Hi Brian,

I haven’t experimented too systematically with larger grid sizes, but I think the intuition that the convolutions might not be fine-grained enough is probably right. It is also my experience that it takes longer to train generally for larger grid sizes, simply as a matter of there being a larger state space for DQN to learn Q values for. If you happen to find parameters of the network that allow for good performance on a 10x10 grid though, please don’t hesitate to share here, as I would be interested myself.

PhD. Interests include Deep (Reinforcement) Learning, Computational Neuroscience, and Phenomenology.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store