AAAI 2024 Papers — Page 10
AAAI Conference on Artificial Intelligence · 2331 papers
Generative-Based Fusion Mechanism for Multi-Modal Tracking
Zhangyong Tang (Jiangnan University), Josef Kittler (University of Surrey)
Object TrackingConvolutional Neural NetworkTransformerDiffusion modelGenerative Adversarial NetworkImageMultimodality
🎯 What it does: A multi-modal tracking fusion mechanism GMMT based on generative models is proposed, which utilizes a generator under noisy conditions to fuse RGB-T or RGB-D features.
Generator Assisted Mixture of Experts for Feature Acquisition in Batch
Vedang Asgaonkar (Indian Institute of Technology Bombay), Abir De (Indian Institute of Technology Bombay)
OptimizationMixture of ExpertsAuto EncoderGenerative Adversarial NetworkImageBiomedical Data
🎯 What it does: The study achieves optimal feature subset selection in batch feature acquisition through generator assistance, mixed experts, and random hyperplane partitioning.
Geometric-Facilitated Denoising Diffusion Model for 3D Molecule Generation
Can Xu (Zhejiang Gongshang University), Hongyang Chen (Zhejiang Lab)
GenerationData SynthesisDrug DiscoveryTransformerDiffusion modelPoint CloudGraph
🎯 What it does: A 3D molecular generation model based on denoising diffusion, GFMDiff, is proposed, which can simultaneously learn atomic coordinates, types, and bonding structures.
Geometry-Guided Domain Generalization for Monocular 3D Object Detection
Fan Yang (Tsinghua University), Guiguang Ding (Tsinghua University)
Object DetectionDomain AdaptationAutonomous DrivingImage
🎯 What it does: This work proposes the MonoGDG framework, aimed at addressing multi-source domain differences such as camera parameters, pose, and image appearance in monocular 3D object detection through a geometry-guided domain generalization method.
Get a Head Start: On-Demand Pedagogical Policy Selection in Intelligent Tutoring
Ge Gao (North Carolina State University), Min Chi
Reinforcement LearningAuto EncoderTabular
🎯 What it does: An online teaching strategy selection framework named EDUPLANNER is proposed, which divides subgroups based on students' initial logs, performs feature taxation, and uses VAE data augmentation, and selects the optimal teaching strategy through OPE.
Get an A in Math: Progressive Rectification Prompting
Zhenyu Wu (Xi'an Jiaotong University), Chao Shen (Xi'an Jiaotong University)
TransformerLarge Language ModelPrompt EngineeringTextChain-of-Thought
🎯 What it does: A zero-shot reasoning method called Progressive Rectification Prompting (PRP) is proposed, which continuously improves the answers of large language models on mathematical word problems through an iterative verification-correction process.
Ghost Noise for Regularizing Deep Neural Networks
Atli Kosson (École Polytechnique Fédérale de Lausanne), Martin Jaggi (École Polytechnique Fédérale de Lausanne)
OptimizationConvolutional Neural NetworkTransformerImage
🎯 What it does: This study investigates the impact of noise (Ghost Noise) caused by smaller batch sizes in Batch Normalization (BN) on model generalization and proposes Ghost Noise Injection (GNI) as a new regularization technique.
GigaHumanDet: Exploring Full-Body Detection on Gigapixel-Level Images
Chenglong Liu (University of Chinese Academy of Sciences), Lu Fang (Tsinghua University)
Object DetectionConvolutional Neural NetworkImage
🎯 What it does: This paper proposes GigaHumanDet, a detection framework for full-body human detection in ultra-high resolution (Gigapixel) images, based on corner point detection, instance-guided learning, shape-aware matching, and multi-precision aspect ratio embedding.
GIN-SD: Source Detection in Graphs with Incomplete Nodes via Positional Encoding and Attentive Fusion
Le Cheng (Northwestern Polytechnical University), Zhen Wang (Northwestern Polytechnical University)
Graph Neural NetworkGraph
🎯 What it does: A rumor source detection framework GIN-SD is proposed for scenarios with incomplete nodes (missing information) in graphs;
GINN-LP: A Growing Interpretable Neural Network for Discovering Multivariate Laurent Polynomial Equations
Nisal Ranasinghe (University of Melbourne), Saman Halgamuge (University of Melbourne)
Explainability and InterpretabilityComputational EfficiencyTabularPhysics Related
🎯 What it does: An interpretable neural network GINN-LP is proposed, capable of automatically discovering multivariable Laurent polynomial (LP) equations from data, supporting arbitrary degrees and terms, and can automatically determine the required number of terms through a network growth strategy.
GLDL: Graph Label Distribution Learning
Yufei Jin (Florida Atlantic University), Xingquan Zhu (Old Dominion University)
Graph Neural NetworkGraph
🎯 What it does: A Graph Label Distribution Learning (GLDL) framework is proposed to predict the label distribution of each node in a network structure.
GLOP: Learning Global Partition and Local Construction for Solving Large-Scale Routing Problems in Real-Time
Haoran Ye (Soochow University), Fanzhang Li (Tsinghua University)
OptimizationGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: Proposes the GLOP framework, which combines non-autoregressive (NAR) partitioning with autoregressive (AR) refinement to achieve hierarchical solving for large-scale routing problems.
GMMFormer: Gaussian-Mixture-Model Based Transformer for Efficient Partially Relevant Video Retrieval
Yuting Wang (Tsinghua University), Shu-Tao Xia (Tsinghua University)
RetrievalComputational EfficiencyTransformerVideoText
🎯 What it does: This paper studies the Partial Relevant Video Retrieval (PRVR) task and proposes a Gaussian Mixture Model-based Transformer (GMMFormer) to achieve implicit clip modeling, along with a designed query diversity loss to enhance the semantic structure of text representations.
