Self-Organization of Policy by Symmetric Reasoning and its Application to Reinforcement Learning.
Yu KohnoTatsuji TakahashiPublished in: CogSci (2012)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- action selection
- multi agent
- markov decision process
- control policy
- partially observable environments
- actor critic
- markov decision processes
- policy gradient
- partially observable
- reward function
- policy evaluation
- function approximation
- state space
- approximate dynamic programming
- policy iteration
- model free
- markov decision problems
- state and action spaces
- automated reasoning
- function approximators
- partially observable domains
- reasoning process
- knowledge representation
- reinforcement learning problems
- reasoning tasks
- reasoning systems
- state action
- control policies
- policy gradient methods
- average reward
- neural network
- unsupervised learning
- machine learning
- knowledge base
- qualitative reasoning
- reinforcement learning algorithms
- temporal difference
- continuous state
- long run
- finite state
- learning tasks
- state dependent
- transition model
- description logics
- supervised learning
- agent learns
- expert systems
- reinforcement learning methods
- learning algorithm
- agent receives