Title :
Motor primitive discovery
Author :
Thomas, P.S. ; Barto, A.G.
Author_Institution :
Dept. of Comput. Sci., Univ. of Massachusetts Amherst, Amherst, MA, USA
Abstract :
We present a method for autonomous on-line discovery of motor primitives for Markov decision processes with high-dimensional continuous action spaces. These biologically-inspired motor primitives require overhead to compute but form a compressed representation of the action set that allows for improved performance on subsequent learning tasks that have similar dynamics.
Keywords :
Markov processes; learning (artificial intelligence); neurophysiology; Markov decision process; autonomous online discovery; compressed representation; continuous action spaces; motor primitive discovery; Aerospace electronics; Animals; Joints; Muscles; Optimization; Search problems; Vectors;
Conference_Titel :
Development and Learning and Epigenetic Robotics (ICDL), 2012 IEEE International Conference on
Conference_Location :
San Diego, CA
Print_ISBN :
978-1-4673-4964-2
Electronic_ISBN :
978-1-4673-4963-5
DOI :
10.1109/DevLrn.2012.6400845
