AAAI 2023 Papers — Page 6
AAAI Conference on Artificial Intelligence · 1578 papers
Fair Short Paths in Vertex-Colored Graphs
Matthias Bentert (Technische Universitat Berlin), Rolf Niedermeier (Technische Universitat Berlin)
OptimizationGraph Neural NetworkGraph
🎯 What it does: The study focuses on finding the shortest or short paths in a multicolored graph, where the occurrence of each color falls within given upper and lower bounds;
Fair-CDA: Continuous and Directional Augmentation for Group Fairness
Rui Sun (Chinese University of Hong Kong), Zhenguo Li (Huawei)
Recommendation SystemGraph Neural NetworkImageTabular
🎯 What it does: This paper proposes Fair-CDA, a strategy for continuous directional data augmentation at the feature level, aimed at enhancing group fairness while maintaining model accuracy.
FairFed: Enabling Group Fairness in Federated Learning
Yahya H. Ezzeldin (University of Southern California), A. Salman Avestimehr
Federated LearningTabular
🎯 What it does: In the framework of federated learning, the FairFed algorithm is proposed, which dynamically adjusts aggregation weights on the server side based on local and global fairness metrics to enhance group fairness.
Fairness and Explainability: Bridging the Gap towards Fair Model Explanations
Yuying Zhao (Vanderbilt University), Tyler Derr (Vanderbilt University)
OptimizationExplainability and InterpretabilityGraph Neural NetworkGraphTabular
🎯 What it does: This paper proposes a program-oriented fairness metric based on explanation quality and designs a Comprehensive Fairness Algorithm (CFA) that takes into account prediction accuracy, traditional fairness, and explanation fairness.
Fairness and Welfare Quantification for Regret in Multi-Armed Bandits
Siddharth Barman (Indian Institute of Science), Ayush Sawarni (University of Washington)
Optimization
🎯 What it does: In this work, the authors propose the concept of Nash regret for multi-armed bandits (MAB) and design a Nash Confidence Bound algorithm based on improved upper confidence bounds, which can theoretically achieve near-optimal fairness and utility.
Fairness Concepts for Indivisible Items with Externalities
Haris Aziz (University of New South Wales), Toby Walsh (CyberAgent)
Optimization
🎯 What it does: A fair allocation model for indivisible items considering externalities is proposed, along with the concepts of EF1, EFX, GFS, and their corresponding polynomial-time algorithms.
Fairness in Contextual Resource Allocation Systems: Metrics and Incompatibility Results
Nathanael Jo (University of Southern California), Phebe Vayanos (University of Southern California)
Tabular
🎯 What it does: This paper constructs a comprehensive framework to assess fairness at different stages of resource allocation systems (admissions, allocation, and outcomes), and provides theoretical proofs of the compatibility and incompatibility among various fairness metrics, with a particular focus on the housing allocation problem for the homeless in Los Angeles.
FanoutNet: A Neuralized PCB Fanout Automation Method Using Deep Reinforcement Learning
Haiyun Li (Wuhan University of Technology), Mingyu Liu (Huawei Device Co., Ltd.)
OptimizationConvolutional Neural NetworkTransformerReinforcement LearningGraph
🎯 What it does: This paper presents FanoutNet, an automated PCB fanout method based on deep reinforcement learning, which significantly improves the routability of PCBs by pre-allocating layers and routing resources.
Farsighted Probabilistic Sampling: A General Strategy for Boosting Local Search MaxSAT Solvers
Jiongzhi Zheng (Huazhong University of Science and Technology), Jianrong Zhou (Huazhong University of Science and Technology)
OptimizationBenchmark
🎯 What it does: A general Future Probability Sampling (FPS) strategy is proposed to replace the single-variable flip mechanism in MaxSAT local search, thereby improving the performance of (W)PMS solvers.
Fast and Accurate Binary Neural Networks Based on Depth-Width Reshaping
Ping Xue (Hefei University of Technology), Zhen Wei (Hefei University of Technology)
OptimizationNeural Architecture SearchConvolutional Neural NetworkImage
🎯 What it does: This paper proposes the Depth-Width Reshaping (DWR) method, which adjusts the depth and width of existing full-precision network backbones and combines pruning techniques to construct a backbone network more suitable for binary neural networks.
Fast and Interpretable Dynamics for Fisher Markets via Block-Coordinate Updates
Tianlong Nan (Columbia University), Christian Kroer (Columbia University)
OptimizationExplainability and InterpretabilityTabular
🎯 What it does: Research on large-scale Fisher market equilibrium computation, proposing a block coordinate-based stochastic first-order algorithm.
Fast Convergence in Learning Two-Layer Neural Networks with Separable Data
Hossein Taheri (University of California), Christos Thrampoulidis (University of British Columbia)
OptimizationTabular
🎯 What it does: This study investigates the convergence and generalization performance of using Normalized Gradient Descent (NGD) on separable data with two-layer neural networks.
Fast Converging Anytime Model Counting
Yong Lai (Jilin University), Roland H.C. Yap
Benchmark
🎯 What it does: A new approximate model counting algorithm called PartialKC is proposed, which can be terminated midway and converge to exact counts. It utilizes partial knowledge compilation (partial CCDD) to generate random partial CCDD for unbiased estimation.
