Machine intelligence. Personal site of Cosmo Harrigan.

Contact information and profile:
cosmoharrigan@gmail.com
cosmoh@uw.edu
Follow @cosmoharrigan

LinkedIn profile (Contains a list of the companies I founded & various projects)
GitHub profile

I am interested in understanding the nature of cognition and building generally intelligent agents, especially
approximations of universal artificial intelligence and new architectures for deep reinforcement learning.

In addition to being a researcher, I am an internet entrepreneur. Since 2006, I have founded several technology
companies. I am on the board of The Center for Media and Democracy. I graduated from the University of
Washington with a degree in Applied and Computational Mathematical Sciences.

Some recent material:
Survey of Deep Learning for Artificial General Intelligence (Invited talk, AGI 2016)
Overview of Universal AI: Reinforcement Learning via AIXI Approximation
Deep Reinforcement Learning with Regularized Convolutional Neural Fitted Q Iteration (Code)
Neuroevolution as a direct policy search method
Measuring Understanding Using Algorithmic Complexity
Guiding Probabilistic Logical Inference with Nonlinear Dynamical Attention Allocation (Springer LNCS, 2014)