GMP-AR: Granularity Message Passing and Adaptive Reconciliation for Temporal Hierarchy Forecasting
Fan Zhou (Ant Group), Yunhua Hu (TTIC)
Recurrent Neural NetworkTime SeriesFinance Related
🎯 What it does: A temporal hierarchical prediction framework based on granular information message passing and adaptive harmonization (GMP-AR) is proposed, introducing a task optimization module to address practical constraints.
GO-DICE: Goal-Conditioned Option-Aware Offline Imitation Learning via Stationary Distribution Correction Estimation
Abhinav Jain (Rice University), Vaibhav Unhelkar (Rice University)
Robotic IntelligenceReinforcement Learning from Human FeedbackReinforcement LearningSequential
🎯 What it does: This paper proposes an offline imitation learning method for goal-conditioned long-horizon tasks called GO-DICE, which learns hierarchical sub-tasks and action policies from demonstration data, enabling the completion of complex tasks such as multi-object pick-and-place without the need for environment interaction.
Goal Alignment: Re-analyzing Value Alignment Problems Using Human-Aware AI
Malek Mechergui (Colorado State University), Sarath Sreedharan (Colorado State University)
Reinforcement Learning
🎯 What it does: A human cognitive model-based goal alignment framework (HAGL) is proposed, along with an interactive query algorithm designed to infer hidden true goals.
GOALNET: Interleaving Neural Goal Predicate Inference with Classical Planning for Generalization in Robot Instruction Following
Jigyasa Gupta (Indian Institute of Technology Delhi), Mausam
Robotic IntelligenceRecurrent Neural NetworkReinforcement LearningText
🎯 What it does: This paper presents GOALNET, a hybrid framework that alternates between predicting sub-goals using neural networks and classical planners to generate multi-stage robot execution plans from natural language instructions and world states.
GOODAT: Towards Test-Time Graph Out-of-Distribution Detection
Luzhi Wang (Tianjin University), Tat-Seng Chua (National University of Singapore)
Anomaly DetectionGraph Neural NetworkGraphBiomedical Data
🎯 What it does: A testing-time graph structure OOV detection method called GOODAT is proposed, which utilizes a graph masker to learn information bottleneck subgraphs on the test set for unsupervised OOV recognition.
Grab What You Need: Rethinking Complex Table Structure Recognition with Flexible Components Deliberation
Hao Liu (Tencent YouTu Lab), Xing Sun (Tencent YouTu Lab)
RecognitionTransformerTabular
🎯 What it does: For the task of recognizing complex table structures, the GrabTab method is proposed, which achieves end-to-end table structure inference through 'deliberation'-style fusion of various components such as table separators, cell elements, and row-column relationships.
Gradient-Guided Modality Decoupling for Missing-Modality Robustness
Hao Wang (Southern University of Science and Technology), Jianguo Zhang (Southern University of Science and Technology)
ClassificationSegmentationMultimodalityBiomedical DataMagnetic Resonance Imaging
🎯 What it does: Proposes two mechanisms: Gradient-Guided Modality Decoupling (GMD) and Dynamic Sharing (DS) to enhance the robustness of multimodal learning in the presence of missing modalities.
GradTree: Learning Axis-Aligned Decision Trees with Gradient Descent
Sascha Marton (University of Mannheim), Heiner Stuckenschmidt (University of Mannheim)
ClassificationOptimizationTabularElectrocardiogramBenchmark
🎯 What it does: Proposes GradTree, which jointly optimizes all parameters of the hard-axis aligned decision tree through gradient descent;
Gradual Residuals Alignment: A Dual-Stream Framework for GAN Inversion and Image Attribute Editing
Hao Li (University of Science and Technology of China), Zhendong Mao (University of Science and Technology of China)
Image TranslationGenerationTransformerGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes the GradStyle dual-stream framework, which achieves high-detail reconstruction and high editability of GAN inversion and image attribute editing through staged progressive residual injection.
GraFITi: Graphs for Forecasting Irregularly Sampled Time Series
Vijaya Krishna Yalavarthi (University of Hildesheim), Lars Schmidt-Thieme (TU Berlin)
Graph Neural NetworkTime SeriesBiomedical Data
🎯 What it does: A GraFITi model based on graph neural networks is proposed, which transforms irregularly sampled multivariate time series with missing values into a sparse bipartite graph, and predicts future values by estimating the weights of edges in the graph.
Gramformer: Learning Crowd Counting via Graph-Modulated Transformer
Hui Lin (Xi'an Jiaotong University), Deyu Meng (Harbin Institute of Technology)
Graph Neural NetworkTransformerImage
🎯 What it does: This paper presents Gramformer, a model that enhances crowd counting accuracy by adjusting the attention mechanism and node features of the Transformer through a graph structure.
Graph Context Transformation Learning for Progressive Correspondence Pruning
Junwen Guo (Tongji University), Jun Yu (Fuzhou University)
Pose EstimationGraph Neural NetworkTransformerImage
🎯 What it does: A GCT-Net based on graph context transformation is proposed for stepwise correspondence point removal and pose estimation in two-view matching.
Graph Contrastive Invariant Learning from the Causal Perspective
Yanhu Mo (Beijing University of Posts and Telecommunications), Chuan Shi (Beijing University of Posts and Telecommunications)
Representation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: From a causal perspective, the GCIL method is proposed, which uses spectral graph enhancement to simulate non-causal interventions and incorporates dimension-level invariance and independence objectives to achieve causal invariant representation learning in graph contrastive learning.
