GreedyGQ(lambda) with Cart pole balancing problem

33 views

Skip to first unread message

Sam

unread,

Mar 18, 2012, 10:01:58 PM3/18/12

to RLPark

Hi All,

I have successful results using GreedyGQ(lambda) with Mountain car
benchmark (avg. steps around 106). I am trying to get the same results
as Sarsa(lambda) for Cart pole benchmark with GreedyGQ(lambda). So far
my attempts are unsatisfactory. I am wondering whether RLpark contains
results for Cart pole benchmark with GreedyGQ(lambda).

Please let me know.

Thank you!

Sam

Thomas Degris

unread,

Mar 19, 2012, 4:18:55 AM3/19/12

to github...@googlegroups.com, Hamid Reza Maei

Hello Sam,

I never tested GreedyGQ(lambda) on cart pole. Maybe you can try with the second step-size in GreedyGQ set to 0, or find in the literature some examples of qlearning on cart pole to give you a starting point for the parameters.