Continuous adaptive critic designs
Tools
Tools
Tools
Hanselmann, T., Noakes, L. and Zaknich, A. (2005) Continuous adaptive critic designs. In: International Joint Conference on Neural Networks, IJCNN 2005, 31 July - 4 August, Montreal, Canada.
![[img]](http://researchrepository.murdoch.edu.au/style/images/fileicons/application_pdf.png)
| PDF - Published Version Download (1397kB) | Preview |
Link to Published Version: http://dx.doi.org/10.1109/IJCNN.2005.1556403
*Subscription may be required
*Subscription may be required
Abstract
A continuous formulation of an adaptive critic design (ACD) is investigated. Connections to the discrete case are made, where backpropagation through time (BPTT) and realtime recurrent learning (RTRL) are prevalent. A second order actor adaptation, based on Newton's method, is established for fast actor convergence. Also, a fast critic update for concurrent actor-critic training is outlined that keeps the Bellman optimality correct to first order approximation after actor changes.
| Publication Type: | Conference Paper |
|---|---|
| Murdoch Affiliation: | School of Engineering |
| Publisher: | IEEE |
| Copyright: | © 2005 IEEE |
| Notes: | In Proceedings of the International Joint Conference on Neural Networks, 2005. IJCNN '05, Pages 3001-3006. |
| URI: | http://researchrepository.murdoch.edu.au/id/eprint/11935 |
| Item Control Page |
