IJCAI 2024 Papers — Page 6
International Joint Conference on Artificial Intelligence · 790 papers
Multi-Modality Spatio-Temporal Forecasting via Self-Supervised Learning
Jiewen Deng (Southern University of Science and Technology), Xuan Song (Southern University of Science and Technology)
Graph Neural NetworkContrastive LearningMultimodalityTime Series
🎯 What it does: Propose a multi-modal spatiotemporal prediction framework called MoSSL based on self-supervised learning, integrating multi-modal data augmentation, global self-supervised learning, and modal self-supervised learning to excavate and quantify spatial, temporal, and modal heterogeneity.
Multi-Relational Graph Attention Network for Social Relationship Inference from Human Mobility Data
Guangming Qin (Ocean University of China), Junyu Dong (Ocean University of China)
Graph Neural NetworkGraph
🎯 What it does: This paper proposes a Multi-Relationship Graph Attention Network (MRGAN) for inferring social relationships from human mobility trajectory data.
Multi-scale Context-Aware Networks Based on Fragment Association for Human Activity Recognition
Zhiqiong Wang (Northeastern University), Junchang Xin (Northeastern University)
RecognitionComputational EfficiencyConvolutional Neural NetworkTransformerTime Series
🎯 What it does: Proposes a lightweight multi-scale context-aware network (MSC-CA) for sensor-based human activity recognition, consisting of an internal feature extraction module and a context-aware module.
Multi-TA: Multilevel Temporal Augmentation for Robust Septic Shock Early Prediction
Hyunwoo Sohn (North Carolina State University), Min Chi (North Carolina State University)
ClassificationRecurrent Neural NetworkTransformerGenerative Adversarial NetworkTabularBiomedical DataElectronic Health Records
🎯 What it does: Propose a multi-layer temporal enhancement framework called Multi-TA, combining t-BERT representation learning and constrained worst-case transformation to achieve temporal robustness for early prediction of septic shock.
MultifacetEval: Multifaceted Evaluation to Probe LLMs in Mastering Medical Knowledge
Yuxuan Zhou (Tsinghua University), Ji Wu (Tsinghua University)
Large Language ModelPrompt EngineeringBiomedical DataBenchmarkChain-of-Thought
🎯 What it does: Propose a multi-faceted evaluation framework, MultifactEval, which designs four evaluation dimensions (comparison, correction, differential diagnosis, verification) based on the same medical knowledge point, and constructs two multi-faceted medical evaluation datasets, MultiDiseK and MultiMedQA, to systematically assess the depth and breadth of LLMs' mastery of medical knowledge.
Multimodal Representation Distribution Learning for Medical Image Segmentation
Chao Huang (Sun Yat-sen University), Zhihua Wang (Shenzhen MSU-BIT University)
SegmentationConvolutional Neural NetworkTransformerVision Language ModelMultimodalityBiomedical Data
🎯 What it does: This paper proposes a multimodal medical image segmentation method that enhances segmentation performance by fusing medical text annotations with image features.
Multiplex Graph Representation Learning via Bi-level Optimization
Yudi Huang (Guangxi Normal University), Xiaofeng Zhu (Guangxi Normal University)
OptimizationRepresentation LearningGraph Neural NetworkGraph
🎯 What it does: This paper proposes a multi-view graph representation learning framework named MGBO based on two-layer optimization, which addresses the edge hunger problem by learning a self-expressive matrix to capture global positive and negative relationships.
NanoAdapt: Mitigating Negative Transfer in Test Time Adaptation with Extremely Small Batch Sizes
Shiji Zhao (Nanjing University of Aeronautics and Astronautics), Sheng-Jun Huang (Nanjing University of Aeronautics and Astronautics)
Domain AdaptationImage
🎯 What it does: Propose NanoAdapt to address the negative transfer problem in Test Time Adaptation (TTA) under extremely small batch sizes (e.g., N=1).
Natural Language Decomposition and Interpretation of Complex Utterances
Harsh Jhamtani (Microsoft), Benjamin Van Durme (Microsoft)
Explainability and InterpretabilityAI Code AssistantTransformerLarge Language ModelTextBenchmarkRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Propose a method called DECINT, which achieves interpretation of complex user requests by first decomposing complex natural language instructions into a series of simple natural language steps, and then progressively generating programs using a language-to-program parser.
Natural Language-centered Inference Network for Multi-modal Fake News Detection
Qiang Zhang (University of Science and Technology of China), Zheng-Jun Zha (University of Science and Technology of China)
ClassificationTransformerLarge Language ModelPrompt EngineeringVision Language ModelImageTextMultimodalityRetrieval-Augmented Generation
🎯 What it does: This paper proposes a Natural Language-Centric Reasoning Network (NLIN), which converts news images and background knowledge into text and leverages LLM for reasoning, constructing a large-scale Chinese multimodal fake news dataset CFND across multiple platforms and domains.
Navigating Continual Test-time Adaptation with Symbiosis Knowledge
Xu Yang (Xidian University), Cheng Deng (Xidian University)
Domain AdaptationContrastive LearningImage
🎯 What it does: Propose a dual-stream network in continual testing temporal adaptation, separately optimizing BN layers and full parameters with independent optimization, combined with weight averaging and soft parameter alignment to achieve symbiosis between source knowledge and target knowledge.