Fast Counterfactual Inference for History-Based Reinforcement Learning
Haichuan Gao (Tsinghua University), Feng Chen (Tsinghua University)
Recurrent Neural NetworkReinforcement LearningSequential
🎯 What it does: Proposes Tree-based History Counterfactual Inference (T-HCI), which effectively compresses historical space through coarse and fine-grained causal inference in historical reinforcement learning, improving sample efficiency.
Fast Fluid Simulation via Dynamic Multi-Scale Gridding
Jinxian Liu (Shanghai Jiao Tong University), Xiaoyang Huang (Huawei Hisilicon)
Computational EfficiencyConvolutional Neural NetworkPoint CloudOrdinary Differential Equation
🎯 What it does: A dynamic multi-scale grid method and adaptive Runge-Kutta integration are proposed to achieve real-time high-fidelity simulations based on Lagrangian particle fluid.
Fast Offline Policy Optimization for Large Scale Recommendation
Otmane Sakhi (Criteo AI Lab), Alexandre Gilotte (Criteo AI Lab)
Recommendation SystemOptimizationReinforcement LearningTabular
🎯 What it does: In the offline context of the multi-armed bandit framework, self-normalized importance sampling and approximate maximum inner product search (MIPS) are utilized to achieve offline policy optimization for large-scale recommendation systems, significantly reducing dependence on the size of the product catalog.
Fast Online Hashing with Multi-Label Projection
Wenzhe Jia (Ocean University of China), Jie Gui (Southeast University)
RetrievalOptimizationComputational EfficiencyImage
🎯 What it does: A Fast Online Hashing (FOH) method is proposed, which constructs a query pool, employs a neighbor maintenance algorithm, and utilizes layer sampling. By leveraging a multi-label similarity matrix and label projection loss, it achieves online hashing updates that only modify a small number of database entries, significantly reducing query time.
Fast Regularized Discrete Optimal Transport with Group-Sparse Regularizers
Yasutoshi Ida (NTT Computer and Data Science Laboratories), Yasuhiro Fujiwara (NTT Communication Science Laboratories)
OptimizationComputational EfficiencyImage
🎯 What it does: A fast solving algorithm for the discrete optimal transport problem with group sparse regularization is proposed, significantly reducing the gradient computation cost by skipping zero gradients and pre-selecting non-zero gradient groups.
Fast Saturating Gate for Learning Long Time Scales with Recurrent Neural Networks
Kentaro Ohno (NTT), Yasutoshi Ida (NTT)
Recurrent Neural NetworkSequential
🎯 What it does: A new fast gate function is proposed, which achieves faster saturation by embedding the hyperbolic sine function into the sigmoid, thereby enhancing the ability of RNNs to learn extremely long time scales.
FastAMI – a Monte Carlo Approach to the Adjustment for Chance in Clustering Comparison Metrics
Kai Klede (Friedrich-Alexander Universitat Erlangen-Nrnberg), Björn Eskofier (Friedrich-Alexander Universitat Erlangen-Nrnberg)
OptimizationComputational EfficiencyBenchmark
🎯 What it does: This paper proposes FastAMI, an algorithm based on Monte Carlo sampling for the rapid approximation of Adjusted Mutual Information (AMI) and Standardized Mutual Information (SMI).
FASTDIAGP: An Algorithm for Parallelized Direct Diagnosis
Viet-Man Le (Graz University of Technology), Thi Ngoc Trang Tran (University of Sevilla)
OptimizationComputational EfficiencyTabular
🎯 What it does: This paper presents FASTDIAGP—a diagnostic method that parallelizes the traditional FASTDIAG direct diagnosis algorithm, significantly improving execution speed through speculative programming and early parallel computation of consistency checks.
Faster Adaptive Federated Learning
Xidong Wu (University of Pittsburgh), Heng Huang (Nanjing University of Aeronautics and Astronautics)
OptimizationFederated LearningImageText
🎯 What it does: FAFED is proposed, an adaptive optimization algorithm that uses shared adaptive learning rates and momentum variance reduction in a heterogeneous data environment for cross-device federated learning, significantly accelerating the convergence of non-convex objectives.
Faster Fair Machine via Transferring Fairness Constraints to Virtual Samples
Zhou Zhai (Nanjing University of Information Science and Technology), Bin Gu (MBZUAI)
ClassificationOptimizationComputational EfficiencyTabular
🎯 What it does: This paper proposes a method to transform fairness constraints into 'virtual samples', allowing traditional unconstrained classifiers (such as SVM) to be efficiently trained under various fairness constraints (both linear and nonlinear covariance constraints).
Fault-Tolerant Offline Multi-Agent Path Planning
Keisuke Okumura (Tokyo Institute of Technology), Sébastien Tixeuil (Sorbonne University)
OptimizationRobotic IntelligenceReinforcement Learning
🎯 What it does: The research proposes the offline fault-tolerant path planning problem MAPPCF in multi-robot systems that may experience failures, and provides theoretical analysis and solution methods.
Feature Distribution Fitting with Direction-Driven Weighting for Few-Shot Images Classification
Xin Wei (Nanchang University), Weidong Min (Nanchang University)
ClassificationConvolutional Neural NetworkImage
🎯 What it does: A Direction-Driven Weighting Method (DDWM) is proposed to enhance few-shot image classification performance by fitting feature distributions.
Feature Normalization and Cartography-Based Demonstrations for Prompt-Based Fine-Tuning on Emotion-Related Tasks
Mahshid Hosseini (University of Illinois Chicago), Cornelia Caragea (University of Illinois Chicago)
ClassificationTransformerSupervised Fine-TuningPrompt EngineeringText
🎯 What it does: Under a small number of labeled samples, prompt-based fine-tuning is performed by combining feature matrix (mean and standard deviation) exchange (MoEx) and training dynamic-based example selection (CBDemo) to enhance the performance of sentiment and empathy classification tasks.
