Title :
Fast reinforcement learning using asymmetric probability density function
Author :
Umesako, K. ; Obayashi, M. ; Kobayashi, K.
Author_Institution :
Graduate Sch. of Sci. & Eng., Yamaguchi Univ., Ube, Japan
Abstract :
We propose an asymmetric probability density function (PDF) to select an effective action on reinforcement learning (RL). The proposed method utilizing the information of search direction enables RL to reduce the number of trials. Furthermore, the proposed method can be applied easily to various methods of RL, for example, actor-critic, stochastic gradient ascent method. The performance of our proposed method is demonstrated by computer simulations.
Keywords :
learning (artificial intelligence); neural nets; PDF; acceleration of learning; asymmetric probability density function; neural network; reinforcement learning; stochastic gradient ascent method; temporal difference method; Acceleration; Animals; Computer errors; Computer simulation; Concrete; Neural networks; Probability density function; Search methods; Stochastic processes; Unsupervised learning;
Conference_Titel :
SICE 2002. Proceedings of the 41st SICE Annual Conference
Print_ISBN :
0-7803-7631-5
DOI :
10.1109/SICE.2002.1195260