How to implement LSTM layers in MATLAB's DDPG agent

I am trying to include LSTM layers in my DDPG agent to compensate for modeling in a non-markovian environment. I would assume I would add them in both the actor and critic networks and was wondering if MATLAB's DDPG algorithm would automatically handle the change to R-DDPG and if not, I would appreciate insight on the process of modifying the current algorithm (on the MATLAB side of modifying the algorithm to what I want).

Thank you

Answer 1

[Emmanouil Tzorakoleftherakis]

Hello,

You can use lstm layers directly in both actors and critics and the built-in DDPG agent will handle the rest. Take a look here:

https://www.mathworks.com/help/reinforcement-learning/ref/rl.agent.rlddpgagent.html#mw_692a723c-8e84-4d86-b43d-248fb989e57c