Messaouda Benzaoui
Teacher, Institute of Electrical and Electronic Engineering, Université M’hamed Bougara, Boumerdès
- Algeria
Evolution of evaluation function, Reinforcement and TD error -FACL algorithm-
This presents the influence of the learning rate and the discount factor.
FACL and FQL reinforcement learning algorithms for a mobile robot
The objective is to obtain a behaviour that is as ‘rewarding’ as possible. Most commonly used methods of learning in the Machine Learning community are: Q-Learning and Learning and Actor-Critic Learning. Both are based on temporal differences .