK identifying performance on a real-world mathematical CK dataset MaCKT.
- Are Message Passing Neural Networks Really Helpful for Knowledge Graph Completion?17
It is found that surprisingly, simple MLP models are able to achieve comparable performance to MPNNs, suggesting that MP may not be as crucial as previously believed.
- 17
- Educational Research in the Twenty-First Century17
This study investigates how computational science through machine learning can be leveraged to characterize the rigor of educational resources teachers curate, using a revised Bloom’s taxonomy, at scale within Pinterest.
- Neural Multi-task Learning for Teacher Question Detection in Online Classrooms17
An end-to-end neural framework that automatically detects questions from teachers’ audio recordings and strengthens the understanding of semantic relations among different types of questions by incorporating multi-task learning techniques.
- Personalized Multimodal Feedback Generation in Education17
A novel Personalized Multimodal Feedback Generation Network (PMFGN) armed with a modality gate mechanism and a personalized bias mechanism to address the challenges of personalized feedback generation is proposed.
- 17
- 17
- Recommendation in social media17
This tutorial aims to provide a comprehensive overview of various recommendation tasks in social media, especially their recent advances and new frontiers, and reviews state-of-the-art algorithms, and deliberate the emerging challenges and opportunities.
- Learning on Graphs with Large Language Models(LLMs): A Deep Dive into Model Robustness16
This work investigates the robustness against graph structural and textual perturbations in terms of two dimensions: LLMs-as-Enhancers and LLMs-as-Predictors and finds that both LLMs-as-Enhancers and LLMs-as-Predictors offer superior robustness against structural and textual attacks.
- 16
- Yet Meta Learning Can Adapt Fast, it Can Also Break Easily16
This paper formally defines key elements of adversarial attacks unique to meta learning and proposes the first attacking algorithm against meta learning under various settings and evaluates the effectiveness of the proposed attacking strategy as well as the robustness of several representative meta learning algorithms.
- Deep Reinforcement Learning for Online Advertising in Recommender Systems16
This paper develops a reinforcement learning based framework that can continuously update its advertising strategies and maximize reward in the long run and demonstrates the effectiveness of the proposed framework based on real-world data.
- 16
- 16
- DANCE: a deep learning library and benchmark platform for single-cell analysis15
The DANCE platform is introduced, the first standard, generic, and extensible benchmark platform for accessing and evaluating computational methods across the spectrum of benchmark datasets for numerous single-cell analysis tasks.
- 15
- A General Single-Cell Analysis Framework via Conditional Diffusion Generative Models15
This work proposes a general single-cell analysis framework by unifying common computational tasks as posterior estimation problems and introduces scDiff through the proposed framework and illustrates the flexibility of scDiff by incorporating prior information through large language models and graph neural networks.
- 15
- Social Science–guided Feature Engineering15
This research is mainly guided by three social science theories, Emotional Information, Diffusion of Innovations, and Individual Personality, and extracts three categories of related features and leverage them for signed link analysis.
- Detecting Harmful Online Conversational Content towards LGBTQIA+ Individuals14
This work introduces a real-world dataset, implements two baseline machine learning models, and fine-tune 3 pre-trained large language models that can achieve very promising performance on detecting online Anti-LGBTQIA+ conversational content detection tasks.
- 14
- 14
- Predicting Online Protest Participation of Social Media Users14
A framework to predict from the user's past status messages and interactions whether the next post of the user will be a declaration of protest is proposed and evaluated using data from the social media platform Twitter on protests during the recent Nigerian elections.
- Graph Enhanced BERT for Query Understanding13
A novel graph-enhanced pre-training framework, GE-BERT, is proposed, which leverages both query content and the query graph to capture both semantic information and users' search behavioral information of queries.
- 13
- Self-Comparison for Dataset-Level Membership Inference in Large (Vision-)Language Model12
It is found that a member prefix followed by a non-member suffix (paraphrased from a member suffix) can further trigger the model's memorization on training data, and a novel dataset-level membership inference method based on Self-Comparison is proposed.
- Automate Knowledge Concept Tagging on Math Questions with LLMs12
It is demonstrated that LLMs are a promising tool for concept tagging in math questions because the zero/few-shot learning capability of LLMs makes them well-suited for application in educational scenarios, which often face challenges in collecting large-scale, expertise-annotated datasets.
