Aligning as Debiasing: Causality-Aware Alignment via Reinforcement Learning with Interventional Feedback.
Yu XiaTong YuZhankui HeHandong ZhaoJulian J. McAuleyShuai LiPublished in: NAACL-HLT (2024)
Keyphrases
- reinforcement learning
- learning algorithm
- function approximation
- image alignment
- state space
- reinforcement learning algorithms
- relevance feedback
- causal reasoning
- model free
- supervised learning
- action selection
- multi agent reinforcement learning
- optimal policy
- feedback mechanisms
- active exploration
- causal inference
- policy search
- neural network
- temporal difference
- patient specific
- user feedback
- computer assisted
- transfer learning
- image registration
- image analysis
- multi agent
- machine learning