Emergence of Multi-step Discrete State Transition through Reinforcement Learning with a Recurrent Neural Network.
Mohamad Faizal Bin SamsudinYoshito SawatsubashiKatsunari ShibataPublished in: ICONIP (2) (2012)
Keyphrases
- multi step
- state transition
- reinforcement learning
- state space
- transition model
- state transitions
- state transition model
- input output
- black box
- probability distribution
- reinforcement learning algorithms
- hidden markov models
- markov chain
- single step
- optimal policy
- markov decision processes
- function approximation
- finite state machines
- partially observable
- knn
- temporal difference
- dynamic programming
- database
- k nearest neighbor
- semi supervised
- markov decision problems
- machine learning
- model free
- linear programming
- supervised learning
- nearest neighbor
- action space
- high dimensional
- learning algorithm