Graph Disentangled Contrastive Learning with Personalized Transfer for Cross-Domain Recommendation
Jing Liu (Tianjin University), Yuting Su (Tianjin University)
Recommendation SystemMeta LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: A cross-domain recommendation model GDCCDR based on graph decoupling and contrastive learning is proposed, utilizing a meta-network to achieve personalized transfer of transferable features within domains.
Graph Invariant Learning with Subgraph Co-mixup for Out-of-Distribution Generalization
Tianrui Jia (Beijing University of Posts and Telecommunications), Chuan Shi (China Mobile Information Technology Co. Ltd.)
ClassificationDomain AdaptationGraph Neural NetworkMixture of ExpertsGraph
🎯 What it does: This paper proposes a graph invariant learning framework based on subgraph co-mixup, combining environment mixup and invariant mixup to achieve OOD generalization for graph structures.
Graph Learning in 4D: A Quaternion-Valued Laplacian to Enhance Spectral GCNs
Stefano Fiorini (Italian Institute of Technology), Enza Messina (University of Milano-Bicocca)
Graph Neural NetworkGraph
🎯 What it does: This paper proposes a quaternion Laplacian matrix and the corresponding spectral graph convolutional network QuaterGCN, designed to handle directed graphs with arbitrary weights, arbitrary signs, and bidirectional edges (digons).
Graph Neural Networks with Soft Association between Topology and Attribute
Yachao Yang (Beijing University of Technology), Baocai Yin
ClassificationRepresentation LearningGraph Neural NetworkGraph
🎯 What it does: This paper proposes a new semi-supervised node classification graph neural network GNN-SATA, which utilizes a soft association mechanism to separate attribute and topology representations, and dynamically adjusts adjacency relationships through graph pruning (GPM) and graph augmentation (GAM) modules.
Graph Neural Prompting with Large Language Models
Yijun Tian (University of Notre Dame), Panpan Xu (Amazon)
Graph Neural NetworkLarge Language ModelPrompt EngineeringTextGraphBiomedical Data
🎯 What it does: This paper proposes a prompting method based on Graph Neural Networks—Graph Neural Prompting (GNP)—to inject structured knowledge from Knowledge Graphs (KG) into pre-trained large language models (LLM) through soft prompts, thereby enhancing their performance in tasks such as question answering.
Graph of Thoughts: Solving Elaborate Problems with Large Language Models
Maciej Besta (ETH Zurich), Torsten Hoefler (Cledar)
Large Language ModelPrompt EngineeringTextChain-of-Thought
🎯 What it does: The Graph of Thoughts (GoT) framework is proposed, modeling the reasoning process of LLMs as an arbitrary graph structure, allowing thoughts (LLM outputs) to combine, aggregate, and iterate based on graph relationships, thereby enhancing multi-step reasoning capabilities.
Graph Reasoning Transformers for Knowledge-Aware Question Answering
Ruilin Zhao (Huazhong University of Science and Technology), Guandong Xu (University of Technology Sydney)
Graph Neural NetworkTransformerTextMultimodality
🎯 What it does: This paper proposes the Graph Reasoning Transformers (GRT) model, which enhances knowledge-augmented question answering performance through a triplet-level graph encoder, cross-modal representation alignment pre-training, and attention bias in cross-modal fusion.
Graph-Aware Contrasting for Multivariate Time-Series Classification
Yucheng Wang (Institute for Infocomm Research, A*STAR), Zhenghua Chen (Nanyang Technological University)
ClassificationGraph Neural NetworkTransformerContrastive LearningTime Series
🎯 What it does: A graph contrastive learning framework for multivariate time series (TS-GAC) is proposed, achieving spatial consistency and temporal consistency through graph structures.
Graph-Based Prediction and Planning Policy Network (GP3Net) for Scalable Self-Driving in Dynamic Environments Using Deep Reinforcement Learning
Jayabrata Chowdhury (Indian Institute of Science), PB Sujit (Indian Institute of Science Education and Research)
Autonomous DrivingGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: Proposes the GP3Net framework, which combines graphical prediction with deep reinforcement learning to achieve adaptive safe driving.
Greedy-Based Online Fair Allocation with Adversarial Input: Enabling Best-of-Many-Worlds Guarantees
Zongjun Yang (Peking University), Christian Kroer (Columbia University)
OptimizationAdversarial Attack
🎯 What it does: This study investigates the problem of online fair allocation. Under aggressive input conditions (while satisfying non-extreme values and the total utility of a single agent approaching infinity), it proves that the Greedy and PACE algorithms can achieve a balance between fairness and efficiency over an infinite period, and provides asymptotic controllable guarantees for the Best-of-Many-Worlds scenario.
GridFormer: Point-Grid Transformer for Surface Reconstruction
Shengtao Li (Tsinghua University), Ming Gu (Tsinghua University)
RestorationSegmentationGenerationConvolutional Neural NetworkTransformerPoint CloudMesh
🎯 What it does: Proposes GridFormer, which utilizes a point-grid transformer to bridge point clouds and space through attention between point and grid features, generating a continuous occupancy field;
GroundVLP: Harnessing Zero-Shot Visual Grounding from Vision-Language Pre-training and Open-Vocabulary Object Detection
Haozhan Shen (Zhejiang University), Jianwei Yin (Zhejiang University)
RecognitionObject DetectionTransformerVision Language ModelContrastive LearningImageTextMultimodality
🎯 What it does: This paper proposes GroundVLP, a zero-shot visual localization method that utilizes existing visual language pre-trained models (VLP) and open vocabulary object detectors (OVD) to accomplish referential expression recognition and phrase localization tasks without the need for additional annotated data.
