AAAI 2023 Papers — Page 7
AAAI Conference on Artificial Intelligence · 1578 papers
Generic and Dynamic Graph Representation Learning for Crowd Flow Modeling
Liangzhe Han (Beihang University), Tongyu Zhu (Beihang University)
Representation LearningGraph Neural NetworkGraphTime Series
🎯 What it does: A general dynamic graph representation learning framework GDCF is constructed for various crowd flow prediction tasks.
GENNAPE: Towards Generalized Neural Architecture Performance Estimators
Keith G. Mills (University of Alberta), Di Niu (University of Alberta)
Neural Architecture SearchGraph Neural NetworkContrastive LearningGraphBenchmark
🎯 What it does: A general neural network architecture performance estimator, GENNAPE, is proposed, which can make accurate predictions on unseen network structures.
Geometric Inductive Biases for Identifiable Unsupervised Learning of Disentangled Representations
Ziqi Pan (Shanghai Jiao Tong University), Liqing Zhang (Shanghai Jiao Tong University)
GenerationRepresentation LearningAuto EncoderImage
🎯 What it does: This paper proposes two types of geometric inductive biases (α-structure and β-structure) and designs the GDRAE model, unifying PCA bias and geometric bias within an unsupervised decoupled learning framework.
Geometry-Aware Network for Domain Adaptive Semantic Segmentation
Yinghong Liao (Chinese University of Hong Kong), Shuguang Cui (University of Hong Kong)
SegmentationDomain AdaptationGenerative Adversarial NetworkPoint Cloud
🎯 What it does: A two-stage geometry-aware network GANDA is proposed, which predicts the depth of the target domain using the depth information from the source domain, constructs RGB-D point clouds, performs semantic and geometric segmentation using 3D point clouds, and achieves domain adaptation through geometric alignment and coordinate-color decoupling.
GLCC: A General Framework for Graph-Level Clustering
Wei Ju (Peking University), Ming Zhang (Peking University)
Graph Neural NetworkContrastive LearningGraphBiomedical Data
🎯 What it does: Proposed the Graph-Level Contrastive Clustering (GLCC) framework to achieve graph-level clustering in multi-graph scenarios;
Global Concept-Based Interpretability for Graph Neural Networks via Neuron Analysis
Han Xuanyuan (University of Cambridge), Pietro Liò (University of Cambridge)
Explainability and InterpretabilityGraph Neural NetworkGraph
🎯 What it does: This study investigates the concept detection capability of neurons in Graph Neural Networks (GNN) and proposes a global interpretability framework.
Global Convergence of Two-Timescale Actor-Critic for Solving Linear Quadratic Regulator
Xuyang Chen (National University of Singapore), Lin Zhao (National University of Singapore)
OptimizationReinforcement Learning
🎯 What it does: The paper studies the convergence of the Single-sample Two-Timescale Natural Actor-Critic algorithm on the Linear Quadratic Regulator (LQR) problem, providing a global convergence proof and a sample complexity upper bound of O(ε⁻²·⁵).
Global Dilated Attention and Target Focusing Network for Robust Tracking
Yun Liang (South China Agricultural University), Fumian Long (South China Agricultural University)
Object TrackingConvolutional Neural NetworkImageVideo
🎯 What it does: A new visual tracking framework GdaTFT is proposed, which includes Global Dilated Attention (GDA) and Target Focusing Network (TFN) to enhance feature semantic representation and strengthen target embedding.
Global Mixup: Eliminating Ambiguity with Clustering
Xiangjin Xie (Shenzhen International Graduate School Tsinghua University), Hai-Tao Zheng (Shenzhen International Graduate School Tsinghua University)
ClassificationData SynthesisConvolutional Neural NetworkRecurrent Neural NetworkTransformerText
🎯 What it does: Global Mixup is proposed, which splits the sample generation of data augmentation and label determination into two stages, and re-labels virtual samples using global clustering relationships, addressing the label ambiguity problem of traditional Mixup.
Global-Local Characteristic Excited Cross-Modal Attacks from Images to Videos
Ruikui Wang (Beihang University), Yunhong Wang (Beihang University)
Adversarial AttackConvolutional Neural NetworkImageVideo
🎯 What it does: A cross-modal adversarial attack method from images to videos is proposed, enhancing the cross-modal transferability of adversarial samples through global interaction enhancement and local relevant perturbation techniques.
GLT-T: Global-Local Transformer Voting for 3D Single Object Tracking in Point Clouds
Jiahao Nie (Hangzhou Dianzi University), Jing Zhang (University of Sydney)
Object TrackingAutonomous DrivingTransformerPoint Cloud
🎯 What it does: A global-local Transformer voting framework GLT-T is proposed to improve the seed point voting of VoteNet for higher quality 3D single-object tracking.
GLUECons: A Generic Benchmark for Learning under Constraints
Hossein Rajaby Faghihi (Michigan State University), Parisa Kordjamshidi (University of Pennsylvania)
ImageTextBenchmark
🎯 What it does: The GLUECons benchmark is proposed for systematically evaluating the effects of integrating constraint knowledge into deep learning models across different tasks.
