AAAI 2024 Papers — Page 2
AAAI Conference on Artificial Intelligence · 2331 papers
Adversarial Attacks on the Interpretation of Neuron Activation Maximization
Geraldin Nanfack (Concordia University), Eugene Belilovsky (Princeton University)
Explainability and InterpretabilityKnowledge DistillationAdversarial AttackConvolutional Neural NetworkTransformerSupervised Fine-TuningImage
🎯 What it does: This paper proposes a method for adversarial attacks on neuron activation maximization feature visualization, which can alter the interpretation results without significantly affecting model performance.
Adversarial Purification with the Manifold Hypothesis
Zhaoyuan Yang (General Electric Research), Peter Tu (Australian National University)
ClassificationAdversarial AttackAuto EncoderImage
🎯 What it does: A manifold hypothesis-based adversarial purification framework is proposed, achieving a robust model without adversarial training during testing through variational inference.
Adversarial Robust Safeguard for Evading Deep Facial Manipulation
Jiazhi Guan (Tsinghua University), Youjian Zhao (Tsinghua University)
Adversarial AttackConvolutional Neural NetworkGenerative Adversarial NetworkImage
🎯 What it does: A learning-based dual protection model (ARS) is proposed, which generates adversarial noise through a single forward inference to preemptively protect facial images from deepfake attacks and is robust against common post-processing perturbations (such as blurring, JPEG compression, cropping, and scaling).
Adversarial Socialbots Modeling Based on Structural Information Principles
Xianghua Zeng (Beihang University), Angsheng Li (Beihang University)
Adversarial AttackGraph Neural NetworkReinforcement LearningGraphTabular
🎯 What it does: A framework called SIASM based on the principle of structural information is proposed to actively model the attack behavior of social robots, maximizing network influence and enhancing concealment.
Adversarially Balanced Representation for Continuous Treatment Effect Estimation
Amirreza Kazemi (Simon Fraser University), Martin Ester (Simon Fraser University)
Representation LearningAdversarial AttackGenerative Adversarial NetworkTextBiomedical Data
🎯 What it does: A framework for adversarial balanced representation learning (ACFR) is proposed for estimating continuous treatment effects. By minimizing the KL divergence of the representation layer and using a cross-attention network to predict outcomes, it can achieve both covariate balance and make full use of treatment value information.
AE-NeRF: Audio Enhanced Neural Radiance Field for Few Shot Talking Head Synthesis
Dongze Li (University of Chinese Academy of Sciences), Tieniu Tan (Chinese Academy of Sciences)
GenerationData SynthesisTransformerNeural Radiance FieldVideoAudio
🎯 What it does: This paper proposes AE-NeRF, which utilizes an audio-aware aggregation module and dual NeRF to separate audio-related and non-related areas, achieving high-quality audio-driven character head synthesis with a small amount of training video.
AesFA: An Aesthetic Feature-Aware Arbitrary Neural Style Transfer
Joonwoo Kwon (Seoul National University), Jiook Cha (Seoul National University)
Image TranslationComputational EfficiencyConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: A lightweight AesFA model is proposed to achieve aesthetic feature-aware arbitrary neural style transfer.
Agile Multi-Source-Free Domain Adaptation
Xinyao Li (University of Electronic Science and Technology of China), Ke Lu (University of Electronic Science and Technology of China)
Domain AdaptationImage
🎯 What it does: A lightweight Bi-ATEN module is proposed to achieve multi-source unsupervised domain adaptation, enabling adaptation to the target domain using a small number of trainable parameters without extensive fine-tuning of the source model.
Agile-Quant: Activation-Guided Quantization for Faster Inference of LLMs on the Edge
Xuan Shen (Northeastern University), Yanzhi Wang (Northeastern University)
OptimizationComputational EfficiencyTransformerLarge Language ModelText
🎯 What it does: A quantization framework called Agile-Quant based on activation guidance is proposed, which can simultaneously quantize the weights and activations of LLMs on edge devices, and optimize the quantization effect through activation-aware token pruning.
AGS: Affordable and Generalizable Substitute Training for Transferable Adversarial Attack
Ruikui Wang (Beihang University), Yunhong Wang (Beihang University)
Object DetectionSegmentationAdversarial AttackConvolutional Neural NetworkTransformerContrastive LearningImage
🎯 What it does: Developed a self-supervised alternative model training framework (AGS) that trains the alternative model using only unlabeled data, thereby achieving transferable black-box adversarial attacks.
Ahpatron: A New Budgeted Online Kernel Learning Machine with Tighter Mistake Bound
Yun Liao (Tianjin University), Jianwu Dang (Tianjin University)
ClassificationOptimizationTabular
🎯 What it does: A new budget online kernel learning model, Ahpatron, is proposed, which significantly improves the error bounds of previous work and addresses an open problem related to the upper bound of hypothesis space constraints.
Aleth-NeRF: Illumination Adaptive NeRF with Concealing Field Assumption
Ziteng Cui (Shanghai AI Laboratory), Tatsuya Harada (University of Oxford)
RestorationGenerationNeural Radiance FieldImage
🎯 What it does: Train NeRF for unsupervised learning under low light and overexposure conditions, and generate new views with normal lighting by removing/adding the Concealing Field during inference.