GSDD: Generative Space Dataset Distillation for Image Super-resolution
Haiyu Zhang (Northwestern Polytechnical University), Yanning Zhang (Northwestern Polytechnical University)
Data SynthesisSuper ResolutionKnowledge DistillationGenerative Adversarial NetworkImage
🎯 What it does: This paper studies a generative spatial data distillation method based on GAN inversion, called GSDD, which compresses the training set while retaining key information for low-level image super-resolution tasks.
GSENet:Global Semantic Enhancement Network for Lane Detection
Junhao Su (Southeast University), Zhihuai Xie (Fudan University)
Autonomous DrivingConvolutional Neural NetworkTransformerImage
🎯 What it does: This paper proposes GSENet, which enhances lane detection accuracy in complex scenarios using a global semantic enhancement network.
GSN: Generalisable Segmentation in Neural Radiance Field
Vinayak Gupta (Indian Institute of Technology), P. J. Narayanan
SegmentationKnowledge DistillationTransformerNeural Radiance FieldContrastive LearningImage
🎯 What it does: Achieve multi-view semantic segmentation of unknown scenes using a general NeRF model without the need to retrain for each scene.
GSO-Net: Grid Surface Optimization via Learning Geometric Constraints
Chaoyun Wang (Xi'an Jiaotong University), Caigui Jiang (Xi'an Jiaotong University)
OptimizationConvolutional Neural NetworkMesh
🎯 What it does: Proposes encoding mesh surfaces as geometric images and completing developability, unfolding, and denoising optimization through a self-supervised learning network called GSO-Net.
Guiding a Harsh-Environments Robust Detector via RAW Data Characteristic Mining
Hongyang Chen (Xi'an Jiaotong University), Kaisheng Ma (Tsinghua University)
Object DetectionImage
🎯 What it does: This study investigates the feasibility of using RAW images for object detection in harsh environments and proposes a complete RAW detection pipeline (PRD), a raw noise degradation benchmark (RCB), and a nonlinear regularization method (FR).
GxVAEs: Two Joint VAEs Generate Hit Molecules from Gene Expression Profiles
Chen Li (Nagoya University), Yoshihiro Yamanishi (Nagoya University)
GenerationData SynthesisDrug DiscoveryRecurrent Neural NetworkAuto EncoderBiomedical Data
🎯 What it does: This study proposes a dual VAE model (ProfileVAE + MolVAE) to generate 'hit' molecules with targeted activity and drug similarity from gene expression profiles, utilizing latent features of gene expression to guide molecular generation.
H-ensemble: An Information Theoretic Approach to Reliable Few-Shot Multi-Source-Free Transfer
Yanru Wu (Tsinghua University), Yang Li (Tsinghua University)
ClassificationDomain AdaptationOptimizationImage
🎯 What it does: The H-ensemble framework is proposed, utilizing the information theory H-score for dynamic linear ensemble optimization in multi-source few-shot transfer learning.
H2GFormer: Horizontal-to-Global Voxel Transformer for 3D Semantic Scene Completion
Yu Wang (Beihang University), Chao Tong (Beihang University)
SegmentationAutonomous DrivingTransformerImagePoint Cloud
🎯 What it does: Proposes H2GFormer, which utilizes RGB images to achieve 3D semantic scene completion through a Transformer, focusing on handling horizontal differences and the importance of object edges;
HACDR-Net: Heterogeneous-Aware Convolutional Network for Diabetic Retinopathy Multi-Lesion Segmentation
QiHao Xu (Shenzhen University), Yong Xu (Harbin Institute of Technology)
SegmentationConvolutional Neural NetworkImage
🎯 What it does: This paper proposes a network based on heterogeneous perception convolution (HACDR-Net) and noise adjustment loss (NALoss) to improve the accuracy of multi-lesion segmentation in diabetic retinopathy.
HAGO-Net: Hierarchical Geometric Message Passing for Molecular Representation Learning
Hongbin Pei (Xi'an Jiaotong University), Xiaohong Guan (Xi'an Jiaotong University)
Representation LearningDrug DiscoveryGraph Neural NetworkGraph
🎯 What it does: Proposed the HAGO-Graph hierarchical geometric graph and HAGO-Net model for complete and hierarchical representation of molecular geometry and learning molecular representations;
Hand-Centric Motion Refinement for 3D Hand-Object Interaction via Hierarchical Spatial-Temporal Modeling
Yuze Hao (Zhejiang University), Hehe Fan (Zhejiang University)
Pose EstimationRobotic IntelligenceTransformerAuto EncoderPoint Cloud
🎯 What it does: Improved the motion refinement method for 3D hand-object interaction, refining rough gesture sequences affected by noise and jitter to make interactions more natural and coherent.
Hard Regularization to Prevent Deep Online Clustering Collapse without Data Augmentation
Louis Mahon (University of Edinburgh), Thomas Lukasiewicz (Vienna University of Technology)
OptimizationRepresentation LearningConvolutional Neural NetworkImage
🎯 What it does: This paper proposes an online deep clustering method that does not require data augmentation. It prevents clustering collapse by applying Bayesian regularization to hard assignments within batches during training, achieving high-quality representation learning.
Hardness of Random Reordered Encodings of Parity for Resolution and CDCL
Leroy Chew (TU Wien), Stefan Szeider (TU Wien)
Graph
🎯 What it does: This paper proves that viewing randomly permuted Parity encoding as a Tseitin formula leads to exponential length/time requirements in Resolution and CDCL solving due to the linear tree width of its graph. It further introduces a random Parity Addition formula, demonstrating the same exponential difficulty; at the same time, it provides O(n log n) DRAT short proofs for these formulas.