GMDNet: A Graph-Based Mixture Density Network for Estimating Packages’ Multimodal Travel Time Distribution
Xiaowei Mao (Beijing Jiaotong University), Youfang Lin (Cainiao Network)
Graph Neural NetworkMixture of ExpertsGraphTime Series
🎯 What it does: This paper proposes a Graph Mixture Density Network (GMDNet) for accurately predicting the multimodal travel time distribution of packages in logistics networks based on routes and graph structures.
Goal-Conditioned Generators of Deep Policies
Francesco Faccio (AI Initiative), Jürgen Schmidhuber
GenerationReinforcement LearningSequential
🎯 What it does: This paper proposes a goal-conditioned policy generator GoGePo, which directly generates deep neural network policies based on the 'expected return' command using Fast Weight Programmer (FWP) and hypernetworks;
Goal-Conditioned Q-learning as Knowledge Distillation
Alexander Levine (University of Maryland), Soheil Feizi (University of Maryland)
Knowledge DistillationReinforcement Learning
🎯 What it does: Two target-conditioned offline reinforcement learning algorithms based on gradient attention transfer, ReenGAGE and Multi-ReenGAGE, are proposed, and their performance is validated in high-dimensional target spaces and multi-target sparse reward environments.
GOHSP: A Unified Framework of Graph and Optimization-Based Heterogeneous Structured Pruning for Vision Transformer
Miao Yin (Rutgers University), Bo Yuan (Samsung Research America)
CompressionOptimizationTransformerImage
🎯 What it does: Proposed and implemented the GOHSP framework, which utilizes graph models and optimization methods for heterogeneous structured pruning of Vision Transformers.
Good Helper Is around You: Attention-Driven Masked Image Modeling
Zhengqi Liu (Southeast University), Hao Luo (Alibaba Group)
Object DetectionSegmentationRepresentation LearningTransformerContrastive LearningImage
🎯 What it does: A mask and drop strategy based on Vision Transformer self-attention (AMT) is proposed in self-supervised image pre-training, guided by unsupervised saliency maps to enhance representation learning effectiveness and efficiency.
GPTR: Gestalt-Perception Transformer for Diagram Object Detection
Xin Hu (Shaanxi Provincial Key Laboratory of Big Data Knowledge Engineering, Xi'an Jiaotong University), Yaqiang Wu (Lenovo Research)
Object DetectionTransformerImage
🎯 What it does: A Transformer based on Gestalt perception (GPTR) is proposed for chart object detection, utilizing the sparse visual features and low-frequency categories of charts.
Gradient Corner Pooling for Keypoint-Based Object Detection
Xuyang Li (Xidian University), Guangming Shi (Xidian University)
Object DetectionConvolutional Neural NetworkImage
🎯 What it does: A new keypoint detection framework called Gradient Corner Pooling (GCP) is studied to improve the localization accuracy of corner-based object detection when similar objects are densely arranged.
Gradient Estimation for Binary Latent Variables via Gradient Variance Clipping
Russell Z. Kunes (Columbia University), Simon Tavaré (Columbia University)
OptimizationReinforcement LearningTabular
🎯 What it does: This paper presents a technique for improving gradient estimation in Bernoulli discrete latent variable models, addressing the issue of variance explosion at parameter boundaries with traditional estimators.
Gradient-Adaptive Pareto Optimization for Constrained Reinforcement Learning
Zixian Zhou (Chinese Academy of Sciences), Qing He (Huawei)
OptimizationReinforcement LearningSequentialBenchmark
🎯 What it does: A new constrained reinforcement learning (CRL) method called GCPO is proposed, which utilizes Pareto optimization ideas and achieves a balance between rewards and costs through two gradient recalibration techniques to meet constraints.
Gradient-Based Graph Attention for Scene Text Image Super-resolution
Xiangyuan Zhu (Central South University), Gerald Schaefer (Loughborough University)
RestorationSuper ResolutionGraph Neural NetworkImageBenchmark
🎯 What it does: A gradient map-based graph attention network is proposed, combining group attention and channel attention to achieve text image super-resolution.
Gradient-Variation Bound for Online Convex Optimization with Constraints
Shuang Qiu (University of Chicago), Mladen Kolar (University of Chicago)
Optimization
🎯 What it does: This paper studies online convex optimization problems with multiple functional constraints and proposes a new online primal-dual mirror-approximation algorithm aimed at achieving low regret and constraint violation within a time step T.
GradPU: Positive-Unlabeled Learning via Gradient Penalty and Positive Upweighting
Songmin Dai (Shanghai University), Tong Liu (Shanghai University)
ClassificationConvolutional Neural NetworkImage
🎯 What it does: A new positive-negative semi-supervised learning method called GradPU is proposed, which utilizes gradient penalty and positive sample weighting to alleviate the overfitting problem of positive classes.
Graph Anomaly Detection via Multi-Scale Contrastive Learning Networks with Augmented View
Jingcan Duan (National University of Defense Technology), Zhibin Dong (National University of Defense Technology)
Anomaly DetectionGraph Neural NetworkContrastive LearningGraph
🎯 What it does: This paper proposes the GRADATE framework for the task of graph anomaly detection, employing multi-scale contrastive learning (node-subgraph, node-node, subgraph-subgraph) and achieving multi-view enhancement of graphs through edge modification to improve the robustness of subgraph embeddings and detection performance.
