Q(λ)-learning algorithm question
- From: kartoun@xxxxxxxxx
- Date: Sun, 08 Jan 2006 00:48:51 GMT
Hi,
For Watkin's Q(\lambda)-learning algorithm; why would someone
prefer higher values of \lambda than lower? My experiments both in
simulation and on real robot show that \lambda=0.5 gives a solution
that tends to converge. For lower and higher \lambda values (0.25
and 0.75) learning diverges. Is it task specific?
Thanks,
Uri.
http://www.compactech.com/kartoun
[ comp.ai is moderated. To submit, just post and be patient, or if ]
[ that fails mail your article to <comp-ai@xxxxxxxxxxxxxxxxxx>, and ]
[ ask your news administrator to fix the problems with your system. ]
.
- Prev by Date: Proofs of Reinforcement Learning Algorithms
- Next by Date: Algorithm/Theory help: Patterns, comparing, calculating 'distance' between?
- Previous by thread: Proofs of Reinforcement Learning Algorithms
- Next by thread: Algorithm/Theory help: Patterns, comparing, calculating 'distance' between?
- Index(es):
Relevant Pages
|
Loading