Negative Prompt Driven Complementary Parallel Representation for Open-World 3D Object Retrieval
Yang Xu (Tsinghua University), Yue Gao (Tsinghua University)
RetrievalRepresentation LearningGraph Neural NetworkPrompt EngineeringAuto EncoderMultimodalityPoint CloudMesh
🎯 What it does: This paper proposes the Negative Prompt Driven Complementary Parallel Representation (NPCP) framework, which generates bidirectional embeddings using negative prompts to achieve more robust retrieval in open-world 3D object retrieval.
Negative-Binomial Randomized Gamma Dynamical Systems for Heterogeneous Overdispersed Count Time Sequences
Rui Huang (Great Bay University), Heinz Koeppl (Technische Universitaet Darmstadt)
Explainability and InterpretabilityTime Series
🎯 What it does: This paper proposes a negative binomial randomized gamma dynamics system for modeling heteroscedastic count time series, achieving an interpretable transition structure.
NegativePrompt: Leveraging Psychology for Large Language Models Enhancement via Negative Emotional Stimuli
Xu Wang (School of Artificial Intelligence, Jilin University), Yuan Wu (School of Artificial Intelligence, Jilin University)
TransformerLarge Language ModelPrompt EngineeringTextBenchmark
🎯 What it does: Propose the NegativePrompt strategy, design and test 10 negative emotion stimuli, combine psychological theories (cognitive dissonance, social comparison, stress and coping) to enhance LLM prompts, and conduct experiments on five mainstream LLMs (Flan-T5-Large, Vicuna, Llama 2, ChatGPT, GPT-4).
NELLIE: A Neuro-Symbolic Inference Engine for Grounded, Compositional, and Explainable Reasoning
Nathaniel Weir, Benjamin Van Durme (Johns Hopkins University)
Explainability and InterpretabilityTransformerLarge Language ModelTextRetrieval-Augmented GenerationChain-of-Thought
🎯 What it does: Proposes a neural symbolic reasoning engine NELLIE based on large language models, which constructs interpretable proof trees through backward chaining search in natural language fact corpora to answer multiple-choice questions.
No Regularization Is Needed: Efficient and Effective Incomplete Label Distribution Learning
Xiang Li (Nanjing University of Aeronautics and Astronautics), Songcan Chen (Nanjing University of Aeronautics and Astronautics)
Classification
🎯 What it does: This paper proposes a weighted incomplete label distribution learning method (WInLDL) without explicit regularization, modeling real-world scenarios where labels with small degrees are more prone to missingness.
Nonconvex Multiview Subspace Clustering Framework with Efficient Method Designs and Theoretical Analysis
Zhi Wang (Southwest University), Tao Jia (Southwest University)
OptimizationComputational EfficiencyRepresentation LearningMultimodalityBenchmark
🎯 What it does: This paper proposes a multi-view subspace clustering framework based on non-convex ℓq regularization, and provides an efficient solution algorithm along with convergence analysis;
Nonparametric Detection of Gerrymandering in Multiparty Plurality Elections
Dariusz Stolicki (Jagiellonian University), Stanisław Szufa (AGH University)
Anomaly DetectionTabular
🎯 What it does: Proposes a nonparametric method for detecting district delineation bias in multi-party systems and partially contentious districts.
Normative Testimony and Belief Functions: A Formal Theory of Norm Learning
Taylor Olson (Northwestern University), Kenneth D. Forbus (Northwestern University)
🎯 What it does: This paper proposes a formalized norm learning framework for learning the internal normative beliefs of a group from normative testimonies.
Nukplex: An Efficient Local Search Algorithm for Maximum K-Plex Problem
Rui Sun (Northeast Normal University), Minghao Yin (Northeast Normal University)
OptimizationComputational EfficiencyGraphBenchmark
🎯 What it does: Proposed a local search algorithm called Nukplex for solving the maximum k-plex problem.
OD-DETR: Online Distillation for Stabilizing Training of Detection Transformer
Shengjian Wu (East China Normal University), Qingli Li (East China Normal University)
Object DetectionKnowledge DistillationConvolutional Neural NetworkTransformerImage
🎯 What it does: Proposed an online distillation framework OD-DETR to stabilize and accelerate the training of DETR series models;
Off-Agent Trust Region Policy Optimization
Ruiqing Chen (Peking University), Yaodong Yang (Peking University)
Reinforcement Learning
🎯 What it does: Proposed an offline agent (Off-Agent) strategy optimization framework that allows multiple agents to selectively share their experiences during learning, and provides theoretical guarantees of approximate monotonic improvement.
Offline Policy Learning via Skill-step Abstraction for Long-horizon Goal-Conditioned Tasks
Donghoon Kim (Sungkyunkwan University), Honguk Woo (Sungkyunkwan University)
Robotic IntelligenceReinforcement LearningAuto EncoderWorld ModelSequential
🎯 What it does: Built a framework named GLVSA in an offline environment, learning long-term goal-conditioned strategies through skill-step abstraction, and achieving model-driven trajectory expansion and modular policy hierarchies.
Offline Reinforcement Learning with Behavioral Supervisor Tuning
Padmanaba Srinivasan (Imperial College London), William Knottenbelt (Imperial College London)
Reinforcement LearningBenchmark
🎯 What it does: Propose an offline reinforcement learning algorithm called TD3-BST, which trains a Morse neural network to estimate uncertainty and dynamically uses it as a behavior supervisor to guide the policy in selecting actions within the dataset-supported range.
