Modifying RL Policies with Imagined Actions: How Predictable Policies Can Enable Users to Perform Novel Tasks.
Isaac S. SheidlowerReuben M. AronsonElaine ShortPublished in: CoRR (2023)
Keyphrases
- optimal policy
- multiagent reinforcement learning
- reinforcement learning
- decision processes
- hierarchical reinforcement learning
- transfer learning
- markov decision process
- reinforcement learning agents
- control policies
- user interface
- policy search
- reward function
- markov decision processes
- end users
- action selection
- decision theoretic
- partially observable markov decision processes
- human users
- action space
- decision problems
- stochastic games
- user requests
- dynamic programming
- social media
- recommender systems
- macro actions