Aligner²: Enhancing Joint Multiple Intent Detection and Slot Filling via Adjustive and Forced Cross-Task Alignment
Zhihong Zhu (Peking University), Yuexian Zou (Peking University)
Recurrent Neural NetworkReinforcement LearningText
🎯 What it does: The Aligner 2 framework is proposed, which includes the Adjustive Cross-Task Aligner (ACA) before task interaction and the Forced Cross-Task Aligner (FCA) that enforces alignment through reinforcement learning after interaction, used for range and prediction alignment in multi-intent spoken language understanding (SLU).
Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination
Qingwang Zhang (Shenzhen University), Yingying Zhu (Shenzhen University)
RetrievalOptimizationConvolutional Neural NetworkTransformerImage
🎯 What it does: A framework (FRGeo) is proposed for aligning geometric spatial layouts in cross-view geographic localization through feature recombination.
ALISON: Fast and Effective Stylometric Authorship Obfuscation
Eric Xing (Washington University in St. Louis), Dongwon Lee (Pennsylvania State University)
Adversarial AttackTransformerLarge Language ModelText
🎯 What it does: A method for author identity obfuscation based on grammatical features, ALISON, is proposed, which utilizes a lightweight internal author discriminator and a masked language model to perform semantic-preserving style replacement of multiple word segments at once.
All Beings Are Equal in Open Set Recognition
Chaohua Li (Nanjing University of Aeronautics and Astronautics), Songcan Chen (Nanjing University of Aeronautics and Astronautics)
ClassificationRecognitionContrastive LearningImage
🎯 What it does: The K+K scheme is proposed, using Target-Aware Universum (TAU) to generate K classes of pseudo-unknown samples, and treating known and unknown categories equally through Dual Contrastive Loss, improving open-set recognition.
All Should Be Equal in the Eyes of LMs: Counterfactually Aware Fair Text Generation
Pragyan Banerjee (Indian Institute of Technology), Sumit Bhatia (MDSR Labs, Adobe)
GenerationTransformerLarge Language ModelText
🎯 What it does: A fair reasoning framework based on adversarial instances, CAFIE, is proposed, which adjusts the probability distribution of adversarial contexts for different groups during inference to generate fairer text.
Almost Envy-Free Allocations of Indivisible Goods or Chores with Entitlements
Max Springer (University of Maryland), Hadi Yami (Microsoft Corporation)
🎯 What it does: This paper studies how to fairly allocate indivisible goods or tasks to agents with different weights, proposing fair allocation algorithms and impossibility results under unequal rights.
AltDiffusion: A Multilingual Text-to-Image Diffusion Model
Fulong Ye (Beijing University of Posts and Telecommunications), Ledell Wu (Beijing Academy of Artificial Intelligence)
GenerationKnowledge DistillationDiffusion modelImageTextBenchmark
🎯 What it does: A multilingual text-to-image diffusion model called AltDiffusion has been proposed and trained, supporting 18 languages, directly processing multilingual prompts to generate high-quality images.
AltNeRF: Learning Robust Neural Radiance Field via Alternating Depth-Pose Optimization
Kun Wang (Nanjing University of Science and Technology), Jian Yang (Nanjing University of Science and Technology)
GenerationPose EstimationDepth EstimationOptimizationNeural Radiance FieldVideo
🎯 What it does: Proposes the AltNeRF framework, which alternately optimizes self-supervised monocular depth estimation (SMDE) and NeRF to generate high-quality neural radiance fields from monocular videos without pose annotations;
Altruism in Facility Location Problems
Houyu Zhou (City University of Hong Kong), Minming Li (University of Nebraska Lincoln)
Optimization
🎯 What it does: In this paper, the authors study the scenario of altruistic agents in facility location problems (FLPs) where each agent may belong to multiple overlapping groups. They propose a new mechanism design approach that ensures strategy independence in a multi-group altruistic environment and achieves approximately optimal results across various objectives (community cost, maximum cost, maximum total group cost, maximum average group cost).
Amalgamating Multi-Task Models with Heterogeneous Architectures
Jidapa Thadajarassiri (Srinakharinwirot University), Elke Rundensteiner (Worcester Polytechnic Institute)
Knowledge DistillationRepresentation LearningConvolutional Neural NetworkImage
🎯 What it does: The VENUS method is proposed to address the knowledge fusion problem under heterogeneous architectures of multi-task teacher models, training a unified multi-task student with unlabeled data.
AMD: Anatomical Motion Diffusion with Interpretable Motion Decomposition and Fusion
Beibei Jing (Huazhong University of Science and Technology), Wei Yang (Huazhong University of Science and Technology)
GenerationData SynthesisRetrievalLarge Language ModelSupervised Fine-TuningDiffusion modelTextMultimodality
🎯 What it does: The AMD model is proposed, which uses LLM to decompose complex texts into anatomical short sentences, and generates realistic actions through a dual-branch diffusion fusion.
AMD: Autoregressive Motion Diffusion
Bo Han (Zhejiang University), Chang Xu (University of Sydney)
GenerationData SynthesisPose EstimationDiffusion modelVideoTextMultimodality
🎯 What it does: Proposes an autoregressive diffusion model AMD for generating high-quality human motion sequences based on long text prompts;
Amodal Scene Analysis via Holistic Occlusion Relation Inference and Generative Mask Completion
Bowen Zhang (Australian Institute for Machine Learning, University of Adelaide), Yifan Liu (Adobe Research)
SegmentationGenerationTransformerDiffusion modelImage
🎯 What it does: This paper proposes a new framework for panoramic occlusion relationship reasoning and generative occlusion completion (HORI+GMC) for efficient inference of invisible shapes and determination of occlusion relationships.