Feature-Level Debiased Natural Language Understanding
Yougang Lyu (Shandong University), Zhaochun Ren (Shandong University)
ClassificationContrastive LearningText
🎯 What it does: A debiasing method based on contrastive learning (DCT) is proposed, which reduces the model's dependence on biased features by dynamically sampling positive and negative biased samples.
FedABC: Targeting Fair Competition in Personalized Federated Learning
Dui Wang (Wuhan University), Dacheng Tao (Nanyang Technological University)
Federated LearningImage
🎯 What it does: The FedABC framework is proposed, which constructs binary classification tasks for each category at each client and uses binary classification loss to alleviate the issues of class imbalance and missing positive samples under non-IID conditions.
FedALA: Adaptive Local Aggregation for Personalized Federated Learning
Jianqing Zhang (Shanghai Jiao Tong University), Haibing Guan (Shanghai Jiao Tong University)
Federated LearningConvolutional Neural NetworkImageText
🎯 What it does: The FedALA method is proposed, which enhances local training performance in personalized federated learning by initializing the local model through an Adaptive Local Aggregation (ALA) module that integrates the global model and local model using element-level weighting.
Federated Generative Model on Multi-Source Heterogeneous Data in IoT
Zuobin Xiong (Georgia State University), Zhipeng Cai (Georgia State University)
GenerationFederated LearningGenerative Adversarial NetworkImage
🎯 What it does: A federated generative model framework designed for multi-source heterogeneous IoT data, supporting both feature-related and label-related scenarios, and providing synchronous and asynchronous update methods.
Federated Learning on Non-IID Graphs via Structural Knowledge Sharing
Yue Tan (Australian Artificial Intelligence Institute), Chengqi Zhang (Australian Artificial Intelligence Institute)
Federated LearningGraph Neural NetworkGraph
🎯 What it does: A federated graph learning framework named FedStar is proposed, specifically designed to enhance model performance on non-IID graph data by sharing structural knowledge.
Federated Robustness Propagation: Sharing Adversarial Robustness in Heterogeneous Federated Learning
Junyuan Hong (Michigan State University), Jiayu Zhou (University of Texas at Austin)
Federated LearningAdversarial AttackImage
🎯 What it does: This study investigates how to achieve the propagation of adversarial robustness in a heterogeneous federated learning environment through batch normalization techniques, proposing the FedRBN algorithm.
FedGS: Federated Graph-Based Sampling with Arbitrary Client Availability
Zheng Wang (Xiamen University), Cheng Wang (Xiamen University)
Federated LearningGraph Neural NetworkGraph
🎯 What it does: This paper proposes the FEDGS (Federated Graph-based Sampling) framework, which achieves stable global model updates and reduces long-term bias by constructing a data distribution-dependent graph (3DG) under arbitrary client availability.
FEditNet: Few-Shot Editing of Latent Semantics in GAN Spaces
Mengfei Xia (Tsinghua University), Yong-Jin Liu (Tsinghua University)
GenerationData SynthesisGenerative Adversarial NetworkContrastive LearningImage
🎯 What it does: Learn interpretable latent semantics and achieve attribute editing using a small number of labeled samples in a pre-trained GAN.
FedMDFG: Federated Learning with Multi-Gradient Descent and Fair Guidance
Zibin Pan (Chinese University of Hong Kong), Junhua Zhao (Chinese University of Hong Kong)
OptimizationFederated LearningImage
🎯 What it does: This paper proposes FedMDFG, which utilizes multi-gradient descent and fairness guidance to simultaneously solve for a fair descent direction and an appropriate step size in federated learning, thereby enhancing the fairness and convergence speed of the global model.
FedNP: Towards Non-IID Federated Learning via Federated Neural Propagation
Xueyang Wu (Hong Kong University of Science and Technology), Qian Xu (Hong Kong University of Science and Technology)
Federated LearningImageAudio
🎯 What it does: The FedNP method is proposed, which incorporates a neural propagation task into local training in federated learning to explicitly estimate the global data distribution, thereby alleviating the performance degradation caused by non-IID data.
FeedFormer: Revisiting Transformer Decoder for Efficient Semantic Segmentation
Jae-hun Shim (Sogang University), Suk-Ju Kang (Sogang University)
SegmentationTransformerImage
🎯 What it does: This paper proposes FeedFormer, a model that enhances high-level structural information and completes semantic segmentation by utilizing a Transformer decoder, using low-level features as keys/values and high-level features as queries.
Few-Shot 3D Point Cloud Semantic Segmentation via Stratified Class-Specific Attention Based Transformer Network
Canyu Zhang (University of South Carolina), Song Wang (University of South Carolina)
SegmentationTransformerPoint Cloud
🎯 What it does: A hierarchical class-specific attention Transformer network is proposed for few-shot 3D point cloud semantic segmentation, directly utilizing multi-scale point cloud features without relying on pooling or graph construction, improving matching accuracy and inference speed.
Few-Shot Composition Learning for Image Retrieval with Prompt Tuning
Junda Wu (New York University), Ricardo Henao (Duke University)
RetrievalTransformerPrompt EngineeringContrastive LearningImage
🎯 What it does: This study focuses on few-shot ensemble learning for image retrieval, proposing PromptCLIP, which combines visual prompts and text prompts using CLIP, and enhances the learning efficiency of visual prompts through self-supervised auxiliary tasks.
