This website is used to document and publicise ongoing work on my Evolving Atari project, the purpose of which is to extend research in evolutionary reinforcement learning.

The current focus of my work is on a hybrid (deep learning + neuroevolution) learning method for creating general Atari game playing agents, that harnesses the representational learning power of deep neural networks, and the creativity and exploration of evolutionary algorithms.

If you’re interested in discussing any of my research or related ideas, please reach out to me.