What is: Reinforcement Learning (Interactive Learning in Decision Processes)?
— Is there a way to learn by interacting
— i.e. interact have experience and use the experience to learn (predict the future)
— Interact to explore and utilize what makes learning (goal/outcome) enhanced
— The computation approach of this method is Reinforcement Learning (Interactive Learning in Decision Processes)?
— it is a goal oriented learning from interactions
— it has it’s root in Markov decision process (MDP)
— Markov decision process (MDP) is a model for sequential decision making when outcomes are uncertain.[1] [Wikipedia]
Reinforcement Learning (Interactive Learning in Decision Processes) Involves:
Markov decision processes
Dynamic Programming
Monte Carlo methods
Temporal-difference learning
Function approximation methods
Monte Carlo methods
Solves problems with repeated random sampling.
Temporal-difference learning: combination of the Monte Carlo (MC) method and the Dynamic Programming (DP) method.
Function approximation methods
