state-of-the-art approaches
Strategies and methodologies employed in AI research that lead to the highest known performance in given tasks, often building on or innovating beyond existing state-of-the-art models.
- A High-Dimensional Statistical Method for Optimizing Transfer Quantities in Multi-Source Transfer Learning
- A Latent Multilayer Graphical Model For Complex, Interdependent Systems
- Bayesian Optimization with Preference Exploration using a Monotonic Neural Network Ensemble
- CoCoA: A Minimum Bayes Risk Framework Bridging Confidence and Consistency for Uncertainty Quantification in LLMs
- Constructing an Optimal Behavior Basis for the Option Keyboard
- Correlated Low-Rank Adaptation for ConvNets
- DERD-Net: Learning Depth from Event-based Ray Densities
- FineRS: Fine-grained Reasoning and Segmentation of Small Objects with Reinforcement Learning
- From Forecasting to Planning: Policy World Model for Collaborative State-Action Prediction
- Novel Exploration via Orthogonality
- Reinforcement Learning Meets Masked Generative Models: Mask-GRPO for Text-to-Image Generation
- Removing Concepts from Text-to-Image Models with Only Negative Samples
- Resource-Constrained Federated Continual Learning: What Does Matter?
- Robust Hallucination Detection in LLMs via Adaptive Token Selection
- STaRFormer: Semi-Supervised Task-Informed Representation Learning via Dynamic Attention-Based Regional Masking for Sequential Data