- Sharpness-Aware Data Poisoning Attack12
A novel attack method called Sharpness-Aware Data Poisoning Attack (SAPA), which leverages the concept of DNNs' loss landscape sharpness to optimize the poisoning effect on the worst re-trained model, regardless of the specific retraining procedure employed.
- Localized Graph Collaborative Filtering12
This work introduces a novel perspective to build GNN-based CF methods for recommendations which leads to the proposed framework Localized Graph Collaborative Filtering (LGCF), and empirical results demonstrate that LGCF provides complementary information to the embedding-basedCF model which can be utilized to boost recommendation performance.
- 12
- Link and interaction polarity predictions in signed networks12
This paper proposes a framework that jointly solves the link and interaction polarity prediction problem based on newly found understanding of how these two problems are correlated, and performs experiments on a real-world signed network to demonstrate the effectiveness of the proposed approach.
- Towards Improving Sample Representativeness of Teachers on Online Social Media: A Case Study on Pinterest12
This paper surveys teachers in the United States as seed samples and devise a heuristic that automatically identifies other Pinterest accounts that are likely to be teachers thus improving the sample representativeness of teachers on Pinterest.
- Cloud Information Retrieval: Model Description and Scheme Design12
A cloud information retrieval framework characterized by its retrieval risk formula, which enables, for the very first time, an effective retrieval of keywords from encrypted cloud data without undermining key word privacy and retrieval performance.
- 12
- AutoG: Towards automatic graph construction from tabular data11
This work introduces a set of datasets to formalize and evaluate graph construction methods, and proposes an LLM-based solution, AutoG, automatically generating high-quality graph schemas without human intervention, which can rival those produced by human experts.
- 11
- Siamese Neural Networks for Class Activity Detection11
This work proposes a Siamese neural framework to automatically identify teacher and student utterances from classroom recordings and demonstrates that the approach is superior on the prediction tasks for both online and offline classroom environments.
- Understanding and Predicting Weight Loss with Mobile Social Networking Data11
The initial investigation to understand weight loss with a large-scale mobile social networking dataset with near 10 million users is conducted and a number of interesting findings are revealed that help to build a meaningful model to predict weight loss automatically.
- Knowledge Tagging System on Math Questions via LLMs with Flexible Demonstration Retriever10
This paper explores automating the knowledge tagging for questions task using Large Language Models (LLMs), in response to the inability of prior encoding-based methods to deal with the hard cases which involve strong domain knowledge and complicated concept definitions.
- Exploring Query Understanding for Amazon Product Search10
This work conducts a comprehensive study and shares its year-long journey investigating how the query understanding service impacts Amazon Product Search, and proposes a query understanding-based multi-task learning framework for ranking.
- Alternately Optimized Graph Neural Networks10
This work proposes a new optimization framework for semi-supervised learning on graphs that can be conveniently solved by the alternating optimization algorithms, resulting in significantly improved efficiency.
- 10
- Characterizing Teacher Connections in Online Social Media10
A thorough and large-scale investigation of teacher connections in online social media, which is lacking in the literature, and compares the physical and virtual network of teachers on Pinterest using several graph-related metrics.
- 10
- 10
- Aligning large language models and geometric deep models for protein representation9
Experimental results reveal that GDMs incorporating both graph and 3D structural information align better with LLMs, larger LLMs demonstrate improved alignment capabilities, and protein rarity significantly impacts alignment performance.
- 9
- 9
- S patial CTD: A Large-Scale Tumor Microenvironment Spatial Transcriptomic Dataset to Evaluate Cell Type Deconvolution for Immuno-Oncology9
A large-scale spatial transcriptomic deconvolution benchmark dataset named SpatialCTD is introduced, encompassing 1.8 million cells and 12,900 pseudo spots from the human tumor microenvironment across the lung, kidney, and liver, and a graph neural network-based deconvolution method (i.e., GNNDeconvolver) is proposed.
- 9
- Exploiting Hierarchical Structures for Unsupervised Feature Selection9
This work provides a principled method to exploit hierarchical structures of features and proposes a novel framework HUFS, which utilizes the given hierarchical structures to help select features without labels.
- Price Recommendation on Vacation Rental Websites9
This paper identifies some important factors that influence the rental prices and proposes a framework that consists of Multi-Scale Affinity Propagation to cluster houses, Nash Equilibrium to remove unreasonable price and Linear Regression model with Normal Noise to predict the reasonable prices.
