Jaeyoung Do

reinforcement learning (1) off-policy evaluation (1) reward modeling (1) concept-based explanation (1) visual saliency (1) weakly supervised learning (1) conversational ai (1) multimodal learning (1) automatic speech recognition (1) language model alignment (1) task generalization (1) reinforcement learning from human feedback (1) model alignment (1) policy improvement (1) off-policy reinforcement learning (1) lifelong learning (1) safe reinforcement learning (1) image classifier (1) proximal policy optimization (1) text embedding (1)