Different results from RL TD3 in MATLAB 2023a and 2024a

19 ビュー (過去 30 日間)
Azadeh
Azadeh 2024 年 5 月 3 日
回答済み: aditi bagora 2024 年 5 月 6 日
I am running the same Simulink model with the same RL agent and settings, but I get different results running it in MATLAB 2023a and MATLAB 2024a. while reapeating in the same MATLAB gives me the same results. Does anyone have any idea what is the reason and how can I fix it?

回答 (1 件)

aditi bagora
aditi bagora 2024 年 5 月 6 日
Hello Azadeh,
I understand that you are running the Simulink model with same RL agent and settings in different versions of MATLAB i.e. R2023a and R2024a and you are getting different results.
The variation in the output could arise because the random seed for the RL agent or environment has not been specifically set or controlled. MATLAB versions may approach default seeding in unique manners, leading to different results.
In order to fix the issue, you can explicitly set the random seed to ensure consistent initialization across versions.
Please refer to the following documentation link for more information.

カテゴリ

Help Center および File ExchangeReinforcement Learning についてさらに検索

タグ

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by