Learning with reinforcement learning

My friend and I explored reinforcement learning by implementing three important algorithms: value iteration, Q learning and deep Q learning. These go from more traditional reinforcement learning methods towards deep learning. We demonstrated these algorithm through building a gridworld environment in JavaScript, after which I went onto experimenting with these algorithms on more complex environments (i.e. playing ping pong).

Reinforcement learning is extremely important for my field of study in human-computer interaction and tangible media, yet I find a lack of presence in these branches of AI. If this is something you're interested in pursuing, please do get in touch.

For technical details and collaborations, please see my Github repo above.

Handcrafted in New York City, design & code by
Dora Jambor