／var／logmarcus chiu

Multi／K-Armed Bandit Problem

Created on

Feedback Types

Description

evaluative feedback

pure evaluative feedback indicates

  • depends on action
  • how good the action is
  • not whether it’s the best action

instructive feedback

pure instructive feedback indicates

  • independent of action
  • whether it’s the best action
  • not how good the action is

nonassociative - does not involve learning to act in more than one situation

k-armed bandit problem - exploration v exploitation

https://github.com/TheRealMarcusChiu/machine-learning-master-example/tree/master/reinforcement-learning/multi-arm-bandit