Graph Component Contrastive Learning for Concept Relatedness Estimation
Yueen Ma (Chinese University of Hong Kong), Irwin King (Chinese University of Hong Kong)
Representation LearningGraph Neural NetworkTransformerContrastive LearningGraph
🎯 What it does: Construct a concept-level graph called ConcreteGraph, and use it for data augmentation and graph component contrastive learning to address the high-order relationships in concept relevance estimation and the issue of data scarcity.
Graph Knows Unknowns: Reformulate Zero-Shot Learning as Sample-Level Graph Recognition
Jingcai Guo (Hong Kong Polytechnic University), Fushuo Huo (Hong Kong Polytechnic University)
ClassificationRecognitionGraph Neural NetworkImage
🎯 What it does: Each image is split into several fine-grained components, and these components and their mutual influences form a sample-level graph, which is then mapped to the semantic space using a graph neural network to achieve zero-shot classification.
Graph Ordering Attention Networks
Michail Chatzianastasis (Ecole Polytechnique), Michalis Vazirgiannis (Ecole Polytechnique)
Recurrent Neural NetworkGraph Neural NetworkGraph
🎯 What it does: A new graph neural network layer (GOAT) is proposed, which captures the collaborative information between neighbors by learning neighbor ranking and using recurrent networks for aggregation.
Graph Structure Learning on User Mobility Data for Social Relationship Inference
Guangming Qin (Ocean University of China), Junyu Dong (Ocean University of China)
Recommendation SystemGraph Neural NetworkGraph
🎯 What it does: Proposes the SRINet framework, which utilizes graph structure learning to remove noise edges from user mobility data and applies GCN for social relationship inference on multi-semantic meeting graphs.
Graphix-T5: Mixing Pre-trained Transformers with Graph-Aware Layers for Text-to-SQL Parsing
Jinyang Li (University of Hong Kong), Yongbin Li (Alibaba Group)
Graph Neural NetworkTransformerLarge Language ModelText
🎯 What it does: The GRAPHIX-T5 model is proposed, which integrates the pre-trained T5 with graph neural networks to enhance the reasoning and structured generation capabilities from cross-domain text to SQL.
GraphPrompt: Graph-Based Prompt Templates for Biomedical Synonym Prediction
Hanwen Xu (University of Washington), Sheng Wang (University of Washington)
Drug DiscoveryGraph Neural NetworkPrompt EngineeringGraphBiomedical Data
🎯 What it does: By transforming the ontology graph structure into prompt templates, a masked language model is utilized to complete the biomedical synonym prediction task.
Graphs, Constraints, and Search for the Abstraction and Reasoning Corpus
Yudong Xu (University of Toronto), Scott Sanner (University of Toronto)
Graph Neural NetworkGraph
🎯 What it does: This paper proposes an object-centric framework called ARGA, based on graph abstraction and constraint-guided search, to solve ARC tasks.
GraphSR: A Data Augmentation Algorithm for Imbalanced Node Classification
Mengting Zhou (University of Macau), Zhiguo Gong (University of Macau)
ClassificationGraph Neural NetworkReinforcement LearningGraph
🎯 What it does: The GraphSR algorithm is designed to automatically complete minority class nodes from a large number of unlabeled nodes using self-supervised training combined with similarity screening and reinforcement learning, addressing the class imbalance issue in node classification.
GRASMOS: Graph Signage Model Selection for Gene Regulatory Networks
Angelina Brilliantova (Rochester Institute of Technology), Ivona Bezáková (Rochester Institute of Technology)
Graph Neural NetworkGraph
🎯 What it does: This paper proposes a maximum likelihood-based graph identification model selection method (GRASMOS) for inferring the distribution of positive and negative edges in a gene regulatory network with a given topology.
GRIP: Graph Representation of Immune Repertoire Using Graph Neural Network and Transformer
Yongju Lee (Seoul National University), Sunghoon Kwon (Seoul National University)
Graph Neural NetworkTransformerGraphBiomedical Data
🎯 What it does: This paper proposes a method to represent the immune receptor repertoire as a hierarchical graph structure and uses graph neural networks and Transformers to predict the survival risk of cancer patients.
GRLSTM: Trajectory Similarity Computation with Graph-Based Residual LSTM
Silin Zhou (University of Electronic Science and Technology of China), Peng Han (University of Electronic Science and Technology of China)
Recurrent Neural NetworkGraph Neural NetworkGraphTime SeriesSequential
🎯 What it does: This paper proposes the GRLSTM framework, which implements trajectory similarity calculation based on road networks using knowledge graph embedding, graph attention networks, and residual multi-layer LSTM.
Grouped Knowledge Distillation for Deep Face Recognition
Weisong Zhao (Institute of Information Engineering, Chinese Academy of Sciences), Zhen Lei (Institute of Automation, Chinese Academy of Sciences)
RecognitionKnowledge DistillationImage
🎯 What it does: This paper proposes a new knowledge distillation method—Grouped Knowledge Distillation (GKD), specifically designed for large-scale face recognition tasks.
