performance improvement
Performance improvement refers to advancements in the accuracy, efficiency, or other metrics of an AI model over time. This can result from optimizing algorithms, refining architectures, or using better training techniques.
- 1000 Layer Networks for Self-Supervised RL: Scaling Depth Can Enable New Goal-Reaching Capabilities
- 3D Interaction Geometric Pre-training for Molecular Relational Learning
- A Bayesian Approach to Contextual Dynamic Pricing using the Proportional Hazards Model with Discrete Price Data
- A*-Thought: Efficient Reasoning via Bidirectional Compression for Low-Resource Settings
- Are Pixel-Wise Metrics Reliable for Computerized Tomography Reconstruction?
- Automated Detection of Visual Attribute Reliance with a Self-Reflective Agent
- Availability-aware Sensor Fusion via Unified Canonical Space
- BraVE: Offline Reinforcement Learning for Discrete Combinatorial Action Spaces
- ComPO: Preference Alignment via Comparison Oracles
- Combinatorial Ski Rental Problem: Robust and Learning-Augmented Algorithms
- Continual Knowledge Adaptation for Reinforcement Learning
- Continuous Subspace Optimization for Continual Learning
- CoreaSpeech: Korean Speech Corpus via JAMO-based Coreset Selection for Efficient and Robust Korean Speech Generation
- Cost-Sensitive Freeze-thaw Bayesian Optimization for Efficient Hyperparameter Tuning
- D2SA: Dual-Stage Distribution and Slice Adaptation for Efficient Test-Time Adaptation in MRI Reconstruction
- DBLoss: Decomposition-based Loss Function for Time Series Forecasting
- Deferring Concept Bottleneck Models: Learning to Defer Interventions to Inaccurate Experts
- Defining and Discovering Hyper-meta-paths for Heterogeneous Hypergraphs
- Dual Alignment Framework for Few-shot Learning with Inter-Set and Intra-Set Shifts
- Efficient Prompt Compression with Evaluator Heads for Long-Context Transformer Inference
- Efficient Training-Free Online Routing for High-Volume Multi-LLM Serving
- End-to-End Low-Light Enhancement for Object Detection with Learned Metadata from RAWs
- Enhancing Bioactivity Prediction via Spatial Emptiness Representation of Protein-ligand Complex and Union of Multiple Pockets
- Enhancing Infrared Vision: Progressive Prompt Fusion Network and Benchmark
- Enhancing Sample Selection Against Label Noise by Cutting Mislabeled Easy Examples
- Escaping Collapse: The Strength of Weak Data for Large Language Model Training
- Escaping the SpuriVerse: Can Large Vision-Language Models Generalize Beyond Seen Spurious Correlations?
- FADRM: Fast and Accurate Data Residual Matching for Dataset Distillation
- Few-Shot Knowledge Distillation of LLMs With Counterfactual Explanations
- Forecasting in Offline Reinforcement Learning for Non-stationary Environments
- Functional Virtual Adversarial Training for Semi-Supervised Time Series Classification
- Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Free
- Generalizing Experience for Language Agents with Hierarchical MetaFlows
- GoRA: Gradient-driven Adaptive Low Rank Adaptation
- HMARL-CBF – Hierarchical Multi-Agent Reinforcement Learning with Control Barrier Functions for Safety-Critical Autonomous Systems
- HelpSteer3-Preference: Open Human-Annotated Preference Data across Diverse Tasks and Languages
- How Well Can Differential Privacy Be Audited in One Run?
