Input normalization using a reinforcement learning DQN agent
6 ビュー (過去 30 日間)
古いコメントを表示
Margarita Cabrera
2021 年 3 月 26 日
Hi all
I have built a DQN agent to solv a custom reinforcement problem.
Following mathworks examples I have no used any kind of normalization applied to the critic input.
In fact, as far as I could check, all examples of RL that use a DNN to create an actor or a critic especify 'Normalization', 'none' at the input layers of the Actor and Critic.
My question is, is it possible to use a normalization as for instance "zscore" at the input layers of a critic or of an actor when these are based on a DNN?
I'have tried to applied zscore normalization, but then, the agent does not work.
thanks
0 件のコメント
採用された回答
Emmanouil Tzorakoleftherakis
2021 年 3 月 26 日
Hello,
Normalization through the input layers is not supported for RL training. As a workaround, you can scale the observations rewards on the environment side.
2 件のコメント
H. M.
2022 年 12 月 3 日
@Emmanouil Tzorakoleftherakis
Could you explain more, the way you mentioned about normalization. I want to do it, but I can't figure it out.
Regards
その他の回答 (0 件)
参考
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!