dqn-hfo
dqn-hfo copied to clipboard
Try fewer state features?
Hi @mhauskn. Thanks a lot for your work. I’m curious about one thing. By HFO manual, there are 58(+1) state features for only-one-offense-player setting, but these features are over complete, as written in the manual. For example, we only need 2 features to locate the position of one player, but there are much more Landmark and Proximity features. Have you considered or tried using fewer state features?
Hi Hongjie, you are correct - the features in HFO are designed to be overcomplete, because neural networks are able to handle redundant information gracefully. I have not tried using fewer state features, but I suspect that it would be entirely possible.
On Tue, Aug 7, 2018 at 3:15 PM Hongjie [email protected] wrote:
Hi @mhauskn https://github.com/mhauskn. Thanks a lot for your work. I’m curious about one thing. By HFO manual, there are 58(+1) state features for only-one-offense-player setting, but these features are over complete, as written in the manual. For example, we only need 2 features to locate the position of one player, but there are much more Landmark and Proximity features. Have you considered or tried using fewer state features?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/mhauskn/dqn-hfo/issues/35, or mute the thread https://github.com/notifications/unsubscribe-auth/AABNOeaeTBD2opkdU4l3E4cImpNXqD44ks5uOhGHgaJpZM4Vy-Wt .