Reinaldo A. C. Bianchi,  Anna H. R. Costa. 

This work presents a new class of algorithms that allows the use of heuristics to speed up Reinforcement Learning (RL) algorithms. This class of algorithms, called “Heuristically Accelerated Learning” (HAL) is modeled using a convenient mathematical formalism known as Markov Decision Processes. A heuristic function H influences the agent’s choice of the actions to perform during the learning process. As the heuristic is used only when choosing the action to be taken, the RL algorithm operation is not modified and many proprieties of the RL algorithms are preserved. Experiments that use HAL algorithms to solve problems in a number of domains – including robotic navigation – are presented. Experimental results allow to conclude that even a very simple heuristic results in a significant performance increase in the learning rate of the reinforcement learning algorithm.

http://www.lbd.dcc.ufmg.br/colecoes/ctd/2005/015.pdf

Processando consulta.