Establishing a Dialog Agent Policy using Deep Reinforcement Learning in the Psychotherapy Domain.
Chester Holt-QuickJim WarrenPublished in: ACSW (2021)
Keyphrases
- reinforcement learning
- action selection
- reward function
- optimal policy
- markov decision process
- multi agent
- partially observable
- state action
- agent receives
- intelligent agents
- conversational agent
- conversational agents
- function approximation
- partially observable domains
- markov decision processes
- policy search
- state space
- action space
- partially observable markov decision process
- multi agent systems
- multi agent environments
- agent learns
- markov decision problems
- domain specific
- selective perception
- state and action spaces
- learned knowledge
- control policy
- learning agent
- decision making
- partially observable markov decision processes
- domain independent
- learning capabilities
- natural language
- agent architecture
- dynamic programming
- transfer learning
- reinforcement learning agents
- autonomous agents
- inverse reinforcement learning
- state abstraction
- continuous state spaces
- decision theoretic
- single agent
- policy iteration
- optimal control
- policy gradient
- multiple agents
- reinforcement learning problems
- control policies
- function approximators
- partially observable environments
- multiagent systems
- reward signal
- mobile agents
- learning process