Grouping Matrix Based Graph Pooling with Adaptive Number of Clusters
Sung Moon Ko (LG AI Research), Honglak Lee (LG AI Research)
Drug DiscoveryGraph Neural NetworkGraph
🎯 What it does: A new differentiable graph pooling method called GMPOOL is proposed, which can automatically determine the appropriate number of clusters without pre-setting the clustering number;
GuidedMixup: An Efficient Mixup Strategy Guided by Saliency Maps
Minsoo Kang (Korea University), Suhyun Kim (Korea Institute of Science and Technology)
ClassificationData SynthesisComputational EfficiencyImage
🎯 What it does: This paper proposes GuidedMixup, an efficient Mixup data augmentation method based on saliency maps, which first minimizes saliency region conflicts using a greedy pairing algorithm, and then generates mixed images according to pixel-level mixing ratios.
H-TSP: Hierarchically Solving the Large-Scale Traveling Salesman Problem
Xuanhao Pan (Huazhong University of Science and Technology), Jiang Bian (University of Science and Technology of China)
OptimizationConvolutional Neural NetworkTransformerReinforcement LearningGraph
🎯 What it does: An end-to-end framework H-TSP based on hierarchical reinforcement learning is proposed to solve large-scale (up to 10,000 nodes) traveling salesman problems;
HALOC: Hardware-Aware Automatic Low-Rank Compression for Compact Neural Networks
Jinqi Xiao (Rutgers University), Bo Yuan (Washington State University)
CompressionNeural Architecture SearchConvolutional Neural NetworkImage
🎯 What it does: The HALOC (Hardware-Aware Automatic Low-Rank Compression) framework is proposed, which automatically selects the rank of low-rank decomposition for each layer of the convolutional network in a differentiable manner using the idea of NAS, thereby compressing the network while considering hardware performance.
Handling Missing Data via Max-Entropy Regularized Graph Autoencoder
Ziqi Gao (Hong Kong University of Science and Technology), Jia Li (Hong Kong University of Science and Technology)
Graph Neural NetworkAuto EncoderGraph
🎯 What it does: This paper proposes a Maximum Entropy-based Graph Autoencoder (MEGAE) that alleviates the spectral concentration problem in traditional Graph Autoencoders (GAE) by maximizing graph entropy, achieving more accurate imputation of missing graph attributes.
Hard Sample Aware Network for Contrastive Deep Graph Clustering
Yue Liu (National University of Defense Technology), Cancan Chen (National University of Defense Technology)
Graph Neural NetworkContrastive LearningGraph
🎯 What it does: A Hard Sample Aware Network (HSAN) is designed for deep graph clustering, improving clustering performance through dynamic weighting of hard positive and negative samples.
HAVEN: Hierarchical Cooperative Multi-Agent Reinforcement Learning with Dual Coordination Mechanism
Zhiwei Xu (Institute of Automation Chinese Academy of Sciences), Guoliang Fan (Institute of Automation Chinese Academy of Sciences)
Reinforcement LearningSequential
🎯 What it does: A multi-agent hierarchical value decomposition framework named HAVEN is proposed, which uses a dual-layer coordination mechanism to achieve simultaneous learning of high-level and low-level policies.
Head-Free Lightweight Semantic Segmentation with Linear Transformer
Bo Dong (Alibaba Group), Fan Wang (Amazon Prime Video)
SegmentationTransformerImage
🎯 What it does: A headless lightweight semantic segmentation framework AFFormer is proposed, which directly utilizes prototype learning and adaptive frequency filtering on high-resolution features to achieve semantic recovery, completely eliminating the traditional decoder.
Heterogeneous Graph Learning for Multi-Modal Medical Data Analysis
Sein Kim (Korea Advanced Institute of Science and Technology), Chanyoung Park (Korea Advanced Institute of Science and Technology)
ClassificationSegmentationRecommendation SystemGraph Neural NetworkContrastive LearningMultimodalityBiomedical DataMagnetic Resonance ImagingAlzheimer's Disease
🎯 What it does: A multi-layer graph learning-based HetMed framework is proposed to integrate medical imaging and non-imaging multimodal data for clinical decision-making.
Heterogeneous Graph Masked Autoencoders
Yijun Tian (University of Notre Dame), Nitesh V. Chawla (Brandeis University)
GenerationRepresentation LearningGraph Neural NetworkAuto EncoderGraph
🎯 What it does: This paper proposes HGMAE, a generative self-supervised masked autoencoder for heterogeneous graphs, aimed at learning node representations in the absence of labels.
Heterogeneous Region Embedding with Prompt Learning
Silin Zhou (University of Electronic Science and Technology of China), Peng Han (University of Electronic Science and Technology of China)
Graph Neural NetworkPrompt EngineeringGraph
🎯 What it does: This paper proposes the HREP framework, which constructs a heterogeneous graph containing human flow, POI, and geographical adjacency relationships. It utilizes a relation-aware GCN and self-attention fusion to obtain high-quality regional embeddings, and enhances performance in downstream tasks through prefix-tuning for prompt learning.