Few-Shot Defect Image Generation via Defect-Aware Feature Manipulation
Yuxuan Duan (Shanghai Jiao Tong University), Liqing Zhang (Shanghai Jiao Tong University)
GenerationData SynthesisAnomaly DetectionGenerative Adversarial NetworkImage
🎯 What it does: This paper proposes a few-shot defect image generation method called DFMGAN, which addresses the scarcity of industrial defect images. It first trains StyleGAN2 on defect-free images as a base, and then applies defect-aware residual blocks to manipulate features in specific defect areas, generating realistic and diverse defect images while simultaneously producing corresponding defect masks, achieving automated augmentation of defect images.
Few-Shot Object Detection via Variational Feature Aggregation
Jiaming Han (Wuhan University), Gui-Song Xia (Wuhan University)
Object DetectionMeta LearningAuto EncoderImage
🎯 What it does: A meta-learning framework is proposed, which includes two new feature aggregation schemes: Class-Agnostic Aggregation (CAA) and Variational Feature Aggregation (VFA) for few-shot object detection.
FinalMLP: An Enhanced Two-Stream MLP Model for CTR Prediction
Kelong Mao (Renmin University of China), Zhenhua Dong (Tsinghua University)
Recommendation SystemTabularBenchmark
🎯 What it does: A two-stream MLP structure called FinalMLP is proposed, with the addition of stream-specific feature selection and multi-head bilinear aggregation layers to enhance CTR prediction performance.
Find Beauty in the Rare: Contrastive Composition Feature Clustering for Nontrivial Cropping Box Regression
Zhiyu Pan (Huazhong University of Science and Technology), Weicai Zhong (Huawei)
Object DetectionConvolutional Neural NetworkContrastive LearningImage
🎯 What it does: Proposes the Contrastive Composition Clustering (C2C) mechanism, allowing the image cropping regression model to learn richer compositional features from rare samples, avoiding the generation of trivial cropping boxes.
Finding Fair Allocations under Budget Constraints
Siddharth Barman (Indian Institute of Science), K. V. N. Sreenivas (Indian Institute of Science)
Optimization
🎯 What it does: A polynomial-time algorithm based on density greedy is proposed to solve the fair allocation of indivisible items under budget constraints;
Finding Good Partial Assignments during Restart-Based Branch and Bound Search
Hongbo Li (Northeast Normal University), Jimmy H.M. Lee (Chinese University of Hong Kong)
OptimizationTabularBenchmark
🎯 What it does: This paper proposes an algorithm that dynamically generates and utilizes good partial assignments in restart-based branch-and-bound search to accelerate the solution of general constraint optimization problems.
Fine-Grained Position Helps Memorizing More, a Novel Music Compound Transformer Model with Feature Interaction Fusion
Zuchao Li (Wuhan University), Kehua Su (Wuhan University)
ClassificationRecognitionTransformerSequentialAudio
🎯 What it does: In the music notation understanding task, the authors proposed the Feature Interaction Fusion (FiF) module and the Rotational Absolute-Relative Position Encoding (RoAR), improving the Composite Word Transformer (CP+Transformer) model to better capture the interrelationships among multiple attributes in music events and obtain finer-grained positional information.
Fine-Grained Retrieval Prompt Tuning
Shijie Wang (Dalian University of Technology), Qi Tian (University of Sydney)
RetrievalPrompt EngineeringImage
🎯 What it does: This paper proposes a Fine-Grained Retrieval Prompt Tuning (FRPT) method based on frozen pre-trained models, utilizing sample prompts and feature adaptation to achieve fine-grained image retrieval.
Finite Based Contraction and Expansion via Models
Ricardo Guimarães (University of Bergen), Jandson S. Ribeiro (University of Hagen)
🎯 What it does: This paper proposes a new belief change paradigm that uses a finite base to represent the agent's knowledge state and models external information using arbitrary sets, thereby enabling reception and eviction operations on the finite base.
Fisher Markets with Social Influence
Jiayi Zhao (Pomona College), Amy Greenwald (Brown University)
Optimization
🎯 What it does: Incorporating social influence into the Fisher market, this paper constructs an influenced Fisher market model and proves the existence of competitive equilibrium; by constructing a single seller-buyer pseudo-game and a buyer-only pseudo-game, it utilizes variational equilibrium and a dual Stackelberg game system to solve for competitive equilibrium; and proposes a tatonnement algorithm based on the NE oracle that achieves polynomial time solutions under various common utility functions.
FiTs: Fine-Grained Two-Stage Training for Knowledge-Aware Question Answering
Qichen Ye (Peking University), Yuexian Zou (Peking University)
RetrievalKnowledge DistillationGraph Neural NetworkTransformerContrastive LearningText
🎯 What it does: A Fine-grained Two-stage Training (FiTs) framework is proposed for Knowledge-aware Question Answering (KAQA), which first aligns the PLM and KG representations through knowledge adaptation and then fine-tunes the model reasoning through self-supervised objectives (KSD, KBR).
Fixed-Weight Difference Target Propagation
Tatsukichi Shibuya (Tokyo Institute of Technology), Ikuro Sato (Denso IT Laboratory)
ClassificationConvolutional Neural NetworkImage
🎯 What it does: A differential target propagation algorithm with fixed feedback weights is proposed, eliminating the need for feedback network training in traditional target propagation.