- Temporal dynamics in social trust prediction9
This paper model the dynamics of user preferences in two principled ways and incorporates these two types of temporal dynamics into traditional matrix factorization based social trust prediction model, which targets temporal smoothness and weight.
- 8
- 8
- Learning Multi-Level Dependencies for Robust Word Recognition8
A robust word recognition framework that captures multi-level sequential dependencies in noised sentences by employing a sequence-to-sequence model over characters of each word, whose output is given to a word-level bi-directional recurrent neural network.
- Reinforcement Learning for Online Information Seeking8
An overview of deep reinforcement learning for search, recommendation, and online advertising from methodologies to applications, review representative algorithms, and discuss some appealing research directions are given.
- Multi-Faceted Studies on Data Poisoning can Advance LLM Development7
It is argued that multi-faceted studies on data poisoning can advance LLM development and provide valuable insights into LLMs, particularly the interplay between data and model behavior, driving a deeper understanding of their underlying mechanisms.
- 7
- An Efficient Unsupervised Framework for Convex Quadratic Programs via Deep Unrolling7
It is shown that PDQP-net trained in this unsupervised manner can effectively approximate optimal QP solutions, indicating that using PDQP-net predictions to warm-start PDQP can achieve up to 45% acceleration on QP instances.
- Revisiting the Graph Reasoning Ability of Large Language Models: Case Studies in Translation, Connectivity and Shortest Path7
It is suggested that LLMs can fail to understand graph structures through text descriptions and exhibit varying performance for all three fundamental graph tasks: graph description translation, graph connectivity, and the shortest-path problem.
- Gene Expression and Protein Function7
This survey provides an overview of the various types of problems in this domain and the neural architectures that work for these data sets and point out existing deep learning work from other domains that has the potential to be applied to these domains.
- 6
- Whole Page Unbiased Learning to Rank6
A Bias Agnostic whole-page unbiased Learning to rank algorithm, named BAL, is proposed, to automatically find the user behavior model with causal discovery and mitigate the biases induced by multiple SERP features with no specific design.
- 6
- Large-Scale Graph Neural Networks: The Past and New Frontiers6
This tutorial aims to provide a systematic and comprehensive understanding of the challenges and state-of-the-art techniques for scaling GNNs, and to explore new ideas and developments in this rapidly evolving field.
- Towards Adversarial Learning6
This tutorial provides a comprehensive overview of the recent advances of adversarial learning, including both attack methods and defense methods, and presents DeepRobust, a PyTorch adversariallearning library which aims to build a comprehensive and easy-to-use platform to foster this research field.
- Defense Against Gradient Leakage Attacks via Learning to Obscure Data6
This paper proposes a new defense method to protect the privacy of clients' data by learning to obscure data and shows that this method obtains better privacy protection while preserving high accuracy compared with state-of-the-art methods.
- Recent Advances in Multimodal Educational Data Mining in K-12 Education6
This tutorial targets AI researchers and practitioners who are interested in applying state-of-the-art multimodal machine learning techniques to tackle some of the hard-core AIED tasks.
- Unveiling Mode Connectivity in Graph Neural Network5
It is uncovered that GNNs exhibit distinct non-linear mode connectivity, diverging from patterns observed in fully-connected networks or CNNs, and it is demonstrated that graph structure, rather than model architecture, dominates this behavior.
- 5
- SpatialCTD: a large-scale TME spatial transcriptomic dataset to evaluate cell type deconvolution for immuno-oncology5
This work introduces a large-scale spatial transcriptomic dataset named SpatialCTD, encompassing 1.8 million cells from the human tumor microenvironment across the lung, kidney, and liver, and introduces Graph Neural Network-based method GNNDeconvolver, which often outperforms existing state-of-the-art methods by a substantial margin, without requiring single-cell RNA-seq data.
- How does the Memorization of Neural Networks Impact Adversarial Robust Models?5
Benign Adversarial Training (BAT) is proposed which can facilitate adversarial training to avoid fitting "harmful" atypical samples and fit as more "benign" atYPical samples as possible and can achieve better clean accuracy vs. robustness trade-off than baseline methods, in benchmark datasets for image classification.
- Solving ESL Sentence Completion Questions via Pre-trained Neural Language Models5
This work proposes a neural framework to solve SC questions in English examinations by utilizing pre-trained language models and conducts extensive experiments on a real-world K-12 ESL SC question dataset, demonstrating the superiority of the model in terms of prediction accuracy.