AMSP-UOD: When Vortex Convolution and Stochastic Perturbation Meet Underwater Object Detection
Jingchun Zhou (Dalian Maritime University), Chongyi Li (Nankai University)
Object DetectionConvolutional Neural NetworkImage
🎯 What it does: A novel single-stage network for underwater object detection, AMSP-UOD, is proposed to specifically address noise and attenuation issues in underwater images.
An Approximate Skolem Function Counter
Arijit Shaw (Chennai Mathematical Institute), Kuldeep S. Meel (University of Toronto)
TabularBenchmark
🎯 What it does: A novel approximate Skolem function counting algorithm, SkolemFC, is studied, which can estimate the number of Skolem functions without enumerating them.
An Attentive Inductive Bias for Sequential Recommendation beyond the Self-Attention
Yehjin Shin (Yonsei University), Noseong Park (Yonsei University)
Recommendation SystemTransformerSequential
🎯 What it does: This paper proposes a Fourier Transform-based attention mechanism called BSARec, aimed at addressing the low-frequency filtering and over-smoothing issues caused by self-attention in Transformers for sequential recommendation.
An Autoregressive Text-to-Graph Framework for Joint Entity and Relation Extraction
Urchade Zaratiana (FI Group), Thierry Charnois (LIPN - Université Sorbonne Paris Nord - CNRS UMR 7030)
GenerationTransformerText
🎯 What it does: A self-regressive text-to-graph (ATG) framework is proposed, which jointly extracts entities and relationships from unstructured text and achieves this through generating linearized graph structures.
An Eager Satisfiability Modulo Theories Solver for Algebraic Datatypes
Amar Shah (University of California), Sanjit A. Seshia
Benchmark
🎯 What it does: This paper proposes a method for transforming quantifier-free ADT queries into an eager solving approach that only involves UF, and implements a prototype solver called Algaroba.
An Effective Augmented Lagrangian Method for Fine-Grained Multi-View Optimization
Yuze Tan (Sichuan University), Jiancheng Lv (Sichuan University)
OptimizationSupervised Fine-TuningGraph
🎯 What it does: A fine-grained multi-view clustering framework called ALMOND is proposed based on the augmented Lagrangian method, which constructs a consistent graph by assigning weights to each view at the sample level, thereby better preserving the consistency of local structures across views.
An Effective Polynomial Technique for Compiling Conditional Effects Away
Alfonso Emilio Gerevini (Università degli Studi di Brescia), Enrico Scala (Università degli Studi di Brescia)
OptimizationComputational EfficiencyTabularBenchmark
🎯 What it does: A new polynomial compilation technique is proposed, which transforms classical planning problems with conditional effects into simple planning problems with unconditional effects.
An Efficient Knowledge Transfer Strategy for Spiking Neural Networks from Static to Event Domain
Xiang He (Chinese Academy of Sciences), Yi Zeng (Chinese Academy of Sciences)
ClassificationDomain AdaptationSpiking Neural NetworkContrastive LearningImage
🎯 What it does: An efficient knowledge transfer strategy from static images to event domains is proposed to help spiking neural networks (SNNs) improve generalization ability on small sample event data.
An Efficient Subgraph-Inferring Framework for Large-Scale Heterogeneous Graphs
Wei Zhou (Huazhong University of Science and Technology), Hai Jin (Huazhong University of Science and Technology)
Computational EfficiencyGraph Neural NetworkGraph
🎯 What it does: The SubInfer framework is proposed, which achieves efficient learning and inference of large-scale heterogeneous graphs through meta-path subgraph partitioning, global vertex completion, and distributed subgraph training/inference.
An Embedding-Unleashing Video Polyp Segmentation Framework via Region Linking and Scale Alignment
Zhixue Fang (Shenzhen University), Jing Qin (Hong Kong Polytechnic University)
SegmentationTransformerVideo
🎯 What it does: A framework for embedded release is proposed, modeling video segmentation as appearance-level semantic embedding. It utilizes a Proposal Generation Network (PGN) to generate mask proposals and generates background and dynamic semantics through Cross-Scale Region Linking (CRL) and Center-Aware Scale Alignment (CSA) modules in the Appearance Embedding Network (AEN), followed by segmentation through non-parametric semantic interaction.
An Empirical Study of CLIP for Text-Based Person Search
Min Cao (Soochow University), Min Zhang (Harbin Institute of Technology)
RetrievalCompressionTransformerVision Language ModelContrastive LearningImageText
🎯 What it does: This paper conducts systematic experiments on CLIP to explore its potential in the text-driven face search (TBPS) task, proposing a baseline TBPS-CLIP that combines data augmentation, loss functions, and training techniques without using complex modules.
An Exercise in Tournament Design: When Some Matches Must Be Scheduled
Sushmita Gupta (Institute of Mathematical Sciences, HBNI), Peter Strulo (University of Warwick)
OptimizationGraph
🎯 What it does: This paper studies the algorithmic problem of scheduling specified matches in single-elimination tournaments, proposing the Demand-TF problem and providing its formal definition and solution approach.
