Reinforcement Learning 101
Any new subject can be learned in the same way as we learn any new language (Learning new vocabulary → Understanding Language syntax → Conversing with localites) Objective of reinforcement learning is to implement automation, optimization and discovery with minimal human intervention. In this blog I will give an introduction to Reinforcement learning vocab Agent Single Multi Independent Q-learning CTDE MADDPG Environment Stationary Non-Stationary State Action Episode, Visit Reward Long Term Short Term Markov Decision Process Bellman Equation Transition Probability Discounting Factor Policy Value Function State Monte Carlo Method to learn Action Model Free Q-Learning SARSA DQN Based Dyna-Q Monte Carlo Tree Search PILCO Policy Gradient Method REINFORCE Iteration Policy Value Evaluation Policy Improvement Policy Exploration vs Eploitation Exporation Strategies Initialization Termination Actor Critic Methods I'll keep updating this blog whenever I have time
