MATLAB: Visualize Progress in Reinforcement Learning Toolbox

Reinforcement Learning Toolboxvisualization

Hello everyone

Do you know whether it is possible to store an agent's progress within the reinforcement learning toolbox? What I would like to do is to have access to the agent's states in the course of the training, such that I can show how the agent behaved at different points in its learning phase. Is there a built-in functionality like this, or should I try and store every agent individually, which potentially is a lot of data.

Cheers

Best Answer

This is not possible out of the box, but you could implement something like this by setting a counter and saving the current version of the agent as needed. You could do it from the step/reset function

Related Solutions

MATLAB: How to extract a trained RL Agent’s network’s weights and biases

You can get the parameters from the trained's critic representation for DQN agent. In MATLAB R2020a, see getLearnableParameters and getCritic functions (function name changes a bit since R2019b). You can follow similar steps to get the actor's parameters from actor-based agent like DDPG or PPO.

critic = getCritic(agent);
criticParams = getLearnableParameters(critic);

MATLAB: How to visualize episode behaviour with the reinforcement learning toolbox

Hello,

To create a custom MATLAB environment, use the template that pops up after running

rlCreateEnvTemplate('myenv')

In this template there are two methods that can be used for visualization, "plot", and "envUpdatedCallback" (it is called from within "plot"). Use "plot" to create the basic stationary parts of your visualization, and "envUpdatedCallback" to update the coordinates of the moving parts based on your states.

Best Answer

Related Solutions

MATLAB: How to extract a trained RL Agent’s network’s weights and biases

MATLAB: How to visualize episode behaviour with the reinforcement learning toolbox

Related Question