On the Computation of Example-Based Abductive Explanations for Random Forests
Gilles Audemard (Univ. Artois), Nicolas Szczepanski (Univ. Artois)
ClassificationExplainability and InterpretabilityTabular
🎯 What it does: Proposed an example-based inductive explanation method for random forests, along with an algorithm to generate subset-minimal, optimal anchored explanations from instances.
On the Effects of Fairness to Adversarial Vulnerability
Cuong Tran (University of Virginia), Ferdinando Fioretto (University of Virginia)
ClassificationAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: This paper studies the impact of fairness constraints on model adversarial robustness, reveals the contradiction between the two, and proposes a mitigation scheme using bounded Ramp loss.
On the Logic of Theory Change Iteration of KM-Update, Revised
Liangda Fang (Jinan University), Hai Wan (Sun Yat-sen University)
🎯 What it does: Proposed an iterative update logic based on the Darwiche–Pearl belief state framework, addressing the limitations of the Katsuno–Mendelzon update axioms;
On the Power and Limitations of Examples for Description Logic Concepts
Balder ten Cate (University of Amsterdam), Ana Ozaki (University of Bergen)
🎯 What it does: The study investigates the feasibility and complexity of using finite labeled examples to uniquely characterize concepts in different description logic languages, completing an almost complete classification and providing constructions and negative results.
On the Pursuit of EFX for Chores: Non-existence and Approximations
Vasilis Christoforidis (Archimedes / Athena RC), Christodoulos Santorinaios (Archimedes / Athena RC)
Optimization
🎯 What it does: Investigated the fairness issue of allocating indivisible chores to agents, proving that EFX allocations may not exist under general cost functions and the decision problem is NP-hard, while guaranteeing the existence of EFX when the number of items is less than the number of agents + 2, and providing an approximate algorithm under additive costs.
On Using Admissible Bounds for Learning Forward Search Heuristics
Carlos Núñez-Molina (University of Granada), Juan Fernandez-Olivares (University of Granada)
OptimizationGraphBenchmark
🎯 What it does: Propose a statistical learning framework that utilizes verifiable reachability heuristics as a lower bound for truncated Gaussian distributions to improve the learning of forward search heuristics;
One-step Spiking Transformer with a Linear Complexity
Xiaotian Song (Sichuan University), Yanan Sun (Sichuan University)
ClassificationRecognitionSpiking Neural NetworkTransformerImage
🎯 What it does: Proposed One-step Spiking Transformer (OST), achieving a spiking Transformer with single-time-step processing and linear complexity;
Online Combinatorial Optimization with Group Fairness Constraints
Negin Golrezaei (MIT), Fransisca Susan (MIT)
OptimizationTabular
🎯 What it does: Propose a general framework that integrates group fairness constraints into online combinatorial optimization problems (such as product ranking, submodular maximization, shortest path), and provide achievable approximate algorithms;
Online Learning of Capacity-Based Preference Models
Margot Herin (Sorbonne University, CNRS), Nataliya Sokolovska (Sorbonne University, CNRS)
Optimization
🎯 What it does: This paper proposes an online learning algorithm for sparse capacity, applied to the Choquet integral and multilinear models in multicriteria decision making.
Online Learning of Partitions in Additively Separable Hedonic Games
Saar Cohen (Bar-Ilan University), Noa Agmon (Bar-Ilan University)
Optimization
🎯 What it does: This paper studies the coalition formation problem in additive separable hedonic games based on online learning, proposing an algorithm to optimize social cost and minimize static or dynamic regret.
Online Learning with Off-Policy Feedback in Adversarial MDPs
Francesco Bacchiocchi (Politecnico di Milano), Nicola Gatti (Politecnico di Milano)
Reinforcement Learning
🎯 What it does: This paper studies online learning in adversarial Markov decision processes, particularly the challenge of learning using offline feedback. The learner selects a strategy, but the environment is explored by a fixed, possibly unknown strategy (called a colleague strategy).
Online Sampling and Decision Making with Low Entropy
Mohammad Taghi Hajiaghayi (University of Maryland), Jan Olkowski (University of Maryland)
Optimization
🎯 What it does: This paper designs a low-entropy random sequence distribution that can accept k maximum values with nearly optimal competitive ratio (1 - O(√logk/k)) in the free sequence multi-choice secretary problem, where k's range is extended to log n / log log n;
Online Submodular Maximization via Adaptive Thresholds
Zhengchen Yang (Nanjing University of Aeronautics & Astronautics), Jiping Zheng (Nanjing University)
OptimizationVideoTextTabular
🎯 What it does: Propose an online adaptive threshold algorithm ONLINEADAPTIVE for incrementally maximizing submodular functions in large streaming data under cardinality constraints.
Optimal Auction Design with User Coupons in Advertising Systems
Xiaodong Liu (Renmin University of China), Weiran Shen (Renmin University of China)
Recommendation SystemOptimizationFinance Related
🎯 What it does: In online ad auctions, user coupons are considered, and coupon strategies, allocation rules, and payment rules are jointly designed to provide a revenue-optimal mechanism.
