🏢 Nanjing University of Aeronautics and Astronautics
Optimistic Critic Reconstruction and Constrained Fine-Tuning for General Offline-to-Online RL
·2843 words·14 mins·
loading
·
loading
Machine Learning
Reinforcement Learning
🏢 Nanjing University of Aeronautics and Astronautics
This paper introduces OCR-CFT, a novel method for general offline-to-online RL, achieving stable and efficient performance improvements by addressing evaluation and improvement mismatches through opti…
Magnet: We Never Know How Text-to-Image Diffusion Models Work, Until We Learn How Vision-Language Models Function
·3895 words·19 mins·
loading
·
loading
Multimodal Learning
Vision-Language Models
🏢 Nanjing University of Aeronautics and Astronautics
Magnet: Enhancing Text-to-Image Synthesis by Disentangling Attributes in CLIP.
Learning Distinguishable Trajectory Representation with Contrastive Loss
·2347 words·12 mins·
loading
·
loading
Machine Learning
Reinforcement Learning
🏢 Nanjing University of Aeronautics and Astronautics
Contrastive Trajectory Representation (CTR) boosts multi-agent reinforcement learning by learning distinguishable agent trajectories using contrastive loss, thus improving performance significantly.
Forgetting, Ignorance or Myopia: Revisiting Key Challenges in Online Continual Learning
·2870 words·14 mins·
loading
·
loading
Machine Learning
Continual Learning
🏢 Nanjing University of Aeronautics and Astronautics
NsCE framework tackles key OCL challenges: model ignorance (learning effective features in limited time) and myopia (overly simplified features). NsCE integrates non-sparse maximum separation regulari…