Steven’s Homepage

My research focus is on applying deep reinforcement learning to partially observable Markov decision processes (POMDPs) using recurrent models. In other words, I am working to endow intelligent agents with long-term memory capabilities, enabling reasoning under the uncertainty present in the real world.

Under a more philosophical lens, I believe that long-term memory is necessary for truly intelligent agents. Our personality, our decisions – who we ultimately are as people is conditioned on a lifetime of individual experiences. How can you learn from your successes and failures, if you can’t remember them?