HARDVS: Revisiting Human Activity Recognition with Dynamic Vision Sensors
Xiao Wang (Anhui University), Yonghong Tian (Peking University)
RecognitionTransformerVideo
🎯 What it does: A large-scale real event camera dataset HARDVS (300 classes, 100K event sequences) is proposed, and a Transformer-based ESTF model is designed for human action recognition in event streams.
Harnessing Edge Information for Improved Robustness in Vision Transformers
Yanxi Li (University of Sydney), Chang Xu (University of Sydney)
Adversarial AttackTransformerImage
🎯 What it does: A lightweight EdgeNet module is added to the existing high-precision visual Transformer, utilizing Canny edge detection to extract shape and foreground information from images. These edge features are injected into the model through a zero convolution 'interlayer', significantly enhancing resistance to adversarial attacks such as FGSM and PGD, as well as robustness benchmarks like ImageNet-A/R/C, without sacrificing the accuracy of natural images.
Harnessing Holistic Discourse Features and Triadic Interaction for Sentiment Quadruple Extraction in Dialogues
Bobo Li (Wuhan University), Donghong Ji (Wuhan University)
Graph Neural NetworkLarge Language ModelTextBenchmark
🎯 What it does: A model named H2DT is proposed to extract sentiment quadruples (target, aspect, opinion, sentiment polarity) from multi-turn dialogues.
Harnessing Manycore Processors with Distributed Memory for Accelerated Training of Sparse and Recurrent Models
Jan Finkbeiner (Research Center Juelich), Emre Neftci (Graphcore)
Recurrent Neural NetworkSpiking Neural NetworkSequential
🎯 What it does: Implemented a BPTT-based sparse spiking neural network (SNN) training method on Graphcore IPU.
Harnessing the Power of Beta Scoring in Deep Active Learning for Multi-Label Text Classification
Wei Tan (Monash University), Wray Buntine (VinUniversity)
ClassificationConvolutional Neural NetworkRecurrent Neural NetworkSupervised Fine-TuningText
🎯 What it does: A deep active learning framework based on Beta scores, called BESRA, is proposed to address the label imbalance problem in multi-label text classification.
Harnessing the Power of SVD: An SVA Module for Enhanced Signal Classification
Lei Zhai (Xidian University), Quanwei Gao (Xidian University)
ClassificationRecognitionConvolutional Neural NetworkTransformerTime SeriesSequentialBiomedical Data
🎯 What it does: A singular value decomposition-based attention module (SVA) is proposed to enhance communication signal features and suppress noise, integrated into common network backbones.
Hawkes-Enhanced Spatial-Temporal Hypergraph Contrastive Learning Based on Criminal Correlations
Ke Liang (National University of Defense Technology), Xinwang Liu (Nanjing University of Aeronautics and Astronautics)
Contrastive LearningGraphTime Series
🎯 What it does: This paper proposes a spatial-temporal hypergraph framework (HCL) that combines the Hawkes process with contrastive learning for crime prediction, focusing on three types of crime correlations: the co-occurrence of different crime types (type space correlation), the diminishing danger near crime hotspots (neighborhood space correlation), and the Hawkes temporal correlation where events occurring close in time are more related.
HDformer: A Higher-Dimensional Transformer for Detecting Diabetes Utilizing Long-Range Vascular Signals
Ella Lan (Harker School)
ClassificationAnomaly DetectionTransformerMixture of ExpertsTime SeriesBiomedical DataElectronic Health Records
🎯 What it does: A Transformer-based HDformer model is proposed, utilizing optical pulse waveforms (PPG) lasting over 10 minutes for early detection of diabetes.
HDMixer: Hierarchical Dependency with Extendable Patch for Multivariate Time Series Forecasting
Qihe Huang (University of Science and Technology of China), Yang Wang (University of Science and Technology of China)
Time Series
🎯 What it does: A pure MLP structure called HDMixer is proposed, which implements multivariate time series forecasting using scalable chunking (LEP) and hierarchical dependency detectors (HDE).
HEAP: Unsupervised Object Discovery and Localization with Contrastive Grouping
Xin Zhang (National University of Singapore), Robby T. Tan (National University of Singapore)
Object DetectionSegmentationRetrievalTransformerContrastive LearningImage
🎯 What it does: A hierarchical unsupervised object discovery framework HEAP based on contrastive clustering is proposed, which can achieve multi-class object segmentation within a single image.
Heterogeneous Graph Reasoning for Fact Checking over Texts and Tables
Haisong Gong (Chinese Academy of Sciences), Liang Wang (Chinese Academy of Sciences)
Graph Neural NetworkLarge Language ModelTextTabular
🎯 What it does: This paper proposes a fact-checking model based on a word-level heterogeneous graph, HeterFC, which performs authenticity reasoning on text and table evidence.
Heterogeneous Test-Time Training for Multi-Modal Person Re-identification
Zi Wang (Anhui University), Ran He (Chinese Academy of Sciences Institute of Automation)
RecognitionDomain AdaptationTransformerContrastive LearningImageMultimodality
🎯 What it does: A Heterogeneous Testing Time Training (HTT) framework is proposed, which utilizes the Cross-Identity Cross-Modal Distance Loss (CIM) and Multi-Modal Testing Time Training (MTT) self-supervised strategy to significantly enhance the generalization performance of multi-modal person re-identification in unknown domains.
HGE: Embedding Temporal Knowledge Graphs in a Product Space of Heterogeneous Geometric Subspaces
Jiaxin Pan (University of Stuttgart), Steffen Staab (University of Stuttgart)
Graph Neural NetworkGraphTime Series
🎯 What it does: A Temporal Knowledge Graph Embedding model HGE based on heterogeneous geometric subspaces is proposed, utilizing the multiplication of complex numbers, split-complex numbers, and dual numbers, along with two attention mechanisms for integration.