FLAME: Free-Form Language-Based Motion Synthesis & Editing
Jihoon Kim (Korea University), Sungjoon Choi (Kakao Brain)
GenerationData SynthesisPose EstimationTransformerDiffusion modelVideoText
🎯 What it does: A diffusion model named FLAME is proposed, achieving free-text driven 3D action generation and editing;
Flexible 3D Lane Detection by Hierarchical Shape Matching
Zhihao Guan (Xi'an Jiaotong University), Shuqi Mei (Tencent)
Object DetectionAutonomous DrivingConvolutional Neural NetworkTransformerPoint Cloud
🎯 What it does: A hierarchical end-to-end 3D lane detection framework (FHLD) has been designed and implemented, capable of simultaneously predicting global parameter curves and local segment shapes on the bird's-eye view (BEV) of point clouds, thereby outputting a flexible and accurate set of lane points.
Flexible Budgets in Restless Bandits: A Primal-Dual Algorithm for Efficient Budget Allocation
Paula Rodriguez Diaz, Milind Tambe (Harvard University)
OptimizationReinforcement Learning from Human Feedback
🎯 What it does: This paper proposes a flexible budget recreational multi-armed bandit (F-RMAB) model and provides a concave-convex minimax optimization framework based on Lagrangian relaxation, utilizing the primal-dual gradient algorithm (PDGS) to solve the problem, thereby obtaining high-quality resource allocation strategies.
Flora: Dual-Frequency LOss-Compensated ReAl-Time Monocular 3D Video Reconstruction
Likang Wang (Hong Kong University of Science and Technology), Lei Chen (Hong Kong University of Science and Technology)
SegmentationData SynthesisDepth EstimationRecurrent Neural NetworkSimultaneous Localization and MappingVideo
🎯 What it does: A real-time monocular 3D video reconstruction method called Flora is proposed, which achieves high-quality and complete 3D reconstruction through dual-frequency distortion compensation aggregation and loss compensation.
Flow to Control: Offline Reinforcement Learning with Lossless Primitive Discovery
Yiqin Yang (Tsinghua University), Chongjie Zhang (Tsinghua University)
Reinforcement LearningFlow-based ModelTabularBenchmark
🎯 What it does: The study enhances performance in offline reinforcement learning through hierarchical skill discovery, proposing a reversible low-level skill learning method (Lossless Primitive Discovery, LPD) and combining it with high-level offline algorithms.
Flow-Based Robust Watermarking with Invertible Noise Layer for Black-Box Distortions
Han Fang (National University of Singapore), Ee-Chien Chang (National University of Singapore)
RestorationFlow-based ModelImage
🎯 What it does: A digital watermarking framework based on reversible networks has been designed and implemented, integrating reversible noise layers to achieve high robustness against both white-box and black-box distortions.
FlowFace: Semantic Flow-Guided Shape-Aware Face Swapping
Hao Zeng (Netease Fuxi AI Lab), Xin Yu (University of Technology Sydney)
Image TranslationGenerationGenerative Adversarial NetworkOptical FlowImage
🎯 What it does: A two-stage shape-aware face swapping framework called FlowFace is proposed, which first reshapes the target face shape through semantic flow, and then projects the internal features of the source face onto the reshaped face.
fmLRE: A Low-Resource Relation Extraction Model Based on Feature Mapping Similarity Calculation
Peng Wang (Southeast University), Wenjun Ke (Southeast University)
Reinforcement LearningText
🎯 What it does: In the low-resource relation extraction task, fmLRE is proposed, which generates pseudo-labels through self-training and calculates the similarity between pseudo-labels and true labels in the feature mapping space, using reinforcement learning iterative feedback to filter high-precision pseudo-labels, thereby reducing the bias in pseudo-label selection.
FoPro: Few-Shot Guided Robust Webly-Supervised Prototypical Learning
Yulei Qin (Tencent), Chunhua Shen (Shanghai Jiao Tong University)
ClassificationDomain AdaptationContrastive LearningImage
🎯 What it does: This paper proposes a prototype learning framework called FoPro, guided by a small number of real samples, to jointly train on noisy images crawled from the web and a limited amount of real labeled data, enhancing classification performance in real-world domains.
Forecasting with Sparse but Informative Variables: A Case Study in Predicting Blood Glucose
Harry Rubin-Falcone (University of Michigan), Jenna Wiens (University of Michigan)
Recurrent Neural NetworkTime SeriesBiomedical Data
🎯 What it does: A link encoder/decoder architecture is proposed to handle sparse but informative auxiliary variables (SIV) in time series tasks such as blood glucose prediction.
Foresee What You Will Learn: Data Augmentation for Domain Generalization in Non-stationary Environment
Qiuhao Zeng (University of Western Ontario), Boyu Wang (University of Western Ontario)
Domain AdaptationOptimizationKnowledge DistillationMeta LearningTransformerImage
🎯 What it does: A method for domain generalization under non-stationary environments (DDA) is proposed, which generates feature enhancement for the target domain through an attention domain transformer and captures domain evolution patterns using double-layer optimization and meta-learning.
Formal Verification of Bayesian Mechanisms
Munyque Mittelmann (Universita degli Studi di Napoli Federico II), Laurent Perrussel (Universite Toulouse Capitole)
🎯 What it does: This paper is the first to apply Probabilistic Strategy Logic (PSL) for the formal verification of Bayesian mechanisms by modeling Bayesian mechanisms as stochastic transition systems, thereby transforming various types of Bayesian equilibria (BRE, BNE, NE) and classical mechanism design properties (IR, DSIC, BIC, etc.) into PSL formulas, and utilizing PSL model checking for automated verification.