- Multi-task Learning Based Online Dialogic Instruction Detection with Pre-trained Language Models5
A multi-task paradigm is proposed which enhances the ability to distinguish instances of different classes by enlarging the margin between categories via contrastive loss, and a strategy to fully exploit the misclassified examples during the training stage is designed.
- The Sixth International Workshop on Deep Learning on Graphs - Methods and Applications (DLG-KDD'21)5
This international workshop on "Deep Learning on Graphs: Method and Applications (DLG-KDD'21)" aims to bring together both academic researchers and industrial practitioners from different backgrounds and perspectives to the above challenges.
- 5
- 5
- 5
- Towards the Effect of Examples on In‐Context Learning: A Theoretical Case Study4
A probabilistic model extending from the Gaussian mixture model is introduced to exactly quantify the impact of pre‐training knowledge, label frequency and label noise on the prediction accuracy of ICL.
- Tabula: A Tabular Self-Supervised Foundation Model for Single-Cell Transcriptomics4
T ABULA is presented, a foundation model designed for single-cell transcriptomics, which integrates a novel tabular modeling objective and federated learning framework to enable privacy-preserving pretraining across decentralized datasets.
- 4
- Towards Fair Classification against Poisoning Attacks4
This work proposes a general and theoretically guaranteed framework which accommodates traditional defense methods to fair classification against poisoning attacks, and proves that the proposed defense framework obtains better robustness in terms of accuracy and fairness than representative baseline methods.
- Efficient Machine Learning on Large-Scale Graphs4
This tutorial will cover how to develop and run performant graph algorithms and graph neural network models with TigerGraph, a massively parallel platform for graph analytics, and its Machine Learning Workbench with PyTorch Geometric and DGL support.
- Road to the White House4
This paper collects data during the 2020 Democratic Party presidential primaries pertaining to the candidates and by analyzing this data, it highlights similarities and differences between these two main types of sources, detect the potential impact they have on each other, and understand how this impact relationship can change over time.
- Graph Feature Gating Networks4
This work proposes a general graph feature gating network (GFGN) based on the graph signal denoising problem and then correspondingly introduces three graph filters under GFGN to allow different levels of contributions from feature dimensions.
- 4
- 4
- 4
- Computing Distrust in Social Media4
Trust plays a crucial role in helping online users collect relevant and reliable information, and has been proven to be an effective way to mitigate information overload and credibility problems.
- 4
- 3
- 3
- 3
- Higher-order Structure Boosts Link Prediction on Temporal Graphs3
A Higher-order structure Temporal Graph Neural Network, which incorporates hypergraph representations into temporal graph learning, and achieves superior performance on dynamic link prediction while reducing memory costs by up to 50\% compared to existing methods.
- Sub-graph Based Diffusion Model for Link Prediction3
This paper aims to build a novel generative model for link prediction that treats link prediction between a pair of nodes as a conditional likelihood estimation of its enclosing sub-graph through a dedicated design to decompose the likelihood estimation process via the Bayesian formula.
- Enhancing ID and Text Fusion via Alternative Training in Session-based Recommendation3
AlterRec separates the training of ID and text, thereby avoiding the imbalance issue seen in naive fusion and designs a novel strategy to facilitate the interaction between the two modalities, enabling them to mutually learn from each other and integrate the text more effectively.
- PROPN: Personalized Probabilistic Strategic Parameter Optimization in Recommendations3
This paper forms the personalized probabilistic optimizing problem and compares its solution with deterministic and context-free solutions theoretically to show its superiority, and introduces a novel Personalized pRObabilistic strategic parameter optimizing Policy Network (PROPN) to solve the problem.
- Graph Neural Networks in Computer Vision3
This chapter provides a comprehensive overview of how GNNs are applied to various CV tasks, ranging from single image classification to crossmedia understanding, and provides a discussion of this rapidly growing field from a frontier perspective.
- SUMMER: Bias-aware Prediction of Graduate Employment Based on Educational Big Data3
A framework called SUMMER (biaS-aware gradUate eMployMEnt pRediction) is developed to predict students’ employment status and employment preference while considering biases and a bias-based regularization is designed to smooth the job market biases.
- Summary of Tutorials at The Web Conference 20213
This report summarizes the 23 tutorials hosted at The Web Conference 2021: nine lecture-style tutorials and 14 hands-on tutorials.