Optimal Extended Formulations from Optimal Dynamic Programming Algorithms
Mateus de Oliveira Oliveira (Stockholm University), Wim Van den Broeck (University of Bergen)
Optimization
🎯 What it does: This paper establishes a tight connection between an optimal dynamic programming (DP) algorithm for solving the vertex subset problem (VSP) and the extension complexity of the corresponding polytope, and proposes a theoretical framework that directly translates the size of the DP table into a polytope extension formula.
Optimal Graph Learning and Nuclear Norm Maximization for Deep Cross-Domain Robust Label Propagation
Wei Wang (Sun Yat-sen University), Xiaochun Cao (Sun Yat-sen University)
ClassificationDomain AdaptationConvolutional Neural NetworkGraph Neural NetworkImage
🎯 What it does: This paper proposes a Deep Cross-Domain Robust Label Propagation (DCDRLP) framework, combining optimal graph learning and label propagation with nuclear norm maximization to achieve cross-domain adaptation.
Optimisation and Approximation in Abstract Argumentation: The Case of Stable Semantics
Matthias Thimm (University of Hagen)
OptimizationComputational EfficiencyBenchmark
🎯 What it does: This paper studies two soft stable semantics (k-stable and k*-stable) and analyzes the complexity and approximation algorithms of the corresponding optimization problems;
Optimizing Prosumer Policies in Periodic Double Auctions Inspired by Equilibrium Analysis
Bharat Manvi (TCS Research), Easwar Subramanian (TCS Research)
OptimizationReinforcement Learning
🎯 What it does: This paper studies how prosumers, who both buy and sell in periodic double auctions (PDA), formulate optimal bidding strategies, and proposes an equilibrium solution based on Markov games;
Optimizing Viscous Democracy
Ben Armstrong (University of Waterloo), Nimrod Talmon (Ben Gurion University)
OptimizationGraph
🎯 What it does: The study introduces a viscosity parameter into liquid democracy, defines viscous democracy, and analyzes its impact on collective decision accuracy under different network structures and expert distribution scenarios.
Ordinal Maximin Guarantees for Group Fair Division
Pasin Manurangsi (Google Research), Warut Suksompong (National University of Singapore)
Optimization
🎯 What it does: Study the ordinal maximin share (1-out-of-k MMS) guarantee in fair division of indivisible items, providing asymptotically tight upper and lower bounds for the minimal k (i.e., p MMS), and giving exact non-asymptotic results in two special cases.
OSIC: A New One-Stage Image Captioner Coined
Bo Wang (Hefei University of Technology), Meng Wang (Hefei University of Technology)
GenerationTransformerReinforcement LearningVision Language ModelMultimodality
🎯 What it does: Propose a one-stage image captioning model named OSIC, which extracts multi-layer image features using Swin Transformer, and achieves end-to-end training through dynamic multi-perspective embedding and dual-dimensional non-local refinement, eliminating the task information gap in traditional two-stage models.
OTOcc: Optimal Transport for Occupancy Prediction
Pengteng Li (Shenzhen University), Guang Zhou (Deeproute Inc)
Autonomous DrivingOptimizationComputational EfficiencyPoint Cloud
🎯 What it does: Proposes the OTOcc framework for 3D occupancy prediction by modeling the semantic mapping from pixels to voxels as an Optimal Transport problem;
OUCopula: Bi-Channel Multi-Label Copula-Enhanced Adapter-Based CNN for Myopia Screening Based on OU-UWF Images
Yang Li (Fudan University), Xingtao Zhou (Fudan University)
Convolutional Neural NetworkBiomedical Data
🎯 What it does: This paper achieves multi-label joint prediction of spherical equivalent and axis length on binocular UWF retinal images using a dual-channel adapter and Copula-enhanced CNN to improve myopia screening accuracy.
P2P: Transforming from Point Supervision to Explicit Visual Prompt for Object Detection and Segmentation
Guangqian Guo (Northwestern Polytechnical University), Shan Gao (Northwestern Polytechnical University)
Object DetectionSegmentationTransformerPrompt EngineeringImage
🎯 What it does: Propose a point-supervised object detection and segmentation framework called P2P, which converts point annotations into visual prompts, uses the visual foundation model SAM to generate pseudo-labels, and then re-trains the fully supervised detection/segmentation network with these pseudo-labels.
PACIA: Parameter-Efficient Adapter for Few-Shot Molecular Property Prediction
Shiguang Wu (Tsinghua University), Quanming Yao (Tsinghua University)
Meta LearningDrug DiscoveryGraph Neural NetworkBiomedical Data
🎯 What it does: Proposed PACIA, a parameter-efficient GNN adapter for few-shot molecular property prediction;
ParaILP: A Parallel Local Search Framework for Integer Linear Programming with Cooperative Evolution Mechanism
Peng Lin (Key Laboratory of System Software (Chinese Academy of Sciences) and State Key Laboratory of Computer Science, Institute of Software, Chinese Academy of Sciences), Shaowei Cai (Key Laboratory of System Software (Chinese Academy of Sciences) and State Key Laboratory of Computer Science, Institute of Software, Chinese Academy of Sciences)
OptimizationBenchmark
🎯 What it does: Proposes ParaILP—a parallel local search framework for general integer linear programming (ILP)—that can quickly find high-quality feasible solutions in multi-core environments.
