How to specify the training algorithm of an agent - Reinforcement Learning

Hi!

I would like to train agents in a multi-agent environment. Specifically, I have created two different agents (a DDPG and a TD3). The documentation describes the general algorithm used by the Train Matlab function.

I would like to know if this function customizes the algorithm based on the agent that needs to be trained, implementing a training algorithm as specified in the documentation for DDPG and TD3, for example.

Thank you all in advance!

Answer 1

[Emmanouil Tzorakoleftherakis]

'train' takes an agent object as input, so yes the algorithm will be selected depending on the agent.