Fourier-Net: Fast Image Registration with Band-Limited Deformation
Xi Jia (University of Birmingham), Jinming Duan (Alan Turing Institute)
SegmentationOptimizationComputational EfficiencyConvolutional Neural NetworkOptical FlowImageBiomedical DataMagnetic Resonance Imaging
🎯 What it does: An unsupervised Fourier-Net is proposed, which represents the displacement/velocity field in a low-dimensional frequency domain (band-limited Fourier), trains a simplified U-Net structure, and decodes the complete displacement field through zero-padding + inverse discrete Fourier transform (iDFT), achieving fast medical image registration.
Frame-Level Label Refinement for Skeleton-Based Weakly-Supervised Action Recognition
Qing Yu (University of Tokyo), Kent Fujiwara (LINE Corporation)
RecognitionGraph Neural NetworkSupervised Fine-TuningVideo
🎯 What it does: The S-WTAL (Skeleton-based Weakly-Supervised Temporal Action Localization) problem is proposed, and a framework based on ST-GCN is designed to achieve action segment localization and recognition by learning from video-level labels and iteratively optimizing frame-level pseudo-labels.
FreeEnricher: Enriching Face Landmarks without Additional Cost
Yangyu Huang (Microsoft Research Asia), Dong Chen (Microsoft Research Asia)
Pose EstimationOptimizationConvolutional Neural NetworkImage
🎯 What it does: The FreeEnricher framework is proposed, which expands the density of facial landmarks and improves alignment accuracy using sparse annotated data without increasing additional annotation or inference time.
Frequency Domain Disentanglement for Arbitrary Neural Style Transfer
Dongyang Li (Alibaba Group), Fan Wang (Alibaba Group)
Image TranslationGenerationGenerative Adversarial NetworkContrastive LearningImage
🎯 What it does: This paper proposes a style transfer framework (FDD) that separates and fuses content and style in the frequency domain.
Frequency Selective Augmentation for Video Representation Learning
Jinhyung Kim (LG AI Research), Junmo Kim (KAIST)
Representation LearningContrastive LearningVideo
🎯 What it does: This paper proposes and validates an enhancement method called FreqAug, which randomly removes low-frequency components of videos in the frequency domain to improve the transfer performance of self-supervised video representations.
Frido: Feature Pyramid Diffusion for Complex Scene Image Synthesis
Wan-Cyuan Fan (National Taiwan University), Yu-Chiang Frank Wang (Microsoft Corporation)
GenerationData SynthesisConvolutional Neural NetworkTransformerDiffusion modelImageMultimodality
🎯 What it does: A multi-scale coarse-to-fine progressive diffusion model called Frido is proposed, which generates complex scene images using discrete multi-scale latent encoding and a shared U-Net, supporting multimodal conditions such as text, layout, scene graphs, and labels.
From Coarse to Fine: Hierarchical Pixel Integration for Lightweight Image Super-resolution
Jie Liu (Nanjing University), Gangshan Wu (Nanjing University)
RestorationSuper ResolutionTransformerImage
🎯 What it does: A lightweight single-image super-resolution network HPINet has been designed and implemented, capable of achieving high-quality super-resolution results while maintaining a low parameter count.
From Monopoly to Competition: Optimal Contests Prevail
Xiaotie Deng (Peking University), Hongyi Ling (Peking University)
Optimization
🎯 What it does: The study proves that the optimal competition (maximizing total effort) in a single competition remains an equilibrium solution in a competitive environment, demonstrating its dominance and uniqueness within the framework of multi-contest design.
From Understanding the Population Dynamics of the NSGA-II to the First Proven Lower Bounds
Benjamin Doerr (Ecole Polytechnique), Zhongdi Qu (Ecole Polytechnique)
OptimizationBenchmark
🎯 What it does: This paper presents for the first time the lower bounds of NSGA-II on the benchmark multi-objective problems ONEMINMAX and ONEJUMPZEROJUMP, and proves through probabilistic analysis of population dynamics that this lower bound matches the existing upper bound, deriving an exact constant term.
From Width-Based Model Checking to Width-Based Automated Theorem Proving
Mateus de Oliveira Oliveira (Stockholm University), Farhad Vadiee (University of Bergen)
Graph
🎯 What it does: This paper proposes a general framework that transforms width-based model checking algorithms into an automatic theorem proving method for verifying the truth of graph theory conjectures in graph classes of bounded width.
Frustratingly Easy Truth Discovery
Reshef Meir (Technion Israel Institute of Technology), Lirong Xia (Rensselaer Polytechnic Institute)
ClassificationSegmentationImageText
🎯 What it does: This paper proposes a simple heuristic method based on the average similarity (or distance) between workers to estimate worker accuracy and perform weighted aggregation of answers (P-TD algorithm).
FSR: A General Frequency-Oriented Framework to Accelerate Image Super-resolution Networks
Jinmin Li (Tsinghua University), Shu-Tao Xia (Tsinghua University)
RestorationSuper ResolutionImage
🎯 What it does: This paper proposes a framework for a general frequency domain accelerated super-resolution network called FSR.
