policy iteration

CellStrat > Research/Blog > policy iteration
20
Aug
AlphaZero with Monte Carlo Tree Search

#CellStratAILab #disrupt4.0 #WeCreateAISuperstars #WhereLearningNeverStops In recent weeks, I had presented a session on “AlphaZero with Monte Carlo Tree Search” algorithm at the CellStrat AI Lab. This is an algorithm developed by Google Deepmind in 2016. It mastered the game of GO and beat the 18-time world champion at the time Lee Sedol. Go is an ancient Chinese abstract strategy […]

Posted in: Gaming, Reinforcement Learning, Robotics,
Tags: Actor Critic, AlphaGO, AlphaZero, deepmind, Game Tree, MCTS, Monte Carlo Tree Search, policy evaluation, policy improvement, policy iteration, reinforcement learning, UCT, Upper Confidence Bound,
27
Apr
Temporal Difference methods in RL

This post discusses temporal difference (TD) methods, used in Reinforcement Learning. It contrasts TD methods with Monte Carlo (MC) methods and dynamic programming. You need to have a thorough understanding of Markov Decision Process (MDP) to understand this post. Prediction and Control : In general, RL methods have two components 1) Prediction / Evaluation : where […]

Posted in: Reinforcement Learning, Robotics,
Tags: epsilon greedy technique, exploration vs exploitation, On-policy vs off-policy, policy improvement, policy iteration, Q learning, reinforcement learning, SARSA, SARSAMax, TD control, TD Learning, Temporal Difference, value iteration,