HGPrompt: Bridging Homogeneous and Heterogeneous Graphs for Few-Shot Prompt Learning
Xingtong Yu (University of Science and Technology of China), Xinming Zhang (National University of Singapore)
Graph Neural NetworkPrompt EngineeringGraph
🎯 What it does: Proposes the HGPROMPT framework, unifying pre-training and downstream tasks for homogeneous and heterogeneous graphs, achieving few-shot graph learning through dual templates and dual prompts.
Hidden Follower Detection: How Is the Gaze-Spacing Pattern Embodied in Frequency Domain?
Shu Li (Xidian University), Liang Liao (Nanyang Technological University)
Object TrackingAnomaly DetectionTransformerAudio
🎯 What it does: This study investigates the expression efficiency of hidden tracking behavior in the frequency domain and proposes a new hidden tracking detection framework based on this.
Hierarchical Aligned Multimodal Learning for NER on Tweet Posts
Peipei Liu (Institute of Information Engineering, Chinese Academy of Sciences), Limin Sun (Institute of Information Engineering, Chinese Academy of Sciences)
RecognitionGraph Neural NetworkTransformerVision Language ModelImageTextMultimodality
🎯 What it does: A hierarchical alignment multimodal learning framework called HamLearning is proposed for multimodal named entity recognition (MNER) in tweets;
Hierarchical and Incremental Structural Entropy Minimization for Unsupervised Social Event Detection
Yuwei Cao (University of Illinois Chicago), Philip S. Yu (University of Illinois Chicago)
Graph Neural NetworkTextGraph
🎯 What it does: This paper proposes and implements HISEvent, an unsupervised social event detection framework that first minimizes the incremental one-dimensional structural entropy (1D SE) to supplement semantically similar edges in the message graph, and then utilizes hierarchical two-dimensional structural entropy (2D SE) to minimize the segmentation of the graph, thereby obtaining social event clusters.
Hierarchical Multi-Marginal Optimal Transport for Network Alignment
Zhichen Zeng (University of Illinois Urbana-Champaign), Hanghang Tong (University of Illinois Urbana-Champaign)
OptimizationGraph Neural NetworkGraph
🎯 What it does: A hierarchical optimal transport framework (HOT) is proposed for multi-network alignment.
Hierarchical Planning and Learning for Robots in Stochastic Settings Using Zero-Shot Option Invention
Naman Shah (Arizona State University), Siddharth Srivastava (Arizona State University)
Robotic IntelligenceReinforcement Learning
🎯 What it does: A method is proposed for zero-shot invention of hierarchical options in an unknown dynamic stochastic environment and solving long-delay sparse reward robot planning problems using hierarchical planning.
Hierarchical Topology Isomorphism Expertise Embedded Graph Contrastive Learning
Jiangmeng Li (Institute of Software Chinese Academy of Sciences), Fuchun Sun (Tsinghua University)
Knowledge DistillationRepresentation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: In unsupervised graph representation learning, a method called Hierarchical Topological Isomorphism Knowledge Embedding for Graph Contrastive Learning (HTML) is proposed. This method allows the GNN encoder to learn isomorphic information at the graph level and subgraph level through knowledge distillation, adding isomorphic loss in addition to the original contrastive loss.
Hierarchize Pareto Dominance in Multi-Objective Stochastic Linear Bandits
Ji Cheng (City University of Hong Kong), Qingfu Zhang (City University of Hong Kong)
Recommendation SystemOptimizationReinforcement LearningTabular
🎯 What it does: A multi-objective stochastic linear bandit (MOSLB) algorithm is proposed for mixed Pareto-lexicographic order, along with theoretical analysis and experimental validation of two mixed orders (MPL-PC and MPL-PL).
High-Dimensional Analysis for Generalized Nonlinear Regression: From Asymptotics to Algorithm
Jian Li (Institute of Information Engineering), Weiping Wang (Institute of Information Engineering)
Image
🎯 What it does: This paper proposes a general framework for high-dimensional nonlinear regression analysis, covering various feature mappings such as random features, neural networks, projections, and subsampling. Under this framework, asymptotic risk expressions for ridge regression and ridgeless regression in under/over-parameterized regimes are provided, revealing the relationship between risk and effective dimension. Based on this theory, a trainable random feature regression model (RFRed) is designed, which minimizes effective dimension by simultaneously optimizing feature mapping, regularization parameters, and subsampling matrices, thereby enhancing generalization ability. Experiments on datasets such as MNIST validate the impact of double descent curves, subsampling, and regularization on generalization, and comparisons are made with traditional ridge regression and random projection methods.
High-Fidelity 3D Head Avatars Reconstruction through Spatially-Varying Expression Conditioned Neural Radiance Field
Minghan Qin (Tsinghua University), Haoqian Wang (Tsinghua University)
GenerationData SynthesisNeural Radiance FieldVideo
🎯 What it does: High-fidelity 3D head avatar reconstruction is achieved by introducing Spatially-Varying Expression (SVE) into NeRF, which provides detailed expression and geometric details for NeRF; a coarse-to-fine training strategy (geometric initialization and adaptive importance sampling) is proposed to enhance geometric and rendering quality.