- 3
- Non-IID Graph Neural Networks3
This paper proposes a general non-IID graph neural network framework, i.e., Non-I ID-GNN, which can adapt any existing graph neuralnetwork model to generate a sample-specific model for this graph.
- Graph Convolutional Networks against Degree-Related Biases.3
Self-Supervised-Learning Degree-Specific GCN (SL-DSGCN) is proposed which handles the degree-related biases of GCNs from model and data aspects, and a self-supervised-learning algorithm that assigns pseudo labels with uncertainty scores on unlabeled nodes using a Bayesian neural network improves GCN accuracy dramatically for low-degree nodes.
- 3
- 3
- 3
- An Artificial Intelligence‐Enhanced Assessment Framework for Analyzing Middle School Science Students’ Written Responses2
A multi‐agent system with human‐AI collaboration, in which the AI identifies unclear cases and teachers review and refine uncertain scores shows the framework's potential to enhance classroom assessment by providing timely, reliable scores and feedback to inform teaching and learning.
- A LLM-Driven Multi-Agent Systems for Professional Development of Mathematics Teachers2
I-VIP (Intelligent Virtual Interactive Program), an intelligent tutoring platform for teacher professional development, driven by large language models (LLMs) and supported by multi-agent frameworks, is introduced.
- 2
- Deep Learning on Graphs: Methods and Applications (DLG-KDD2023)2
This international workshop on "Deep Learning on Graphs: Method and Applications (DLG-KDD'23)" aims to bring together both academic researchers and industrial practitioners from different backgrounds and perspectives to above challenges.
- 2
- Robust Graph Neural Networks2
This chapter introduces concepts and definitions of graph adversarial attacks and detail some representative adversarial attack methods on graphs, and discusses representative defense techniques against these adversarialattacks.
- 2
- 2
- Adversarial Robustness in Deep Learning: From Practices to Theories2
This tutorial provides a comprehensive overview on the recent advances of adversarial examples and their countermeasures, from both practical and theoretical perspectives, and introduces DeepRobust, a Pytorch adversarial learning library which aims to build a comprehensive and easy-to-use platform to foster this research field.
- 2
- 2
- MoC-KT: Mixture of Convolutions for Knowledge Tracing1
The proposed MoC-KT framework introduces multi-scale causal convolutional kernels and adaptive segmentation to disentangle learners’ long-term stable progression from short-term fluctuations, such as guessing or fatigue, and incorporates a distance-sensitive decay function to prioritize local dependencies while suppressing irrelevant distant interactions.
- Automated coding of content and pedagogical content knowledge of mathematics using a multi-agent large language model1
Results demonstrate that a multi-agent LLM framework has the potential to code open-ended responses at a level comparable to human coders, and discusses the implications for advancing automated assessment in teacher education and mathematics education.
- scLinguist: A pre-trained hyena-based foundation model for cross-modality translation in single-cell multi-omics1
Systematic benchmarking shows that scLinguist consistently outperforms state-of-the-art methods in predicting protein abundance from RNA across diverse biological contexts and further enables mechanistic and generalizable inference under simulated gene perturbations.
- 1
- Towards Better Benchmark Datasets for Inductive Knowledge Graph Completion1
This paper finds that the current procedure for constructing inductive KGC datasets inadvertently creates a shortcut that can be exploited even while disregarding the relational information, and proposes an alternative strategy for constructing inductive KGC datasets that helps mitigate the PPR shortcut.
- DANCE 2.0: Transforming single-cell analysis from black box to transparent workflow1
An automated and interpretable preprocessing platform featuring two key modules: the Method-Aware Preprocessing (MAP) module, which discovers optimal pipelines for task-specific methods via hierarchical search, and the Dataset-Aware Preprocessing (DAP) module, which recommends pipelines for new datasets via similarity-based matching to a reference atlas.
- The 2nd Workshop on Large Language Models for E-Commerce1
This workshop aims to bring together researchers and industry practitioners to explore both the limitations and opportunities of LLMs in e-commerce to foster collaboration, bridge the gap between academia and industry, and drive innovation in the application of LLMs to E-Commerce.
- Uncovering Problematic Designs Hindering Ubiquitous Cellular Emergency Services Access1
This study shows that, despite sufficient wireless signal coverage, users may still experience prolonged emergency call setup times, call initiation failures, or call drops due to flaws in the design of cellular emergency services.