An Implicit Trust Region Approach to Behavior Regularized Offline Reinforcement Learning
Zhe Zhang (Nanjing University of Aeronautics and Astronautics), Xiaoyang Tan (Nanjing University of Aeronautics and Astronautics)
Reinforcement LearningTabular
🎯 What it does: This paper proposes an implicit trust region method (iTRPO) achieved through reward reshaping in offline reinforcement learning;
An Information-Flow Perspective on Algorithmic Fairness
Samuel Teuber (Karlsruhe Institute of Technology), Bernhard Beckert (Karlsruhe Institute of Technology)
Safty and PrivacyExplainability and InterpretabilityTabular
🎯 What it does: This paper maps the problem of algorithmic fairness into the framework of information flow security, using information flow analysis tools to verify the fairness of program code, and proposes a new quantitative fairness metric—Fairness Spread.
An Interpretable Approach to the Solutions of High-Dimensional Partial Differential Equations
Lulu Cao (Xiamen University), Min Jiang (Xiamen University)
OptimizationExplainability and InterpretabilityComputational EfficiencyTabularPhysics Related
🎯 What it does: A method called HD-TLGP is proposed, which is based on genetic programming symbolic regression, structural transfer, automatic differentiation, and pruning operators, to solve the analytical solutions of high-dimensional partial differential equations.
An Optimal Transport View for Subspace Clustering and Spectral Clustering
Yuguang Yan (Guangdong University of Technology), Michael Kwok-Po Ng (Hong Kong Baptist University)
Anomaly DetectionOptimizationTabular
🎯 What it does: This paper proposes a subspace clustering and spectral clustering explanation based on self-consistent optimal transport, and introduces the spectral optimal transport barycenter model (SOTA), achieving joint learning of the affinity matrix and embedding;
Analytically Tractable Models for Decision Making under Present Bias
Yasunori Akagi (NTT Corporation), Takeshi Kurashima (NTT Corporation)
OptimizationAgentic AI
🎯 What it does: A model is proposed to analyze the behavior of agents with present bias, aimed at understanding human time-inconsistent behavior in the pursuit of long-term goals.
Analyzing Generalization in Policy Networks: A Case Study with the Double-Integrator System
Ruining Zhang (University of Electronic Science and Technology of China), Jian Cheng (Air Force Engineering University)
Reinforcement LearningSequential
🎯 What it does: An analysis and experimental validation of the generalization failure of deep reinforcement learning policy networks for double integrator systems when expanding the state space.
Anchoring Path for Inductive Relation Prediction in Knowledge Graphs
Zhixiang Su (Nanyang Technological University), Lizhen Cui (Shandong University)
TransformerGraphRetrieval-Augmented Generation
🎯 What it does: A framework named Anchoring Path Sentence Transformer (APST) is proposed for inductive relation prediction in knowledge graphs, utilizing Anchoring Paths (AP) and Closed Paths (CP) along with externally retrieved detailed descriptions of entities for reasoning.
ANEDL: Adaptive Negative Evidential Deep Learning for Open-Set Semi-supervised Learning
Yang Yu (Chinese University of Hong Kong), Pheng Ann Heng (National University of Singapore)
Anomaly DetectionRepresentation LearningContrastive LearningImage
🎯 What it does: The ANEDL framework is proposed, utilizing Evidence Deep Learning (EDL) to achieve adaptive anomaly detection and self-training in open set semi-supervised learning.
Angle Robustness Unmanned Aerial Vehicle Navigation in GNSS-Denied Scenarios
Yuxin Wang (Zhejiang University), Mingli Song (Zhejiang University)
Autonomous DrivingTransformerSimultaneous Localization and MappingImage
🎯 What it does: A robust angle navigation model for drone point-to-point navigation is proposed, which directly predicts the direction angle to address flight deviations in GNSS-denied environments.
AnomalyDiffusion: Few-Shot Anomaly Image Generation with Diffusion Model
Teng Hu (Shanghai Jiao Tong University), Chengjie Wang
GenerationAnomaly DetectionDiffusion modelImage
🎯 What it does: AnomalyDiffusion is proposed, a few-shot anomaly image generation method based on diffusion models, capable of synthesizing realistic and diverse anomaly-image-mask pairs on normal images according to a given anomaly mask.
AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models
Zhaopeng Gu (Chinese Academy of Sciences), Jinqiao Wang (Chinese Academy of Sciences)
Anomaly DetectionTransformerPrompt EngineeringVision Language ModelImageMultimodality
🎯 What it does: Utilizing the large visual language model AnomalyGPT to achieve industrial anomaly detection and localization without the need for manual threshold setting.
Another Way to the Top: Exploit Contextual Clustering in Learned Image Coding
Yichi Zhang (Hangzhou Normal University), Zhan Ma (Nanjing University)
CompressionImage
🎯 What it does: A learning-based image compression framework called CLIC is proposed, which utilizes clustering and local attention to achieve global aggregation and local refinement of image features, and introduces GuidedPQF for quantization error compensation.
Any-Size-Diffusion: Toward Efficient Text-Driven Synthesis for Any-Size HD Images
Qingping Zheng (Northwestern Polytechnical University), Hang Xu (Huawei Noah's Ark Lab)
GenerationData SynthesisSuper ResolutionDiffusion modelImageText
🎯 What it does: A two-stage text-to-image synthesis pipeline named Any-Size-Diffusion (ASD) is proposed. Stage-I (ARAD) generates well-synthesized low-resolution images within a limited aspect ratio range through multi-bit ratio training; Stage-II (FSTD) quickly enlarges the images generated in Stage-I to high-definition images of any size using implicit overlapping tile sampling, avoiding seams and significantly reducing memory usage.