Heterogeneous-Branch Collaborative Learning for Dialogue Generation
Yiwei Li (Beijing Institute of Technology), Kan Li (Beijing Institute of Technology)
GenerationKnowledge DistillationTransformerText
🎯 What it does: A collaborative learning framework composed of a main branch and an attribute-specific auxiliary branch is constructed to achieve mutual transmission and integration of multi-attribute knowledge in dialogue generation.
Heuristic Search for Multi-Objective Probabilistic Planning
Dillon Z. Chen (Australian National University), Sylvie Thiébaux (Australian National University)
OptimizationReinforcement Learning from Human FeedbackTabularBenchmark
🎯 What it does: This paper defines the Multi-Objective Stochastic Shortest Path (MOSSP) problem and proposes a heuristic search-based solution algorithm.
HG-SL: Jointly Learning of Global and Local User Spreading Behavior for Fake News Early Detection
Ling Sun (Xi'an Jiaotong University), Yangyang Li (Xi'an Jiaotong University)
ClassificationAnomaly DetectionGraph Neural NetworkGraph
🎯 What it does: This paper proposes a text-free and user identity information-free early detection model for fake news, HG-SL, which utilizes both global and local features of user propagation behavior to distinguish between true and false news.
Hierarchical Consistent Contrastive Learning for Skeleton-Based Action Recognition with Growing Augmentations
Jiahang Zhang (Peking University), Jiaying Liu (Peking University)
RecognitionGraph Neural NetworkTransformerContrastive LearningVideoMultimodality
🎯 What it does: Proposes HiCLR: a hierarchical consistency contrastive learning framework that enhances skeleton action recognition through progressive augmentation and asymmetric consistency loss.
Hierarchical Contrast for Unsupervised Skeleton-Based Action Representation Learning
Jianfeng Dong (Zhejiang Gongshang University), Xun Wang (Zhejiang Gongshang University)
RecognitionRetrievalRepresentation LearningContrastive LearningVideo
🎯 What it does: A hierarchical contrastive learning framework named HiCo is proposed for unsupervised skeleton action representation learning.
Hierarchical Contrastive Learning for Temporal Point Processes
Qingmei Wang (Renmin University of China), Hongteng Xu (Renmin University of China)
TransformerContrastive LearningTime SeriesSequential
🎯 What it does: A hierarchical contrastive learning (HCL) method is proposed to regularize temporal point process models based on maximum likelihood estimation.
Hierarchical ConViT with Attention-Based Relational Reasoner for Visual Analogical Reasoning
Wentao He (University of Nottingham), Xudong Jiang (Nanyang Technological University)
RecognitionConvolutional Neural NetworkTransformerImage
🎯 What it does: A model based on hierarchical ConViT and an attention-based relational reasoner is proposed to address the visual analogy reasoning task in Raven Progressive Matrices (RPM).
Hierarchical Event Grounding
Jiefu Ou (Carnegie Mellon University), Teruko Mitamura (Carnegie Mellon University)
RetrievalTransformerSupervised Fine-TuningText
🎯 What it does: Designed and implemented a hierarchical event alignment task, constructed a multilingual event hierarchy dataset, and proposed a retrieval method based on hierarchical loss.
Hierarchical Mean-Field Deep Reinforcement Learning for Large-Scale Multiagent Systems
Chao Yu (Sun Yat-sen University)
Reinforcement Learning
🎯 What it does: Proposes a Hierarchical Mean-Field (HMF) framework that utilizes intra-group local Q and group mean field to achieve coordinated learning in large-scale multi-agent systems;
Hierarchical Text Classification as Sub-hierarchy Sequence Generation
SangHun Im (Korea University of Technology and Education), Dong Hwan Kim (Korea University of Technology and Education)
ClassificationTransformerText
🎯 What it does: This paper proposes transforming hierarchical text classification into sub-hierarchical sequence generation and designs the HiDEC recursive sub-hierarchical decoder, which utilizes hierarchical-aware masked self-attention and text-hierarchical attention to achieve unstructured encoding of hierarchical information modeling.
High-Dimensional Dueling Optimization with Preference Embedding
Yangwenhui Zhang (East China Normal University), Aimin Zhou (East China Normal University)
OptimizationDrug DiscoveryTabular
🎯 What it does: This paper proposes Preference-Embedded Dual Decision Bayesian Optimization (PE-DBO), which addresses the scalability issue of high-dimensional dual decision black-box optimization by optimizing in a low-dimensional subspace and mapping the pairwise comparisons back to the original high-dimensional space.
High-Level Semantic Feature Matters Few-Shot Unsupervised Domain Adaptation
Lei Yu (Nanjing Normal University), Ming Yang (University of Wollongong)
Domain AdaptationImage
🎯 What it does: The TSECS method is proposed, utilizing task-specific high-level semantic features and cross-domain self-training to achieve few-shot unsupervised domain adaptation for classification and domain alignment.
High-Resolution GAN Inversion for Degraded Images in Large Diverse Datasets
Yanbo Wang (East China Normal University), Yuan Xie (East China Normal University)
Image TranslationRestorationGenerative Adversarial NetworkImage
🎯 What it does: A GAN inverse framework CRi is proposed for degraded image restoration using StyleGAN-XL.