- 1
- 1
- Enhancing Adversarial Training with Feature Separability1
A new concept of adversarial training graph (ATG) is introduced with which the proposed adversarialTraining with feature separability (ATFS) enables to coherently boost the intra- class feature similarity and increase inter-class feature variance.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Longitudinal modality prediction learns gene regulatory patterns: insights from a single-cell competition–
A purpose-built longitudinal multimodal benchmarking dataset that captures early hematopoietic differentiation and the largest single-cell data competition to date is generated, demonstrating that top-performing approaches outperform state-of-the-art methods, and how best-performing models captured biologically meaningful regulatory relationships between modalities.
- –
- Why Retrieval-Augmented Generation Fails: A Graph Perspective–
A model-internal study of retrieval-augmented generation that examines how retrieved evidence influences answer generation, and develops a graph-based error detection framework that uses attribution-graph topology features.
- –
- –
- The Landscape of Single-Cell Foundation Models: Design Principles, Applications, and Open Challenges–
- –
- Towards Understanding Link Predictor Generalizability Under Distribution Shifts–
This work introduces a novel splitting strategy, known as LPShift, which utilizes structural properties to induce a controlled distribution shift, and verifies its effect through empirical evaluation of SOTA LP models on 16 LPShift variants of original dataset splits, with results indicating drastic changes to model performance.
- –
- Machine Learning on Graphs in the Era of Generative Artificial Intelligence–
The first international Workshop on Machine Learning on Graphs in the Era of Generative Artificial Intelligence (MLoG-GenAI) is held in connection with the 31st ACM Conference on Knowledge Discovery and Data Mining, which provides a venue to gather academic researchers and industry practitioners to discuss and picture the development of GML in the new Gen-AI era.
- Large Language Model for E-Commerce Workshop–
This workshop aims to bring together researchers and industry practitioners to explore the opportunities and limitations of LLMs in e-commerce, and foster collaboration, bridge the gap between academia and industry, and drive innovation in applying LLMs to E-Commerce.
- Subgraph Generation for Generalizing on Out-of-Distribution Links–
This work proposes FLEX as a GGM framework which leverages two mechanism: (1) structurally-conditioned graph generation, and (2) adversarial co-training between an auto-encoder and GNN.
- GraphGhost: Tracing Structures Behind Large Language Models–
GraphGhost is proposed, a graph-based framework that models internal token interactions and neuron activations in LLMs as graphs and shows that graph structural properties are closely associated with influential tokens and neuron nodes, and that perturbations to structurally critical nodes lead to measurable changes in reasoning behavior.
- –
- –
- –
- The 1st International Workshop on Graph Foundation Models (GFM)–
The GFM workshop focuses on two critical questions: (1) How can the underlying capabilities of existing foundation models be effectively applied to graph data and (2) What foundational principles should guide the creation of models tailored to the graph domain.
- –
- –
- –
- MEM-GAN: A Pseudo Membrane Generator for Single-cell Imaging in Fluorescent Microscopy–
Inspired by advanced style transfer techniques in computer vision, Mem-GAN styles the content of the cells with missing or weak membranes into cells with integrated membrane staining, to generate high-quality membranes for cells with missing or weak membranes.
- ACM WSDM 2022 report–
The 15th ACM International Conference on Web Search and Data Mining (WSDM 2022) was successfully held on 2/21 - 2/25/2022 online with organization and program details and statistics from Conference Chairs and Program Committee Chairs, as well as a message from the W SDM Steering Committee Chair to call for future WSDM conferences in the US (2024 and Europe (2025).
- Accepted Tutorials at The Web Conference 2022–
This paper summarizes the content of the 20 tutorials that have been given at The Web Conference 2022: 85% of these tutorials are lecture style, and 15% are hands on.
- Learning from Imbalanced Crowdsourced Labeled Data–
A new framework, named “Learning from Imbalanced Crowdsourced Labeled Data” (ICED), is proposed, which simultaneously infers true labels from imbalanced crowdsourced labeled data and achieves high accuracy on downstream tasks such as classification.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Deep Learning on Graphs: An Introduction–
This chapter discusses why and how real-world data is represented as graphs, and introduces what content this book will cover, and provides guidance about who should read this book.
- –
- –
- –
- –
- –
- Exploring Personal Attributes from Unprotected Interactions–
This paper provides a way to mitigate the data sparsity challenge and proposes a novel attribute prediction framework using only the unprotected interactions, which demonstrates that the proposed framework can predict users' personal attributes.
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.