The Effect of Eligibility Traces on Finding Optimal Memoryless Policies in Partially Observable Markov Decision Processes.
John LochPublished in: NIPS (1998)
Keyphrases
- partially observable markov decision processes
- finding optimal
- eligibility traces
- reinforcement learning
- policy evaluation
- optimal policy
- reinforcement learning algorithms
- finite state
- decision problems
- state space
- markov decision processes
- dynamical systems
- dynamic programming
- belief state
- reinforcement learning methods
- multi agent
- planning problems
- partially observable
- temporal difference
- function approximation
- average reward
- approximate solutions
- computational complexity
- decision processes
- model checking
- model free
- machine learning
- dec pomdps
- learning algorithm