Parameterized Analysis of Bribery in Challenge the Champ Tournaments
Juhi Chaudhary (TIFR), Meirav Zehavi (Ben Gurion University Of Negev)
Optimization
🎯 What it does: This paper studies the algorithm and complexity issues of persuasion (bribery) under budget constraints in a 'challenge champion' tournament format to increase the winning probability of designated players, presenting results including weak NP-hardness, W[1]-hardness, and several fixed-parameter tractable outcomes.
Parameterized Complexity of Kidney Exchange Revisited
Úrsula Hébert-Johnson (UC Santa Barbara), Vaishali Surianarayanan (UC Santa Barbara)
OptimizationGraph
🎯 What it does: This paper studies the parameterized complexity of the kidney exchange problem, particularly addressing two unresolved issues from IJCAI '18 and IJCAI '22, proving that the kidney exchange problem is FPT when parameterized by the number of vertex types, and W[1]-hard when parameterized by treewidth.
Pareto Inverse Reinforcement Learning for Diverse Expert Policy Generation
Woo Kyung Kim (Sungkyunkwan University), Honguk Woo (Sungkyunkwan University)
Autonomous DrivingReinforcement LearningDiffusion modelSequential
🎯 What it does: Propose an inverse reinforcement learning framework, ParIRL, which generates a multi-objective Pareto optimal strategy set from only two expert datasets.
ParsNets: A Parsimonious Composition of Orthogonal and Low-Rank Linear Networks for Zero-Shot Learning
Jingcai Guo (Hong Kong Polytechnic University), Song Guo (Hong Kong University of Science and Technology)
ClassificationComputational EfficiencyRepresentation LearningMultimodalityBenchmark
🎯 What it does: Propose a ParsNets framework that decomposes visual-semantic mapping into multiple sparse-activated linear subnetworks, achieving lightweight design and deployability on devices for zero-shot learning.
Partial Optimal Transport Based Out-of-Distribution Detection for Open-Set Semi-Supervised Learning
Yilong Ren (University of Science and Technology of China), S. Kevin Zhou (University of Science and Technology of China)
Domain AdaptationAnomaly DetectionOptimizationConvolutional Neural NetworkImage
🎯 What it does: Model OOD detection in open semi-supervised learning as a partial optimal transport problem, design a binary classifier based on quality scores, and jointly train it with existing SSL frameworks (e.g., FixMatch) to achieve end-to-end OOD detection and classification;
PDENNEval: A Comprehensive Evaluation of Neural Network Methods for Solving PDEs
Ping Wei (Sun Yat-sen University), Qingsong Zou (Sun Yat-sen University)
BenchmarkPhysics Related
🎯 What it does: This paper proposes PDENNEval, a unified benchmark platform for evaluating the performance of 12 neural network methods on 19 multidisciplinary PDE problems.
PEACH: Pretrained-Embedding Explanation across Contextual and Hierarchical Structure
Feiqi Cao (University of Sydney), Hyunsuk Chung (FortifyEdge)
ClassificationExplainability and InterpretabilityTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: Propose PEACH, which constructs decision trees using context embeddings from any pre-trained language model to provide global and local interpretable hierarchical explanations for text classification.
Peptide Vaccine Design by Evolutionary Multi-Objective Optimization
Dan-Xuan Liu (Nanjing University), Chao Qian (Nanjing University)
OptimizationDrug DiscoveryBiomedical Data
🎯 What it does: Propose a peptide vaccine design framework PVD-EMO based on evolutionary multi-objective optimization, converting the original constrained optimization problem into a bi-objective problem and using MOEA for search;
Personalized Federated Learning for Cross-City Traffic Prediction
Yu Zhang (Shandong University), Lizhen Cui (Shandong University)
Federated LearningSafty and PrivacyRecurrent Neural NetworkGraph Neural NetworkTime Series
🎯 What it does: Proposes a personalized federated learning framework named pFedCTP for cross-city traffic prediction, protecting data privacy of source and target cities.
Personalized Heart Disease Detection via ECG Digital Twin Generation
Yaojun Hu (Zhejiang University), Jian Wu (Zhejiang University)
Anomaly DetectionConvolutional Neural NetworkTransformerGenerative Adversarial NetworkTime SeriesBiomedical DataElectrocardiogram
🎯 What it does: Proposes a proactive learning framework called LAVQ-Editor, which enhances the sensitivity and personalized diagnosis of cardiac disease detection models by generating personalized ECG digital twins.
Perturbation Guiding Contrastive Representation Learning for Time Series Anomaly Detection
Liaoyuan Tang (Northwestern Polytechnical University), Feiping Nie (Northwestern Polytechnical University)
Anomaly DetectionRepresentation LearningConvolutional Neural NetworkContrastive LearningTime Series
🎯 What it does: Propose a time series anomaly detection method called PCRTA based on perturbation-guided contrastive learning.
PHSIC against Random Consistency and Its Application in Causal Inference
Jue Li (Shanxi University), Saixiong Liu (Shanxi University)
Tabular
🎯 What it does: Propose Pure HSIC (PHSIC) to eliminate the random consistency of HSIC and apply it to causal inference models.