Any-Stereo: Arbitrary Scale Disparity Estimation for Iterative Stereo Matching
Zhaohuai Liang (China University of Geosciences), Changhe Li (Anhui University of Science and Technology)
Depth EstimationAutonomous DrivingConvolutional Neural NetworkRecurrent Neural NetworkSupervised Fine-TuningImage
🎯 What it does: The AnyStereo module is proposed, which can accurately upsample disparity maps at arbitrary scales in the iterative stereo matching process, addressing the shortcomings of traditional convolutional upsampling in recovering high-frequency details.
Any-Way Meta Learning
JunHoo Lee, Nojun Kwak (Seoul National University)
Meta LearningImage
🎯 What it does: A label-equivalence-based arbitrary category meta-learning framework is proposed, and a semantic classifier and Mixup technique are introduced to enhance generalization ability.
Approval-Based Committee Voting in Practice: A Case Study of (over-)Representation in the Polkadot Blockchain
Niclas Boehmer (Harvard University), Ulrike Schmidt-Kraepelin (TU Eindhoven)
Tabular
🎯 What it does: This paper collects and analyzes approximately 500 large-scale ABC voting instances in the Polkadot blockchain, evaluating the performance of various voting rules in practice;
Approximate Distance Oracle for Fault-Tolerant Geometric Spanners
Kyungjin Cho (Pohang University of Science and Technology), Eunjin Oh (Pohang University of Science and Technology)
OptimizationGraph
🎯 What it does: This paper proposes an approximate distance and shortest path oracle for fault-tolerant geometric support structures, aimed at solving routing problems in real-world road networks.
Approximate Integer Solution Counts over Linear Arithmetic Constraints
Cunjing Ge (Nanjing University)
Optimization
🎯 What it does: The study investigates algorithms for approximately counting integer solutions (i.e., lattice point counting) in convex polyhedra.
Approximating the Shapley Value without Marginal Contributions
Patrick Kolpaczki (Paderborn University), Eyke Hüllermeier
Explainability and InterpretabilityComputational EfficiencyTabular
🎯 What it does: Two approximate algorithms for Shapley values without marginal contributions are proposed (SVARM and Stratified SVARM), achieving efficient and unbiased estimation of Shapley values for all players in cooperative games.
Approximation Algorithms for Preference Aggregation Using CP-Nets
Abu Mohammad Hammad Ali (University of Regina), Sandra Zilles (University of Regina)
Optimization
🎯 What it does: This paper studies the design and analysis of approximate algorithms for the swap problem when using Conditional Preference Networks (CP-nets) for preference aggregation in combinatorial domains. It proposes an approximation ratio of up to 4/3 under specific symmetry conditions and provides a polynomial-time algorithm (Algorithm 1) for constructing the optimal Conditional Preference Table (CPT) on all input parent sets, achieving optimal solutions in most instances.
Approximation Scheme for Weighted Metric Clustering via Sherali-Adams
Dmitrii Avdiukhin (Northwestern University), Grigory Yaroslavtsev (George Mason University)
OptimizationText
🎯 What it does: This paper proposes a Polynomial Time Approximation Scheme (PTAS) for solving the weighted metric clustering problem with a given metric space and arbitrary weight matrix;
AQ-DETR: Low-Bit Quantized Detection Transformer with Auxiliary Queries
Runqi Wang (Beihang University), Baochang Zhang
Object DetectionKnowledge DistillationTransformerImage
🎯 What it does: This paper proposes AQ-DETR, which achieves full 4-bit low-precision quantization of DETR through auxiliary queries and hierarchical distillation.
Arbitrary-Scale Point Cloud Upsampling by Voxel-Based Network with Latent Geometric-Consistent Learning
Hang Du (Hikvision Research Institute), Shiliang Pu (Hikvision Research Institute)
GenerationData SynthesisConvolutional Neural NetworkPoint Cloud
🎯 What it does: A voxel grid-based framework for arbitrary scale point cloud upsampling, PU-VoxelNet, is proposed, which maps sparse point clouds to a predefined grid space and generates high-quality point clouds through density distribution.
Arbitrary-Scale Video Super-resolution Guided by Dynamic Context
Cong Huang (University of Science and Technology of China), Yan Lu (Microsoft Research Asia)
RestorationSuper ResolutionVideo
🎯 What it does: A dynamic context-guided upsampling (DCGU) module is proposed for arbitrary scale video super-resolution, which can efficiently reconstruct high-resolution videos without being constrained by pixel rearrangement.
Are You Concerned about Limited Function Evaluations: Data-Augmented Pareto Set Learning for Expensive Multi-Objective Optimization
Yongfan Lu (East China Normal University), Aimin Zhou (East China Normal University)
OptimizationGenerative Adversarial Network
🎯 What it does: This paper proposes a Data Augmentation Pareto Set Learning (DA-PSL) method to address expensive multi-objective optimization (EMOP) problems, achieving a more accurate approximation of the Pareto front under a limited function evaluation budget.
