For example, a RL algorithm that gains points when a rat presses a lever and loses points when it dispenses a pellet, water, treat, and/or sugar water. After a few days of controlling the rewards given to a rat, all rewards are stopped and the longer/more times the rat presses the lever before giving up, the higher the score.

This would be a situation in which both the input and the outputs are discrete with very low data density over time and with outputs having very long-term affects on the environment.

What kind of RL architecture would be appropriate here?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.