High-Resolution Iterative Feedback Network for Camouflaged Object Detection
Xiaobin Hu (Tencent Youtu Lab), Ling Shao (Terminus Group)
Object DetectionSegmentationTransformerGenerative Adversarial NetworkImage
🎯 What it does: A high-resolution iterative feedback network, HitNet, is proposed to refine low-resolution features through iterative feedback of high-resolution features in covert target detection, thereby improving the segmentation quality of boundaries and details.
How to Cut a Discrete Cake Fairly
Ayumi Igarashi (University of Tokyo)
🎯 What it does: In the fair distribution of discrete cakes, it is proven that for any number of agents with monotonic values, there always exists a connected allocation scheme that satisfies EF1_outer.
HRDoc: Dataset and Baseline Method toward Hierarchical Reconstruction of Document Structures
Jiefeng Ma (University of Science and Technology of China), Cong Liu (iFLYTEK Research)
ClassificationSegmentationRecurrent Neural NetworkTransformerSupervised Fine-TuningTextMultimodality
🎯 What it does: Proposes a hierarchical document structure reconstruction task and implements an end-to-end parsing system (DSPS).
Human Assisted Learning by Evolutionary Multi-Objective Optimization
Dan-Xuan Liu (Nanjing University), Chao Qian (Nanjing University)
OptimizationBiomedical Data
🎯 What it does: A framework based on evolutionary multi-objective optimization, HAL-EMO, is proposed to address the instance allocation problem in human-machine collaborative learning.
Human Joint Kinematics Diffusion-Refinement for Stochastic Motion Prediction
Dong Wei (Nanjing University of Science and Technology), Shengxiang Hu (Tianjin AiForward Science and Technology)
GenerationPose EstimationGraph Neural NetworkTransformerDiffusion modelGenerative Adversarial NetworkVideoMultimodality
🎯 What it does: This paper proposes MotionDiff, a multimodal human motion prediction framework based on diffusion models.
Human-in-the-Loop Vehicle ReID
Zepeng Li (Zhejiang University), Gang Chen (Shanghai Jiao Tong University)
RecognitionRetrievalAutonomous DrivingTransformerContrastive LearningImage
🎯 What it does: This paper proposes a human-machine interactive vehicle ReID framework called IRIN, which directly uses user-provided positive sample feedback as input to the network, allowing for immediate adjustment of the embedded representation of the query image during online inference.
Human-Instructed Deep Hierarchical Generative Learning for Automated Urban Planning
Dongjie Wang (University of Central Florida), Yanjie Fu (University of Central Florida)
GenerationData SynthesisOptimizationGraph Neural NetworkAuto EncoderGenerative Adversarial NetworkGraphTabular
🎯 What it does: Automated urban planning generates land use configurations through human instructions and spatial context, employing a hierarchical generation process that first produces functional areas and then refines grids.
HVTSurv: Hierarchical Vision Transformer for Patient-Level Survival Prediction from Whole Slide Image
Zhuchen Shao (Tsinghua University), Yongbing Zhang (Harbin Institute of Technology)
TransformerImageBiomedical Data
🎯 What it does: Utilizing hierarchical visual Transformers for patient-level survival prediction on whole slide images.
Hybrid CNN-Transformer Feature Fusion for Single Image Deraining
Xiang Chen (Nanjing University of Science and Technology), Hao Li (Shenyang Aerospace University)
RestorationConvolutional Neural NetworkTransformerMixture of ExpertsImage
🎯 What it does: A lightweight hybrid CNN-Transformer feature fusion network HCT-FFN is proposed, achieving single image deraining in a staged progressive manner.
Hybrid Learning with New Value Function for the Maximum Common Induced Subgraph Problem
Yanli Liu (Wuhan University of Science and Technology), Kun He (Yunnan University)
OptimizationReinforcement LearningGraphBiomedical Data
🎯 What it does: A new value function and hybrid branching strategy are proposed to improve the search process of the Maximum Common Induced Subgraph (MCIS) in the Branch-and-Bound algorithm.
Hybrid Pixel-Unshuffled Network for Lightweight Image Super-resolution
Bin Sun (Northeastern University), Yun Fu (Northeastern University)
RestorationSuper ResolutionConvolutional Neural NetworkImage
🎯 What it does: A lightweight image super-resolution network called Hybrid Pixel-Unshuffled Network (HPUN) is proposed, which enhances feature representation through self-residual depthwise separable convolutions and pixel-unshuffle downsampling.
HybridCap: Inertia-Aid Monocular Capture of Challenging Human Motions
Han Liang (ShanghaiTech University), Lan Xu (ShanghaiTech University)
Pose EstimationOptimizationVideoMultimodality
🎯 What it does: This paper proposes HybridCap, which achieves real-time high-quality 3D capture of challenging human actions in a lightweight setup with a single camera and four IMUs.
HybridPrompt: Bridging Language Models and Human Priors in Prompt Tuning for Visual Question Answering
Zhiyuan Ma (Huazhong University of Science and Technology), Guohui Li (Huazhong University of Science and Technology)
TransformerPrompt EngineeringContrastive LearningImageTextMultimodality
🎯 What it does: Proposes the HybridPrompt framework, which combines language models with human priors to fine-tune visual question answering models through a mix of cloze and verify-style prompts;
HyperJump: Accelerating HyperBand via Risk Modelling
Pedro Mendes (Instituto Superior Técnico, Universidade de Lisboa), David Garlan (Carnegie Mellon University)
OptimizationHyperparameter SearchImageTabular
🎯 What it does: This paper proposes HyperJump, a hyperparameter optimization method that accelerates the process through risk modeling based on HyperBand.