- IR-OptSet: An Optimization-Sensitive Dataset for Advancing LLM-Based IR Optimizer
- Impartial Selection with Predictions
- Joint Hierarchical Representation Learning of Samples and Features via Informed Tree-Wasserstein Distance
- LLM-Explorer: A Plug-in Reinforcement Learning Policy Exploration Enhancement Driven by Large Language Models
- LabelAny3D: Label Any Object 3D in the Wild
- Learning Reconfigurable Representations for Multimodal Federated Learning with Missing Data
- Learning from Demonstrations via Capability-Aware Goal Sampling
- LoRASuite: Efficient LoRA Adaptation Across Large Language Model Upgrades
- MERIT: Multilingual Semantic Retrieval with Interleaved Multi-Condition Query
- MM-Agent: LLM as Agents for Real-world Mathematical Modeling Problem
- MTL-KD: Multi-Task Learning Via Knowledge Distillation for Generalizable Neural Vehicle Routing Solver
- MTRec: Learning to Align with User Preferences via Mental Reward Models
- MemEIC: A Step Toward Continual and Compositional Knowledge Editing
- Model-Based Policy Adaptation for Closed-Loop End-to-end Autonomous Driving
- Not All Data are Good Labels: On the Self-supervised Labeling for Time Series Forecasting
- OCN: Effectively Utilizing Higher-Order Common Neighbors for Better Link Prediction
- On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target Stochasticity
- On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target Stochasticity
- Online Mixture of Experts: No-Regret Learning for Optimal Collective Decision-Making
- Optimal Neural Compressors for the Rate-Distortion-Perception Tradeoff
- OverLayBench: A Benchmark for Layout-to-Image Generation with Dense Overlaps
- PINNs with Learnable Quadrature
- Path-Enhanced Contrastive Learning for Recommendation
- Physics-informed Value Learner for Offline Goal-Conditioned Reinforcement Learning
- Pretraining a Shared Q-Network for Data-Efficient Offline Reinforcement Learning
- ProfiX: Improving Profile-Guided Optimization in Compilers with Graph Neural Networks
- QiMeng-NeuComBack: Self-Evolving Translation from IR to Assembly Code
- QuARI: Query Adaptive Retrieval Improvement
- Quantum speedup of non-linear Monte Carlo problems
- RANK++LETR: Learn to Rank and Optimize Candidates for Line Segment Detection
- Rao-Blackwellised Reparameterisation Gradients
- ReDit: Reward Dithering for Improved LLM Policy Optimization
- Recursive Inference Scaling: A Winning Path to Scalable Inference in Language and Multimodal Systems
- Redundancy-Aware Test-Time Graph Out-of-Distribution Detection
- Rethinking Residual Distribution in Locate-then-Edit Model Editing
- Robust Contextual Pricing
- SQLens: An End-to-End Framework for Error Detection and Correction in Text-to-SQL
- STNet: Spectral Transformation Network for Solving Operator Eigenvalue Problem
- SUMO: Subspace-Aware Moment-Orthogonalization for Accelerating Memory-Efficient LLM Training
- Set-LLM: A Permutation-Invariant LLM
- Sparse Polyak: an adaptive step size rule for high-dimensional M-estimation
- Stitch and Tell: A Structured Data Augmentation Method for Spatial Understanding
- Structured Initialization for Vision Transformers
- SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines
- System Prompt Optimization with Meta-Learning
- TANDEM: Bi-Level Data Mixture Optimization with Twin Networks
- TTRL: Test-Time Reinforcement Learning
- The Gaussian Mixing Mechanism: Renyi Differential Privacy via Gaussian Sketches
- The Surprising Effectiveness of Negative Reinforcement in LLM Reasoning
- Towards General Continuous Memory for Vision-Language Models
- TrajAgent: An LLM-Agent Framework for Trajectory Modeling via Large-and-Small Model Collaboration
- Unifying Reconstruction and Density Estimation via Invertible Contraction Mapping in One-Class Classification
- VIBE: Annotation-Free Video-to-Text Information Bottleneck Evaluation for TL;DR
- VIPAMIN: Visual Prompt Initialization via Embedding Selection and Subspace Expansion
- Value-Guided Search for Efficient Chain-of-Thought Reasoning
- Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision
- Why Playing Against Diverse and Challenging Opponents Speeds Up Coevolution: A Theoretical Analysis on Combinatorial Games
- ZPressor: Bottleneck-Aware Compression for Scalable Feed-Forward 3DGS