High-Fidelity Diffusion-Based Image Editing
Chen Hou (University of Science and Technology of China), Zhibo Chen (University of Science and Technology of China)
Image TranslationRestorationGenerationDiffusion modelScore-based ModelRectified FlowImage
🎯 What it does: For the reconstruction and editing tasks of diffusion models, a Rectifier (hypernetwork) is proposed to modulate weights by inserting it into the pre-trained model to bridge the information gap. Additionally, a score-matching-like editing training strategy is introduced, significantly reducing error accumulation during the editing process, thus achieving high-fidelity reconstruction and editing.
High-Fidelity Gradient Inversion in Distributed Learning
Zipeng Ye (Harbin Institute of Technology), Yubo Tang (Harbin Institute of Technology)
RestorationFederated LearningAdversarial AttackConvolutional Neural NetworkImage
🎯 What it does: This paper proposes a two-stage gradient inversion attack, which first uses evolutionary algorithms and feature mutation coefficients to accurately recover the labels of batch training samples, and then achieves high-fidelity batch image recovery through stepwise gradient matching, gradient dropping, and image prior scheduling.
High-Order Structure Based Middle-Feature Learning for Visible-Infrared Person Re-identification
Liuxiang Qiu (Xiamen University), Shunzhi Zhu (University College London)
RecognitionRetrievalConvolutional Neural NetworkTransformerContrastive LearningImageMultimodality
🎯 What it does: HOS-Net is designed, combining short-range and long-range feature extraction, whitening hypergraph learning, and intermediate feature generation to enhance visible-infrared pedestrian re-identification performance.
High-Quality Real-Time Rendering Using Subpixel Sampling Reconstruction
Boyu Zhang (University of California), Hongliang Yuan (Xiaomi Cooperation)
RestorationComputational EfficiencyOptical FlowImageVideo
🎯 What it does: Proposes sub-pixel sampling (sampling only once per 2×2 pixel block, 1/4-spp) along with an SSR network to achieve real-time high-quality rendering.
Higher-Order Graph Convolutional Network with Flower-Petals Laplacians on Simplicial Complexes
Yiming Huang (University of Electronic Science and Technology of China), Linyuan Lü
ClassificationGraph Neural NetworkGraph
🎯 What it does: This paper proposes a high-order graph convolutional network (HiGCN) based on the flower petal-flower petal (FP) model, which models and quantifies high-order interactions through the FP Laplacian and learnable polynomial filters.
HiHPQ: Hierarchical Hyperbolic Product Quantization for Unsupervised Image Retrieval
Zexuan Qiu (Chinese University of Hong Kong), Irwin King (Chinese University of Hong Kong)
RetrievalContrastive LearningImage
🎯 What it does: An unsupervised hierarchical hypersurface product quantization (HiHPQ) method is proposed for efficient image retrieval.
HISR: Hybrid Implicit Surface Representation for Photorealistic 3D Human Reconstruction
Angtian Wang (Johns Hopkins University), Tony Tung (Meta Reality Labs Research)
SegmentationGenerationNeural Radiance FieldGaussian SplattingPoint CloudMesh
🎯 What it does: This paper proposes a Hybrid Implicit Surface Representation (HISR) that combines implicit surfaces (SDF) and volume rendering to achieve layered reconstruction of smooth surfaces and translucent regions.
History Matters: Temporal Knowledge Editing in Large Language Model
Xunjian Yin (Peking University), Xiaojun Wan (Peking University)
TransformerLarge Language ModelPrompt EngineeringTextTime SeriesBenchmark
🎯 What it does: This paper proposes and evaluates the Temporal Knowledge Editing (TKE) task, constructs the ATOKE benchmark dataset, and explores and improves large model knowledge editing methods to retain historical knowledge.
Homophily-Related: Adaptive Hybrid Graph Filter for Multi-View Graph Clustering
Zichen Wen (University of Electronic Science and Technology of China), Lifang He (Lehigh University)
ClassificationOptimizationGraph Neural NetworkAuto EncoderGraph
🎯 What it does: This paper proposes an adaptive hybrid graph filtering method for multi-view graph clustering, which can simultaneously utilize low-frequency (similar) and high-frequency (differential) information to address the clustering challenges of heterogeneous graphs.
HONGAT: Graph Attention Networks in the Presence of High-Order Neighbors
Heng-Kai Zhang (Nanjing University), Yu-Feng Li (Nanjing University)
ClassificationGraph Neural NetworkGraph
🎯 What it does: This paper proposes a graph attention network named HONGAT, which can simultaneously utilize both low-order and high-order neighbor information within a single layer and adjust the aggregation range through a masking mechanism to avoid the over-smoothing problem.
HOP to the Next Tasks and Domains for Continual Learning in NLP
Umberto Michieli (Samsung Research), Mete Ozay (Samsung Research)
ClassificationDomain AdaptationTransformerSupervised Fine-TuningText
🎯 What it does: This paper proposes a continuous learning framework called HOP (High-Order Pooling) for gradually transferring knowledge in multi-task and multi-domain NLP scenarios, avoiding catastrophic forgetting and enhancing forward/backward knowledge transfer.
HORIZON: High-Resolution Semantically Controlled Panorama Synthesis
Kun Yan (Beihang University), Shuai Ma (Microsoft Research Asia)
GenerationData SynthesisTransformerGenerative Adversarial NetworkContrastive LearningImageMultimodality
🎯 What it does: The HORIZON framework is proposed for generating high-resolution 360° panoramic images, achieving semantic control over text and images.