Hypernetworks for Zero-Shot Transfer in Reinforcement Learning
Sahand Rezaei-Shoshtari (McGill University), David Meger (McGill University)
Reinforcement LearningSequential
🎯 What it does: In continuous control tasks, hypernetwork learning is used to map MDP parameters (rewards and dynamics) to near-optimal policies and value functions, achieving zero-shot transfer.
Hypotheses Tree Building for One-Shot Temporal Sentence Localization
Daizong Liu (Huazhong University of Science and Technology), Yu Cheng (Microsoft Research)
RecognitionRepresentation LearningRecurrent Neural NetworkContrastive LearningVideoText
🎯 What it does: A multi-level tree structure model MHST is proposed, which achieves one-shot temporal sentence localization using single-frame annotations, automatically generating multiple candidate segments and performing self-supervised learning.
i-Code: An Integrative and Composable Multimodal Learning Framework
Ziyi Yang (Microsoft Azure Cognitive Services Research), Xuedong Huang (Microsoft Azure Cognitive Services Research)
Representation LearningTransformerMixture of ExpertsContrastive LearningVideoTextMultimodalityAudio
🎯 What it does: This paper proposes i-Code, a self-supervised pre-training framework that can dynamically combine visual, language, and speech modalities, achieving unified representation through cross-modal fusion.
I’m Me, We’re Us, and I’m Us: Tri-directional Contrastive Learning on Hypergraphs
Dongjin Lee (Korea Advanced Institute of Science and Technology), Kijung Shin (Korea Advanced Institute of Science and Technology)
Representation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: This paper proposes a three-way contrastive learning framework, TriCL, for unsupervised hypergraph representation learning, which can simultaneously capture structural information of nodes, hyperedges, and node-hyperedge relationships.
Identification and Estimation of the Probabilities of Potential Outcome Types Using Covariate Information in Studies with Non-compliance
Yuta Kawakami (Yokohama National University), Manabu Kuroki (Yokohama National University)
Tabular
🎯 What it does: Proposes the use of covariate information in random experiments to identify and estimate the probabilities of 16 types of potential outcomes, addressing the issue of non-compliance;
Identify Event Causality with Knowledge and Analogy
Sifan Wu (University of Montreal), Bang Liu (University of Montreal)
RecognitionGraph Neural NetworkTextRetrieval-Augmented Generation
🎯 What it does: This paper studies the task of event causal relationship identification and proposes the KADE framework, which utilizes dual enhancement through external knowledge and internal analogy.
Identifying and Eliminating Majority Illusion in Social Networks
Umberto Grandi (University of Toulouse), Paolo Turrini (University of Warwick)
🎯 What it does: This paper systematically studies the phenomenon of majority illusion in social networks, proving the NP-hardness of its decision and elimination problems, and providing several parameterized solvable solutions.
Identifying Selection Bias from Observational Data
David Kaltenpoth (CISPA Helmholtz Center for Information Security), Jilles Vreeken (CISPA Helmholtz Center for Information Security)
Tabular
🎯 What it does: This paper studies whether it is possible to identify and estimate whether a sample is affected by selection bias with only a single observation sample, and proposes two practical methods: a parameter estimation method based on exponential family models (EXP) and an unsupervised method based on distribution invariance learning (INV).
IKOL: Inverse Kinematics Optimization Layer for 3D Human Pose and Shape Estimation via Gauss-Newton Differentiation
Juze Zhang (ShanghaiTech University), Jingya Wang (ShanghaiTech University)
Pose EstimationOptimizationConvolutional Neural NetworkAuto EncoderImage
🎯 What it does: This paper proposes an Inverse Kinematics Optimization Layer (IKOL) that combines optimization and regression to achieve 3D human pose and shape estimation within an end-to-end framework.
ILSGAN: Independent Layer Synthesis for Unsupervised Foreground-Background Segmentation
Qiran Zou (Tsinghua University), Xiangyang Ji (Tsinghua University)
SegmentationGenerationGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes an Independent Layered Synthesis Generative Adversarial Network (ILSGAN) for unsupervised foreground-background segmentation.
ImageNet Pre-training Also Transfers Non-robustness
Jiaming Zhang (Beijing Jiaotong University), Jian Yu (Beijing Normal University)
ClassificationAdversarial AttackConvolutional Neural NetworkSupervised Fine-TuningImage
🎯 What it does: This paper reveals through experiments and analysis that ImageNet pre-training transfers non-robust features to downstream models in transfer learning, resulting in poor performance of downstream models under adversarial attacks.
Imbalanced Label Distribution Learning
Xingyu Zhao (Southeast University), Xin Geng (Southeast University)
ClassificationDomain AdaptationContrastive LearningTabular
🎯 What it does: This paper studies the imbalance problem in label distribution learning and proposes a model that aligns the distribution of feature representations and label representations to address the distribution shift between the training set and the test set.