FTM: A Frame-Level Timeline Modeling Method for Temporal Graph Representation Learning
Bowen Cao (Peking University), Yuexian Zou (Peking University)
Representation LearningGraph Neural NetworkGraphTime SeriesFinance Related
🎯 What it does: This paper proposes a temporal graph representation learning method called FTM, which is based on frame-level timeline modeling. It uses framing techniques and a timeline aggregator to capture both short-term and long-term features, thereby enhancing the quality of temporal graph representations.
Fully Computer-Assisted Proofs in Extremal Combinatorics
Olaf Parczyk (Freie Universitat Berlin), Tibor Szabó (Zuse Institute Berlin)
OptimizationGraph
🎯 What it does: Using computer-aided search (simulated annealing and tabu search) to find extremal constructions in extremal combinatorics, combined with flagmatic algebra (SDP) to provide lower bounds and stability results, achieving new upper bounds and uniqueness proofs for Ramsey numbers, (3,4) numbers, and g_{4,5}.
Fully Dynamic Online Selection through Online Contention Resolution Schemes
Vashist Avadhanula (Meta), Matteo Russo (Sapienza University)
Optimization
🎯 What it does: A general method for the fully dynamic online selection problem is proposed, utilizing the Online Contest Resolution Scheme (OCRS) to achieve near-optimal competitive ratios, and a no-α-regret algorithm is provided in partially informed environments.
Fully Online Matching with Stochastic Arrivals and Departures
Zihao Li (Nanyang Technological University), Zhenzhen Yan (Nanyang Technological University)
OptimizationGraph
🎯 What it does: The study investigates a model of random arrivals and departures in complete online matching and proposes a matching algorithm based on linear programming.
Fully-Dynamic Decision Trees
Marco Bressan (University of Milan), Mauro Sozio (Institut Polytechnique de Paris)
ClassificationOptimizationTabular
🎯 What it does: Designed and implemented the first fully dynamic decision tree algorithm (FUDYADT) that supports arbitrary insert/delete operations, capable of keeping the split point of each node in the tree within a preset ε of the optimal Gini gain at any moment, and achieving nearly optimal time-space complexity under real-time updates.
Function Approximation for Solving Stackelberg Equilibrium in Large Perfect Information Games
Chun Kai Ling (Carnegie Mellon University), Fei Fang (Carnegie Mellon University)
OptimizationNeural Architecture SearchReinforcement LearningTabular
🎯 What it does: A function approximation-based learning framework is proposed, using neural networks to approximate the enforceable payoff frontier (EPF), thereby solving the Stackelberg extensive form correlated equilibrium (SEFCE) in large complete information two-player games.
Fundamentals of Task-Agnostic Data Valuation
Mohammad Mohammadi Amiri (Massachusetts Institute of Technology), Ramesh Raskar (Massachusetts Institute of Technology)
Data-Centric LearningTabular
🎯 What it does: This paper proposes a task-agnostic data value assessment framework that does not require a validation set. It quantifies diversity and relevance by utilizing the statistical characteristics of the buyer's existing data and the second moment (covariance) differences of the seller's data in the same feature space, thereby providing a benchmark value for data transactions.
GAM: Gradient Attention Module of Optimization for Point Clouds Analysis
Haotian Hu (Zhejiang Leapmotor Technology CO., LTD.), Yanhao Zhang (OPPO Research Institute)
Object DetectionSegmentationOptimizationPoint Cloud
🎯 What it does: A Gradient Attention Module (GAM) is proposed to incorporate gradient information during the local feature aggregation process of point clouds.
Game Implementation: What Are the Obstructions?
Jiehua Chen (TU Wien), Manuel Sorge (TU Wien)
🎯 What it does: This paper studies the game implementation problem (GAME IMPLEMENTATION) of achieving a given set of strategies through payment commitments, proving that it remains NP-hard under various constrained conditions, and provides a complete characterization of zero-budget implementation along with corrections to previously proposed algorithms.
GAN Prior Based Null-Space Learning for Consistent Super-resolution
Yinhuai Wang (Peking University), Jian Zhang (Peking University)
RestorationSuper ResolutionGenerative Adversarial NetworkImage
🎯 What it does: A Pooling-Based Decomposition (PD) method is proposed based on GAN prior and null-space learning, utilizing the pseudo-inverse of average pooling to achieve parameter-free, no additional computation range-zero space decomposition, significantly eliminating low-frequency inconsistencies in super-resolution results.
GANTEE: Generative Adversarial Network for Taxonomy Enterance Evaluation
Zhouhong Gu (Fudan University), Zhong Jian (HUAWEI CBG Edu AI Lab)
GenerationOptimizationComputational EfficiencyTransformerReinforcement LearningGenerative Adversarial NetworkGraph
🎯 What it does: A pluggable Generative Adversarial Network framework called GANTEE is proposed to enhance the effectiveness and efficiency of taxonomy expansion while addressing the evaluation issue of whether new concepts should be added to the taxonomy.
GenéLive! Generating Rhythm Actions in Love Live!
Atsushi Takada (KLab Inc), Daisuke Sakurai (Kyushu University)
GenerationHyperparameter SearchConvolutional Neural NetworkRecurrent Neural NetworkSequentialAudio
🎯 What it does: A rhythm action game score generation model named Gen' eLive! has been developed, and the functionality for automated generation of the first draft has been implemented in KLab's business.