Physics-Informed Neural Networks: Minimizing Residual Loss with Wide Networks and Effective Activations
Nima Hosseini Dashtbayaz (University of Western Ontario), Charles X. Ling (University of Western Ontario)
OptimizationPhysics Related
🎯 What it does: This paper investigates the properties of residual loss in physics-informed neural networks (PINNs), deriving that wide networks can globally minimize residual loss, and pointing out that the higher-order derivatives of activation functions must be bijective to enhance expressiveness, thereby verifying the effectiveness of periodic activation functions (e.g., sine) in solving first- to second-order PDEs.
Physics-Informed Trajectory Prediction for Autonomous Driving under Missing Observation
Haicheng Liao (University of Macau), Chengzhong Xu
Autonomous DrivingTime SeriesPhysics Related
🎯 What it does: Propose a two-stage physics-informed trajectory prediction framework, including a Wavelet Reconstruction Network for recovering missing observations and a Wave Fusion Encoder for interaction modeling, while ensuring the kinematic feasibility of predicted trajectories through a kinematic bicycle model.
Pluggable Watermarking of Deepfake Models for Deepfake Detection
Han Bao (Zhejiang University), Wenzhi Chen (Zhejiang University)
GenerationAnomaly DetectionConvolutional Neural NetworkImage
🎯 What it does: Proposed a pluggable, low-cost active model watermark framework to embed extractable watermarks into the decoders of trained Deepfake generation models, enabling detection of generated images;
Pointsoup: High-Performance and Extremely Low-Decoding-Latency Learned Geometry Codec for Large-Scale Point Cloud Scenes
Kang You (Nanjing University of Aeronautics and Astronautics), Dandan Ding (Hangzhou Normal University)
CompressionGraph Neural NetworkTransformerPoint Cloud
🎯 What it does: Proposed an efficient learning-based geometric encoder-decoder called Pointsoup for compressing and rapidly decoding large-scale point clouds.
PointTFA: Training-Free Clustering Adaption for Large 3D Point Cloud Models
Jinmeng Wu (University of Science and Technology of China), Hanyu Hong (University of Science and Technology of China)
ClassificationTransformerVision Language ModelPoint Cloud
🎯 What it does: Proposes PointTFA, a training-agnostic clustering adaptation on a large-scale 3D point cloud model (ULIP), to enhance zero-shot and few-shot classification performance.
Polynomial Time Presolve Algorithms for Rotation-Based Models Solving the Robust Stable Matching Problem
Sulian Le Bozec-Chiffoleau (Institut Mines Telecom Atlantique), Gilles Simonin (Institut Mines Telecom Atlantique)
OptimizationGraph
🎯 What it does: This paper addresses the Robust Stable Matching (RSM) problem by designing polynomial-time preprocessing algorithms and generalizing them to many-to-many matching instances; it also presents a constraint programming model based on rotation graphs;
Popular and Dominant Matchings with Uncertain and Multimodal Preferences
Gergely Csáji (HUN-REN KRTK KTI)
OptimizationComputational EfficiencyMultimodality
🎯 What it does: Studies the existence and algorithmic issues of popular matchings and dominant matchings in one-sided and two-sided matching markets under uncertain, multi-layered, and robust preference models.
Population-Based Diverse Exploration for Sparse-Reward Multi-Agent Tasks
Pei Xu (CRISE, Institute of Automation, Chinese Academy of Sciences), Kaiqi Huang (CRISE, Institute of Automation, Chinese Academy of Sciences)
Reinforcement LearningBenchmark
🎯 What it does: This paper proposes a method to maintain and guide the exploration of a diverse joint strategy population in multi-agent tasks with sparse rewards.
PoRank: A Practical Framework for Learning to Rank Policies
Pengjie Gu (Nanyang Technological University), Bo An (Nanyang Technological University)
TransformerTime Series
🎯 What it does: Proposes the PoRank framework, which utilizes a cross-policy Transformer for policy comparison and generates weak labels through crowdsourced OPE estimators, enabling the learning of policy ranking without online deployment.
Position Debiasing Fine-Tuning for Causal Perception in Long-Term Dialogue
Shixuan Fan (Huazhong University of Science and Technology), Dangyang Chen (Ping An Property and Casualty Insurance Company of China Ltd)
Explainability and InterpretabilityData-Centric LearningTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: Propose a framework named CPD that utilizes local position awareness and causal perturbation methods to extract causally related sentences from dialogue history, and significantly alleviates positional bias by enhancing the causal awareness of LLMs through IRL and MTE losses during the fine-tuning phase.
PPTFormer: Pseudo Multi-Perspective Transformer for UAV Segmentation
Deyi Ji (University of Science and Technology of China), Feng Zhao (University of Science and Technology of China)
SegmentationTransformerImage
🎯 What it does: Proposes PPTFormer, a pseudo-multi-perspective Transformer network, which enables multi-perspective learning for UAV image semantic segmentation by generating pseudo perspectives in the absence of multi-perspective annotated data.
Practical Anytime Algorithms for Judicious Partitioning of Active Directory Attack Graphs
Yumeng Zhang (University of Adelaide), Hung Nguyen (University of Adelaide)
OptimizationSafty and PrivacyGraph
🎯 What it does: This paper proposes a new graph theory problem—maximizing the number of disconnected pairs between a source and target node by edge deletion under a budget constraint, and applies it to security reinforcement of the Windows Active Directory (AD) system;
Practical Hybrid Gradient Compression for Federated Learning Systems
Sixu Hu (National University of Singapore), Bingsheng He (National University of Singapore)
CompressionFederated LearningSafty and PrivacyImageText
🎯 What it does: Proposed a hybrid gradient compression framework HGC compatible with secure FL, which can efficiently compress gradients in both uplink and downlink communications, significantly reducing communication overhead.