ImGCL: Revisiting Graph Contrastive Learning on Imbalanced Node Classification
Liang Zeng (Tsinghua University), Jian Li (Hong Kong University of Science and Technology)
ClassificationRepresentation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: The ImGCL framework is proposed, which generates pseudo-labels through online clustering and combines node centrality for progressive balanced sampling, addressing the poor representation issue of graph contrastive learning under node class imbalance.
Imperceptible Adversarial Attack via Invertible Neural Networks
Zihan Chen (National University of Defense Technology), Dejian Guan (National University of Defense Technology)
Adversarial AttackFlow-based ModelImage
🎯 What it does: A framework for adversarial attacks called AdvINN is proposed, which utilizes Invertible Neural Networks (INN) to generate adversarial samples with a high attack success rate without significantly altering the visual quality of the images.
Implementing Bounded Revision via Lexicographic Revision and C-revision
Meliha Sezgin (TU Dortmund University), Gabriele Kern-Isberner (TU Dortmund University)
🎯 What it does: This paper proposes and implements 'Bounded Revision'—an iterative belief revision method that limits the degree of acceptance of new information by referencing sentences.
Implicit Stochastic Gradient Descent for Training Physics-Informed Neural Networks
Ye Li (Nanjing University of Aeronautics and Astronautics), Sheng-Jun Huang (Nanjing University of Aeronautics and Astronautics)
OptimizationPhysics RelatedOrdinary Differential Equation
🎯 What it does: This paper proposes the use of Implicit Stochastic Gradient Descent (ISGD) to train Physics-Informed Neural Networks (PINN) to address the training instability of traditional gradient descent in high-frequency or multi-scale problems.
Improved Algorithm for Regret Ratio Minimization in Multi-Objective Submodular Maximization
Yanhao Wang (East China Normal University), Fanxu Meng (Nanjing University)
OptimizationGraph
🎯 What it does: This paper studies the return ratio minimization (RRM) problem in multi-objective submodular maximization and proposes the HS-RRM algorithm based on ε-kernel, δ-net, and HITTINGSET transformation.
Improved Algorithms for Maximum Satisfiability and Its Special Cases
Kirill Brilliantov (Constructor University), Ivan Bliznets (Utrecht University)
Optimization
🎯 What it does: A new algorithm for MAXSAT is proposed, improving the time complexity of the (n,3) and (n,4) versions, and providing a better exponential upper bound for global MAXSAT.
Improved Kernel Alignment Regret Bound for Online Kernel Learning
Junfan Li (Tianjin University), Shizhong Liao (Tianjin University)
OptimizationTabular
🎯 What it does: A new online kernel learning algorithm POMDR is proposed, which improves the kernel alignment regret bound under hinge loss and reduces computational complexity.
Improvement-Focused Causal Recourse (ICR)
Gunnar König (University of Vienna), Moritz Grosse-Wentrup (University of Vienna)
Tabular
🎯 What it does: An improved guided causal recovery (ICR) method is proposed, providing actionable recommendations for individuals affected by adverse decisions, which can both increase the acceptance of predictions (ˆY=1) and genuinely improve the true target Y.
Improving Biomedical Entity Linking with Cross-Entity Interaction
Zhenran Xu (Harbin Institute of Technology), Baotian Hu (Harbin Institute of Technology)
Drug DiscoveryTransformerPrompt EngineeringBiomedical Data
🎯 What it does: This paper proposes a cross-entity interactive re-ranking model based on prompt tuning to address the ambiguity problem in biomedical entity linking.
Improving Crowded Object Detection via Copy-Paste
Jiangfan Deng (Aibee Inc), Feng Zhou (Aibee Inc)
Object DetectionData SynthesisImage
🎯 What it does: Proposes a copy-paste based data augmentation for crowded scenes, combined with consensus learning and pseudo-depth information to alleviate the issues of overlap and deduplication in object detection.
Improving Distantly Supervised Relation Extraction by Natural Language Inference
Kang Zhou (Iowa State University), Qi Li (Iowa State University)
TransformerLarge Language ModelText
🎯 What it does: This paper proposes a relationship extraction framework that combines distant supervision and natural language inference (DSRE-NLI). It automatically mines and filters templates through semi-automatic relation representation (SARV), significantly improving data quality and reducing labor costs.
Improving Dynamic HDR Imaging with Fusion Transformer
Rufeng Chen (Hangzhou Dianzi University), Shanxin Yuan (Queen Mary University of London)
RestorationTransformerImage
🎯 What it does: This paper proposes a Transformer-based HDR fusion framework called HFT, which achieves end-to-end reconstruction from multiple exposed LDR images to HDR images.
Improving End-to-End Speech Translation by Leveraging Auxiliary Speech and Text Data
Yuhao Zhang (Northeastern University), Jingbo Zhu (Northeastern University)
TransformerContrastive LearningTextAudio
🎯 What it does: A multi-step pre-training framework (MSP-ST) is proposed, which incorporates a text encoder into an end-to-end speech translation model. By aligning with a text adapter, employing contrastive learning, and utilizing denoising techniques, the ST encoder gains both audio and text encoding capabilities, further enhancing translation quality using unlabeled speech and text data.