Arithmetic Feature Interaction Is Necessary for Deep Tabular Learning
Yi Cheng (Zhejiang University), Wei Lin (Alibaba Group)
TransformerPrompt EngineeringTabular
🎯 What it does: This paper proposes and validates a Transformer-based deep tabular learning model called AMFormer, which enhances the modeling capability of tabular data through explicit additive and multiplicative feature interactions.
ArtBank: Artistic Style Transfer with Pre-trained Diffusion Model and Implicit Style Prompt Bank
Zhanjie Zhang (Zhejiang University), Huaizhong Lin (Zhejiang University)
Image TranslationGenerationDiffusion modelImage
🎯 What it does: Proposes the ArtBank framework, which achieves artistic style transfer by combining pre-trained large models with an implicit style prompt bank.
Aspect-Based Sentiment Analysis with Explicit Sentiment Augmentations
Jihong Ouyang (Jilin University), Ximing Li (Jilin University)
ClassificationGenerationTransformerLarge Language ModelSupervised Fine-TuningText
🎯 What it does: The ABSA-ESA method is proposed to address the problem of implicit sentiment recognition by generating explicit sentiment incremental sentences.
ASWT-SGNN: Adaptive Spectral Wavelet Transform-Based Self-Supervised Graph Neural Network
Ruyue Liu (Institute of Information Engineering Chinese Academy of Sciences), Weiping Wang (Renmin University of China)
ClassificationRepresentation LearningGraph Neural NetworkContrastive LearningGraph
🎯 What it does: This paper proposes a self-supervised graph neural network ASWT-SGNN, which implements a learnable multi-scale filter through adaptive spectral wavelet transformation, thereby balancing local and global information in node representation learning.
Asymmetric Mutual Alignment for Unsupervised Zero-Shot Sketch-Based Image Retrieval
Zhihui Yin (Xidian University), Cheng Deng (Xidian University)
RetrievalContrastive LearningImage
🎯 What it does: An adaptive asynchronous mutual alignment (AMA) framework for unsupervised zero-shot sketch-based image retrieval (UZS-SBIR) is proposed, achieving cross-modal retrieval of sketches and images under the conditions of no labels and non-overlapping categories.
AT4CTR: Auxiliary Match Tasks for Enhancing Click-Through Rate Prediction
Qi Liu (University of Science and Technology of China), Jun Lei (Meituan)
Recommendation SystemContrastive LearningTabular
🎯 What it does: By incorporating two auxiliary matching tasks (User-Item Matching UIM and Next Item Prediction NIP) into the CTR model, the accuracy of click-through rate prediction is improved, and the issue of data sparsity is alleviated.
Attack Deterministic Conditional Image Generative Models for Diverse and Controllable Generation
Tianyi Chu (Zhejiang University), Huaizhong Lin (Zhejiang University)
GenerationAdversarial AttackTransformerGenerative Adversarial NetworkContrastive LearningImage
🎯 What it does: Proposes a small perturbation attack on a pre-trained deterministic conditional generative model to achieve diversified and controllable generation.
Attacking Transformers with Feature Diversity Adversarial Perturbation
Chenxing Gao (Tencent), Wei Yang (Tencent)
Object DetectionSegmentationPose EstimationDepth EstimationAdversarial AttackConvolutional Neural NetworkTransformerImage
🎯 What it does: This paper proposes a label-free white-box attack method called FDAP, which accelerates the 'feature collapse' of intermediate layer features in Vision Transformers (ViT), suppressing high-frequency information to induce misclassification by the model.
Attacks on Continual Semantic Segmentation by Perturbing Incremental Samples
Zhidong Yu (University of Science and Technology of China), Zhenbo Shi (University of Science and Technology of China)
SegmentationAdversarial AttackConvolutional Neural NetworkGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes an attack method for newly added samples in Continuous Semantic Segmentation (CSS), constructing a task to attack new samples and implementing an offline Class Shift Attack (CS-Attack).
Attention Disturbance and Dual-Path Constraint Network for Occluded Person Re-identification
Jiaer Xia (Xiamen University), Liujuan Cao (Donghua University)
RecognitionRetrievalTransformerGenerative Adversarial NetworkImage
🎯 What it does: By introducing the Attention Disturbance Mask (ADM) and Dual-Path Constraint (DPC) module, the robustness and recognition performance of occluded pedestrian Re-ID are enhanced based on the Transformer (ViT).
Attention Guided CAM: Visual Explanations of Vision Transformer Guided by Self-Attention
Saebom Leem (Sogang University), Hyunseok Seo (Korea Institute of Science and Technology)
Object DetectionExplainability and InterpretabilityTransformerImage
🎯 What it does: A gradient analysis method based on self-attention is proposed for visual explanation and weakly supervised localization of Vision Transformers.
Attention-Induced Embedding Imputation for Incomplete Multi-View Partial Multi-Label Classification
Chengliang Liu (Harbin Institute of Technology), Yong Xu (Sun Yat-sen University)
ClassificationGraph Neural NetworkMultimodality
🎯 What it does: An Attention-Induced Embedding Imputation Network (AIMNet) is proposed to address the multi-view multi-label classification problem with both missing views and missing labels.
Attribute-Missing Graph Clustering Network
Wenxuan Tu (National University of Defense Technology), Xinwang Liu (National University of Defense Technology)
Graph Neural NetworkGraph
🎯 What it does: A unified Attribute-Missing Graph Clustering Network (AMGC) is proposed, which enhances the clustering performance of graphs with missing attributes through alternating optimization of clustering and attribute inference.