PRASS: Probabilistic Risk-averse Robust Learning with Stochastic Search
Tianle Zhang (University of Liverpool), Wenjie Ruan (University of Liverpool)
ClassificationAdversarial AttackImage
🎯 What it does: Propose PRASS—a probability risk-averse robust learning framework based on EVaR, which ensures robustness against both natural and random perturbations.
Pre-DyGAE: Pre-training Enhanced Dynamic Graph Autoencoder for Occupational Skill Demand Forecasting
Xi Chen (University of Science and Technology of China), Hui Xiong (Hong Kong University of Science and Technology (Guangzhou))
Graph Neural NetworkAuto EncoderContrastive LearningTextTime SeriesFinance Related
🎯 What it does: Propose a method based on a pre-trained enhanced dynamic graph autoencoder (Pre-DyGAE) for predicting skill demand from a career perspective.
Pre-training General User Representation with Multi-type APP Behaviors
Yuren Zhang (University of Science and Technology of China), Yang Yu (University of Science and Technology of China)
Recommendation SystemTransformerSupervised Fine-TuningAuto EncoderContrastive LearningTabularSequential
🎯 What it does: Proposed a Multi-Type App Usage Fusion Network (MAFN) that obtains general user representations through pre-training and achieves significant improvements on three types of downstream tasks.
Predictive Accuracy-Based Active Learning for Medical Image Segmentation
Jun Shi (University of Science and Technology of China), Bing Yan (University of Science and Technology of China)
SegmentationConvolutional Neural NetworkBiomedical Data
🎯 What it does: Propose a prediction accuracy-based active learning method PAAL for medical image segmentation
Predictive Modeling with Temporal Graphical Representation on Electronic Health Records
Jiayuan Chen (Ohio State University), Ping Zhang (Ohio State University)
ClassificationExplainability and InterpretabilityGraph Neural NetworkTransformerBiomedical DataElectronic Health Records
🎯 What it does: Proposed a time-heterogeneous graph-based representation method for electronic health records (EHR) and designed the Temporal Graph Transformer (TRANS) model to simultaneously capture structural information of clinical events and temporal dynamics between follow-ups for diagnostic prediction.
Preferred Reasoning in ABA by Cycle-Breaking
Kiet Nguyen Anh (Leipzig University), Markus Ulbricht (Leipzig University)
GraphBenchmark
🎯 What it does: This paper proposes a fixed-parameter tractable algorithm based on backdoors and dependency graphs for suspicious preference reasoning in assumption-based argumentation (ABA).
Primal Grammars Driven Automated Induction
Adel Bouhoula (Arabian Gulf University), Miki Hermann (cole Polytechnique)
Computational EfficiencyBenchmark
🎯 What it does: This paper proposes an automated inductive proof method that can detect and capture divergence during the proof process, generating new lemmas using primal grammars to complete the proof;
Privacy-Preserving UCB Decision Process Verification via zk-SNARKs
Xikun Jiang (Shanghai Jiao Tong University), Yuan Luo (Shanghai Jiao Tong University)
Safty and Privacy
🎯 What it does: Propose a zkUCB that combines zk-SNARKs with the UCB algorithm to achieve privacy-preserving and verifiable decision-making in multi-armed bandit (MAB) problems.
PrivSGP-VR: Differentially Private Variance-Reduced Stochastic Gradient Push with Tight Utility Bounds
Zehan Zhu (Zhejiang University), Jinming Xu (Zhejiang University)
Federated LearningSafty and PrivacyConvolutional Neural NetworkImage
🎯 What it does: Proposed a differential privacy distributed learning algorithm, PrivSGP-VR, implemented in time-varying directed networks, combining variance-reduced stochastic gradient push.
Probabilistic Contrastive Learning for Domain Adaptation
Junjie Li (Beijing University of Posts and Telecommunications), Man Zhang (Beijing University of Posts and Telecommunications)
Domain AdaptationContrastive LearningImage
🎯 What it does: Proposed a probability-based contrastive learning method to address the issue of feature and class weight deviation in domain adaptation.
Probabilistically Robust Watermarking of Neural Networks
Mikhail Pautov (Artificial Intelligence Research Institute), Ivan Oseledets (Artificial Intelligence Research Institute)
Safty and PrivacyKnowledge DistillationAdversarial AttackImage
🎯 What it does: Proposed a probability-robust watermarking framework based on trigger sets for ownership verification of deep learning models.
Prompt Learning for Generalized Vehicle Routing
Fei Liu (City University of Hong Kong), Mingxuan Yuan (Huawei Noah's Ark Lab)
OptimizationMeta LearningTransformerReinforcement LearningPrompt EngineeringGraph
🎯 What it does: Leveraging prompt learning to achieve rapid zero-shot adaptation of pre-trained NCO models for addressing cross-distribution vehicle routing problems.
Prompt Learning with Extended Kalman Filter for Pre-trained Language Models
Quan Li (University of Science and Technology of China), S. Kevin Zhou (University of Science and Technology of China)
Computational EfficiencyRepresentation LearningTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: Propose a neural network framework called CEKFNN based on the conditional extended Kalman filter, designed to dynamically generate prompts and achieve efficient Prompt learning.
