Custom environment in Deep reinforcement learning

Question

Bha Pr 2020-4-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/514500-custom-environment-in-deep-reinforcement-learning

评论： SULAKSHNA DEVI 2020-5-13

I am currently trying to buid to a custom environment for the implementation of deep reinforcement learning. My considered environment has 4 states low, med, high, severe represented by 1,2,3,4 respectively and the actions to be taken are 1,2,3 and rewards are decided on the basis of context like temperature, pressure,humidity which varies with time. So how i can define my reward that changes with time in mystepfunction?

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Ari Biswas 2020-4-20

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/514500-custom-environment-in-deep-reinforcement-learning#answer_427094

One way to solve this is by introducing a property to keep track of elapsed time in your custom MATLAB environment. You can use this property to compute rewards and increment this as needed in the step function.