AUC Optimization from Multiple Unlabeled Datasets
Zheng Xie (Nanjing University), Ming Li (Nanjing University)
OptimizationImage
🎯 What it does: This paper proposes an algorithm U_m-AUC for AUC optimization on multiple unlabeled datasets, which constructs a multi-label AUC optimization problem using the relative class priors of the unlabeled set and achieves training through efficient point optimization.
Audio Generation with Multiple Conditional Diffusion Model
Zhifang Guo (Institute of Computing Technology, Chinese Academy of Sciences), Xiangdong Wang (Institute of Computing Technology, Chinese Academy of Sciences)
GenerationData SynthesisDiffusion modelAudio
🎯 What it does: A multi-condition controlled text-to-audio generation model is proposed, which achieves fine-grained controllable generation of audio content and style by incorporating control conditions such as timestamps, pitch curves, and energy curves.
Audio Scanning Network: Bridging Time and Frequency Domains for Audio Classification
Liangwei Chen (University of Science and Technology of China), Huanhuan Chen (University of Science and Technology of China)
ClassificationAudio
🎯 What it does: This paper proposes the Audio Scanning Network (ASNet), an audio classification framework that utilizes temporal and frequency domain waveform fluctuation features and performs associative learning through Reservoir Kernel Canonical Correlation Analysis (RKCCA).
Auditable Algorithms for Approximate Model Counting
Kuldeep S. Meel (University of Toronto), S. Akshay (Indian Institute of Technology Bombay)
🎯 What it does: A class of auditable approximate model counting algorithms is proposed, and the trade-off between audit complexity and counting algorithm complexity is analyzed.
Augmented Commonsense Knowledge for Remote Object Grounding
Bahram Mohammadi (Australian Institute for Machine Learning), Javen Qinfeng Shi (Australian Institute for Machine Learning)
RecognitionObject DetectionTransformerVision Language ModelMultimodality
🎯 What it does: This paper proposes a model called ACK based on augmented common sense knowledge to enhance visual-language navigation in the REVERIE task.
Auto-Prox: Training-Free Vision Transformer Architecture Search via Automatic Proxy Discovery
Zimian Wei (National University of Defense Technology), Dongsheng Li (National University of Defense Technology)
Neural Architecture SearchTransformerImageBenchmark
🎯 What it does: This paper proposes Auto-Prox, a framework for automatically discovering zero-cost proxies for training-free Vision Transformer (ViT) architecture search.
Automated Defect Report Generation for Enhanced Industrial Quality Control
Jiayuan Xie (Hong Kong Polytechnic University), Qing Li (Hong Kong Polytechnic University)
GenerationAnomaly DetectionRecurrent Neural NetworkLarge Language ModelPrompt EngineeringVision Language ModelImageTextMultimodality
🎯 What it does: Proposed and implemented the Industrial Defect Detection Report Generation (DDRG) task, constructed a defect report dataset for 16 types of materials, and designed the Knowledge-Aware Report Generation (KRG) model;
Automated Design of Affine Maximizer Mechanisms in Dynamic Settings
Michael Curry (University of Zurich), Sven Seuken (ETH Zurich)
OptimizationReinforcement Learning from Human FeedbackTabular
🎯 What it does: This paper proposes an automated method for dynamic mechanism design by modeling the problem as a bi-level optimization that includes game theory and Markov Decision Processes (MDP). It utilizes the Affine Maximization Mechanism (AMA) to ensure strategy-proofness and individual rationality, thereby seeking optimal mechanism parameters under given objectives (such as revenue or task completion time).
Automatic Core-Guided Reformulation via Constraint Explanation and Condition Learning
Kevin Leo (Monash University), Mark Wallace (Monash University)
OptimizationExplainability and InterpretabilityComputational EfficiencyTabular
🎯 What it does: This paper proposes an automated method to reconstruct constraint programming models using core explanations and conditional learning to enhance solving performance.
Automatic Radiology Reports Generation via Memory Alignment Network
Hongyu Shen (Beijing Institute of Technology), Zhaoxing Tian (Beijing Jishuitan Hospital)
GenerationConvolutional Neural NetworkTransformerVision Language ModelImageTextMultimodalityElectronic Health Records
🎯 What it does: A memory alignment-based module is proposed, which aligns image and text features through a memory matrix to improve the quality of medical image report generation.
Autoregressive Omni-Aware Outpainting for Open-Vocabulary 360-Degree Image Generation
Zhuqiang Lu (University of Sydney), Zhiyong Wang (University of Sydney)
GenerationData SynthesisTransformerDiffusion modelImageText
🎯 What it does: AOGNet is proposed, a self-regressive, global-local conditional 360-degree image generation network that can gradually complete the filling of panoramic images by simultaneously utilizing NFoV images and open vocabulary text prompts.