Prompt-enhanced Network for Hateful Meme Classification
Junxi Liu (South China Normal University), Fenghuan Li (Guangdong University of Technology)
ClassificationRecurrent Neural NetworkTransformerLarge Language ModelPrompt EngineeringContrastive LearningMultimodality
🎯 What it does: Proposes a Prompt-Enhanced Network (Pen) framework that extends prompt learning into the feature space, combining multi-view perception and prompt-aware contrastive learning for hate meme classification.
Proportion-based Sensitivity Analysis of Uncontrolled Confounding Bias in Causal Inference
Haruka Yoshida (Yokohama National University), Manabu Kuroki (Yokohama National University)
TabularBiomedical Data
🎯 What it does: Proposes a proportion-based sensitivity analysis (PSA) that evaluates the reliability of causal effects by defining the ratio of uncontrolled confounding bias squared to mean squared error (MSE), and provides theoretical derivations and estimable bounds within the potential outcomes framework and linear structural equation models (SEM).
Protecting Object Detection Models from Model Extraction Attack via Feature Space Coverage
Zeyu Li (Zhejiang University), Shouling Ji (Zhejiang University)
Object DetectionSafty and PrivacyAdversarial AttackContrastive LearningImage
🎯 What it does: Propose a detection framework called OSD based on feature space coverage to identify model extraction attacks targeting object detection models.
Protecting Split Learning by Potential Energy Loss
Fei Zheng (Zhejiang University), Jianwei Yin (Zhejiang University)
ClassificationFederated LearningSafty and PrivacyImageText
🎯 What it does: Propose Potential Energy Loss (PELoss), which introduces repulsive forces on the forward embeddings in split learning, making same-class embeddings cluster near the decision boundary to suppress the model's attack capability.
ProtoPFormer: Concentrating on Prototypical Parts in Vision Transformers for Interpretable Image Recognition
Mengqi Xue (Hangzhou City University), Canghong Jin (Hangzhou City University)
Explainability and InterpretabilityTransformerImage
🎯 What it does: Propose ProtoPFormer, a prototype network for Vision Transformers, addressing the issue of prototypes in ViT being disturbed by backgrounds, thereby enhancing interpretability and accuracy.
Provable Acceleration of Nesterov’s Accelerated Gradient Method over Heavy Ball Method in Training Over-Parameterized Neural Networks
Xin Liu (Army Engineering University Of Pla), Zhisong Pan (Army Engineering University Of Pla)
OptimizationImageBenchmarkOrdinary Differential Equation
🎯 What it does: Studied the convergence differences between Heavy Ball and Nesterov Accelerated Gradient (NAG) momentum methods in overparameterized two-layer ReLU neural networks, and provided rigorous theoretical proofs.
Proximal Curriculum with Task Correlations for Deep Reinforcement Learning
Georgios Tzannetos (Max Planck Institute for Software Systems), Adish Singla (Max Planck Institute for Software Systems)
Reinforcement LearningBenchmark
🎯 What it does: Propose a curriculum learning strategy named PROCURL‑TARGET, addressing the learning objective of target distribution in context-aware multi-task reinforcement learning, balancing task difficulty and progress towards the target task;
PTDE: Personalized Training with Distilled Execution for Multi-Agent Reinforcement Learning
Yiqun Chen (Renmin University of China), Hongxing Chang (Institute of Automation Chinese Academy of Sciences)
Knowledge DistillationReinforcement Learning
🎯 What it does: Proposes a two-phase training paradigm called Personalized Training with Distilled Execution (PTDE), which first enhances agents' decision-making using Global Information Personalization (GIP) with global information, then converts personalized global information into student networks relying only on local observations through knowledge distillation, enabling decentralized execution.
Public Event Scheduling with Busy Agents
Bo Li (Hong Kong Polytechnic University), Ruilong Zhang (University at Buffalo)
Optimization
🎯 What it does: This paper proposes and studies the Public Event Scheduling with Busy Agents (PESBA) problem, aiming to maximize the total time all agents can participate in activities under their work schedules.
Purpose Enhanced Reasoning through Iterative Prompting: Uncover Latent Robustness of ChatGPT on Code Comprehension
Yi Wang (North Carolina State University), Xu Liu (North Carolina State University)
AI Code AssistantTransformerLarge Language ModelPrompt EngineeringText
🎯 What it does: Propose and implement a multi-module prompting framework named Perthept, which enhances the robustness and quality of ChatGPT in code comment generation through two iterative steps: Chain-of-Structure and Reasoning-Enhancement.
Putting Back the Stops: Integrating Syntax with Neural Topic Models
Mayank Nagda (RPTU Kaiserslautern-Landau), Sophie Fellenz (RPTU Kaiserslautern-Landau)
Representation LearningAuto EncoderText
🎯 What it does: Designed a neural topic model named SyConNTM that can simultaneously learn syntactic and semantic topics without preprocessing, and automatically identify stop words through a context network.
QFormer: An Efficient Quaternion Transformer for Image Denoising
Bo Jiang (Northwest A & F University), Bob Zhang (University of Macau)
RestorationTransformerImage
🎯 What it does: Proposed a Quaternion-based Transformer network called QFormer, which uses the Quaternion Transformer Block (QTB) to denoise color image noise;