CORR
1998
Springer
13 years 11 months ago
1998
Springer
A new training algorithm is presented for delayed reinforcement learning problems that does not assume the existence of a critic model and employs the polytope optimization algorit...