AvatarVerse: High-Quality & Stable 3D Avatar Creation from Text and Pose
Huichao Zhang (ByteDance), Min Zheng
GenerationPose EstimationDiffusion modelNeural Radiance FieldImage
🎯 What it does: AvatarVerse is constructed, a complete pipeline for generating high-quality 3D avatars from text descriptions and pose guidance in a zero-shot manner;
AVSegFormer: Audio-Visual Segmentation with Transformer
Shengyi Gao (Nanjing University), Tong Lu (Nanjing University)
SegmentationTransformerVideoMultimodalityAudio
🎯 What it does: A Transformer-based framework called AVSegFormer is proposed for audio-visual segmentation tasks (single sound source, multiple sound sources, and semantic segmentation), achieving pixel-level localization and segmentation of sound source targets.
Axiomatic Aggregations of Abductive Explanations
Gagan Biradar (University of Massachusetts), Yair Zick (University of Massachusetts)
Explainability and InterpretabilityAdversarial AttackTabularFinance Related
🎯 What it does: Design and evaluate three axiomatically-based feature importance metrics for aggregating attributable explanations (AXp) to generate more robust and interpretable model explanations.
B-spine: Learning B-spline Curve Representation for Robust and Interpretable Spinal Curvature Estimation
Hao Wang (Jilin University), Rui Ma (Jilin University)
SegmentationRepresentation LearningConvolutional Neural NetworkTransformerGenerative Adversarial NetworkImage
🎯 What it does: A multi-stage deep learning pipeline is proposed, which first performs spinal segmentation on low-quality X-ray images and utilizes CycleGAN for mask refinement. Then, based on the segmentation mask, it predicts the B-spline curve of the spinal centerline, and finally combines curve slope analysis with a regression model to obtain Cobb angle estimation.
Backdoor Adjustment via Group Adaptation for Debiased Coupon Recommendations
Junpeng Fang (Ant Group), Jun Zhou (Ant Group)
Recommendation SystemTabular
🎯 What it does: This paper proposes the BAGA framework, which implements a coupon recommendation model that removes matching bias by grouping users and coupons and using backdoor adjustment.
Backdoor Attacks via Machine Unlearning
Zihao Liu (Iowa State University), Chenglin Miao (Iowa State University)
OptimizationKnowledge DistillationAdversarial AttackData-Centric LearningConvolutional Neural NetworkImage
🎯 What it does: This paper proposes a method for backdoor attacks through machine unlearning, utilizing withdrawal requests submitted by the attacker to inject triggers into the model, thereby altering the model's output during the inference phase.
Backpropagation Through Agents
Zhiyuan Li (University of Electronic Science and Technology of China), Joni Pajarinen (Aalto University)
Reinforcement LearningAgentic AI
🎯 What it does: This paper proposes the Back-Propagation Through Agents (BPTA) framework, which combines autoregressive joint policies to form the BPPO algorithm for collaborative decision-making in multi-agent reinforcement learning.
Backward Responsibility in Transition Systems Using General Power Indices
Christel Baier (TU Dresden), Jakob Piribauer (TU Dresden)
🎯 What it does: A backward responsibility measure based on general weight indices in cooperative games (such as Shapley value and Banzhaf value) is proposed, distinguishing between optimistic and pessimistic definitions of responsibility, and implementing both exact calculation and random approximation algorithms.
BaCon: Boosting Imbalanced Semi-supervised Learning via Balanced Feature-Level Contrastive Learning
Qianhan Feng (Peking University), Tong Lin (Peking University)
ClassificationContrastive LearningImage
🎯 What it does: A BaCon method is proposed for achieving balanced feature distribution in class-imbalanced semi-supervised learning through feature-level contrastive learning, directly comparing positive samples at the feature center and utilizing reliable negative sample selection and dynamic temperature adjustment.
BadRL: Sparse Targeted Backdoor Attack against Reinforcement Learning
Jing Cui (University of Chinese Academy of Sciences), Junge Zhang (Institute of Automation, Chinese Academy of Sciences)
Adversarial AttackReinforcement LearningSequential
🎯 What it does: A sparse and targeted backdoor injection attack called BadRL is designed to achieve target actions and weaken cumulative rewards in reinforcement learning through triggers.
BAIT: Benchmarking (Embedding) Architectures for Interactive Theorem-Proving
Sean Lamont (Australian National University), Paul Montague (Google DeepMind)
Graph Neural NetworkTransformerReinforcement LearningGraphBenchmark
🎯 What it does: Developed the BAIT framework to unify the experimental environment, data, and models of AI-ITP, systematically comparing formula embedding architectures.
Balancing Humans and Machines: A Study on Integration Scale and Its Impact on Collaborative Performance
Rui Zou (Central China Normal University), Jianwen Sun (Central China Normal University)
ClassificationOptimizationConvolutional Neural NetworkGaussian SplattingImage
🎯 What it does: A statistical framework is proposed to decompose the accuracy of human-machine ensemble into individual accuracy and diversity, and to study the relationship between ensemble scale and performance.
BAND: Biomedical Alert News Dataset
Zihao Fu (University of Cambridge), Nigel Collier (University of Cambridge)
TransformerLarge Language ModelSupervised Fine-TuningTextBiomedical DataBenchmark
🎯 What it does: A medical alert news dataset called BAND was constructed, collecting 1,508 case reports and generating 30 expert questions.
Barely Supervised Learning for Graph-Based Fraud Detection
Hang Yu (Shanghai University), Xiangfeng Luo (Shanghai University)
Anomaly DetectionGraph Neural NetworkContrastive LearningGraphFinance Related
🎯 What it does: This paper proposes a graph-based fraud detection framework that requires very few labeled samples, namely BSL.