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 real...Show MoreMetadata
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.
Date of Conference: 31 July 2005 - 04 August 2005
Date Added to IEEE Xplore: 27 December 2005
Print ISBN:0-7803-9048-2