Hot or Cold? Adaptive Temperature Sampling for Code Generation with Large Language Models
Yuqi Zhu (Academy of Military Sciences), Hong Mei (Peking University)
GenerationAI Code AssistantTransformerLarge Language ModelText
🎯 What it does: Researches decoding strategies in code generation and proposes an adaptive temperature sampling method that dynamically adjusts the temperature based on the difficulty of token predictions;
How to Evaluate Behavioral Models
Greg d'Eon (University of British Columbia), James R. Wright (University of Alberta)
OptimizationExplainability and Interpretability
🎯 What it does: A set of axiom-based evaluation loss function framework is proposed for the assessment of behavioral models, and it is proven that the family of loss functions satisfying these axioms is the Diagonal Bounded Bregman Divergence (DBBD), thus recommending the use of squared L2 error.
How to Evaluate the Generalization of Detection? A Benchmark for Comprehensive Open-Vocabulary Detection
Yiyang Yao (Northwestern Polytechnical University), Qing Wang (Northwestern Polytechnical University)
Object DetectionTransformerVision Language ModelImageMultimodalityBenchmark
🎯 What it does: A new open vocabulary detection (OVD) evaluation benchmark, OVDEval, is proposed, covering 9 sub-tasks, with the design of fine-grained hard negative samples and the introduction of the NMS-AP evaluation metric.
How to Make Knockout Tournaments More Popular?
Juhi Chaudhary (Ben-Gurion University of the Negev), Meirav Zehavi (Ben-Gurion University of the Negev)
Optimization
🎯 What it does: This paper proposes and studies the goal of maximizing competition revenue or popularity through the arrangement of the tournament format in single-elimination matches, constructs the corresponding optimization problem, and provides various algorithms and complexity analyses.
How to Overcome Curse-of-Dimensionality for Out-of-Distribution Detection?
Soumya Suvra Ghosal (University of Wisconsin), Yixuan Li (University of Wisconsin)
Anomaly DetectionConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: A new framework called Subspace Nearest Neighbor (SNN) is proposed for detecting out-of-distribution (OOD) samples, aimed at overcoming the curse of dimensionality in high-dimensional feature spaces.
How to Protect Copyright Data in Optimization of Large Language Models?
Timothy Chu (Google), Chiwun Yang (Sun Yat-sen University)
OptimizationSafty and PrivacyTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: A training method called Copyright Regression is proposed to prevent Transformers/LLMs from outputting content similar to copyrighted data in the training set during optimization.
How to Trade Off the Quantity and Capacity of Teacher Ensemble: Learning Categorical Distribution to Stochastically Employ a Teacher for Distillation
Zixiang Ding (Meituan), Wei Lin
CompressionKnowledge DistillationTransformerImageText
🎯 What it does: A dynamic knowledge distillation framework DynaKD is proposed, which utilizes a learnable adaptive class distribution to randomly sample a single teacher from the teacher ensemble during the distillation process, thereby compressing the BERT model and extending it to image classification tasks.
How to Use the Metropolis Algorithm for Multi-Objective Optimization?
Weijie Zheng (Harbin Institute of Technology), Benjamin Doerr (Institut Polytechnique de Paris)
OptimizationBenchmark
🎯 What it does: Study the multi-objective Metropolis algorithm and propose a new benchmark DLTB, conducting theoretical runtime analysis and experimental comparisons.
HR-Pro: Point-Supervised Temporal Action Localization via Hierarchical Reliability Propagation
Huaxin Zhang (Huazhong University of Science and Technology), Nong Sang (Huazhong University of Science and Technology)
RecognitionObject DetectionContrastive LearningVideo
🎯 What it does: The HR-Pro framework is proposed, which enhances temporal action localization performance by leveraging the reliability of point-level annotations through dual propagation at the segment and instance levels.
HuTuMotion: Human-Tuned Navigation of Latent Motion Diffusion Models with Minimal Feedback
Gaoge Han (Northwest A&F University), Jinglei Tang (Northwest A&F University)
GenerationOptimizationDiffusion modelTextMultimodality
🎯 What it does: This paper proposes HuTuMotion, which optimizes the latent space prior using a small amount of human feedback, thereby generating natural and semantically aligned human actions in latent motion diffusion models; it also supports personalized and stylized action generation.
Hybrid-SORT: Weak Cues Matter for Online Multi-Object Tracking
Mingzhan Yang (Dalian University of Technology), Dong Wang (Dalian University of Technology)
Object TrackingVideoBenchmark
🎯 What it does: This paper proposes Hybrid-SORT, which improves the association process of multi-object tracking by incorporating weak cues such as confidence, height state, and velocity direction based on SORT.
Hybrid-Supervised Dual-Search: Leveraging Automatic Learning for Loss-Free Multi-Exposure Image Fusion
Guanyao Wu (Dalian University of Technology), Risheng Liu (Dalian University of Technology)
Image TranslationRestorationOptimizationNeural Architecture SearchImage
🎯 What it does: This paper proposes a Hybrid-Supervised Dual-Search (HSDS-MEF) framework that automates the design of network structures and loss functions for multi-exposure image fusion, thereby improving fusion quality without relying on manual design.
HybridGait: A Benchmark for Spatial-Temporal Cloth-Changing Gait Recognition with Hybrid Explorations
Yilan Dong (ShanghaiTech University), Jingya Wang (ShanghaiTech University)
RecognitionTransformerVideoMeshBenchmark
🎯 What it does: A gait recognition benchmark CCGait is proposed, covering time and spatial clothing changes in the wild, and robust recognition against clothing and viewpoint variations is achieved based on a three-branch hybrid framework called HybridGait.
Hyp-OW: Exploiting Hierarchical Structure Learning with Hyperbolic Distance Enhances Open World Object Detection
Thang Doan (Bosch Research North America), Liu Ren (Bosch Research North America)
Object DetectionTransformerContrastive LearningImage
🎯 What it does: A method for open-world object detection based on hierarchical structure learning and hyperbolic distance, called Hyp-OW, is proposed.