General Acyclicity and Cyclicity Notions for the Disjunctive Skolem Chase
Lukas Gerlach (TU Dresden), David Carral (Inria)
🎯 What it does: This paper proposes a new method for detecting cycles and acyclicity in Skolem Chase with disjunctive existence rules, extending and improving the existing MFA/MFC detection, which can more effectively determine the termination and non-termination of rule sets.
Generalization Bounds for Inductive Matrix Completion in Low-Noise Settings
Antoine Ledent (Singapore Management University), Marius Kloft (Czech Technical University in Prague)
Recommendation SystemDrug DiscoveryTabular
🎯 What it does: This paper studies the matrix completion (IMC) problem with side information under low noise, providing new exact recovery thresholds and a generalization error upper bound related to the noise level, thereby bridging the theoretical gap between exact recovery and approximate recovery.
Generalized Category Discovery with Decoupled Prototypical Network
Wenbin An (Xi'an Jiaotong University), Ping Chen (University of Massachusetts Boston)
ClassificationTransformerText
🎯 What it does: By decoupling known categories from unknown categories in unlabeled data and utilizing prototype matching and semantic-aware prototype learning, a Decoupled Prototypical Network (DPN) is proposed for Generalized Category Discovery (GCD).
Generalized Cell Type Annotation and Discovery for Single-Cell RNA-Seq Data
Yuyao Zhai (Peking University), Minghua Deng (Peking University)
ClassificationRecognitionAuto EncoderBiomedical Data
🎯 What it does: Proposes the scGAD method for general cell type annotation and discovery in single-cell RNA-seq data.
Generalized Confidence Constraints
Guillaume Perez (Huawei Technologies), Arnaud Lallouet (Universite Cote d'Azur)
OptimizationTabular
🎯 What it does: This paper proposes a general Confidence constraint, utilizing Multi-Valued Decision Diagrams (MDD) to filter and infer the probability of random variables satisfying the constraints, and applies it to the chemical distribution allocation problem.
Generalized Semantic Segmentation by Self-Supervised Source Domain Projection and Multi-Level Contrastive Learning
Liwei Yang (Xi'an Jiaotong University), Jian Sun (Pazhou Laboratory)
SegmentationDomain AdaptationAutonomous DrivingContrastive LearningImage
🎯 What it does: Enhancing the domain generalization performance of semantic segmentation through Self-Supervised Source Domain Projection (SSDP) and Multi-Layer Contrastive Learning (MLCL).
Generalizing Downsampling from Regular Data to Graphs
Davide Bacciu (Universita di Pisa), Francesco Landolfi (Universita di Pisa)
CompressionOptimizationGraph Neural NetworkGraph
🎯 What it does: A graph downsampling/pooling method based on k-MIS is proposed, which achieves controllable compression while preserving the graph structure.
Generalizing Math Word Problem Solvers via Solution Diversification
Zhenwen Liang (University of Notre Dame), Xiangliang Zhang (University of Notre Dame)
Contrastive LearningText
🎯 What it does: A general training framework is proposed, which dynamically generates and filters diverse, quality-controllable answers during the training process of the MWP Solver by introducing an answer buffer and a discriminator, thereby enhancing the model's generalization ability.
Generalizing Multiple Object Tracking to Unseen Domains by Introducing Natural Language Representation
En Yu (Huazhong University of Science and Technology), Wenbing Tao (Huazhong University of Science and Technology)
Object TrackingDomain AdaptationAutonomous DrivingTransformerVision Language ModelVideoText
🎯 What it does: This paper proposes LTrack, a multi-object tracking model that combines natural language descriptions with visual context. It utilizes Visual Context Prompts (VCP) and a Visual-Language Mixing (VLM) module to automatically generate Pseudo Text Descriptions (PTD), thereby enhancing cross-domain generalization performance.
Generating Coherent Narratives by Learning Dynamic and Discrete Entity States with a Contrastive Framework
Jian Guan (Tsinghua University), Minlie Huang (Tsinghua University)
GenerationTransformerContrastive LearningText
🎯 What it does: A discrete entity state representation framework based on contrastive learning (ERIC) is proposed, integrating dynamic updates of entity states and a state attention layer into the Transformer decoder to achieve coherent text generation for stories and news.
Generating Transferable 3D Adversarial Point Cloud via Random Perturbation Factorization
Bangyan He (University of Chinese Academy of Sciences), Xiaochun Cao (Sun Yat-sen University)
Adversarial AttackGraph Neural NetworkPoint Cloud
🎯 What it does: This paper proposes an attack method based on Randomized Perturbation Factorization (PF-Attack), which generates more transferable 3D point cloud adversarial samples by simultaneously optimizing the perturbation and its sub-perturbations.
Generative Image Inpainting with Segmentation Confusion Adversarial Training and Contrastive Learning
Zhiwen Zuo (Zhejiang University), Dongming Lu (Zhejiang University)
RestorationGenerationGenerative Adversarial NetworkContrastive LearningImage
🎯 What it does: A new image inpainting framework is proposed, combining Segmentation Confusion Adversarial Training (SCAT) with contrastive learning, enhancing local texture and global consistency, and capable of handling freely shaped missing regions.
Generative Label Enhancement with Gaussian Mixture and Partial Ranking
Yunan Lu (Nanjing University of Science and Technology), Xiuyi Jia (Nanjing University of Science and Technology)
GenerationData-Centric LearningTabularBiomedical Data
🎯 What it does: This paper proposes a generative label enhancement method based on Gaussian mixture models and partial ranking, called GLEMR, to recover label distributions from data with only logical labels.