Brigham S. Anderson

Homepage

Currently a postdoc at CMU, I work with Prof. Andrew Moore in the  The Auton Lab.

Research Interests

My recent research has had the major themes of fast inference and active inference in graphical models. Inference in graphical models is a fascinating problem with broad scope, since 1) most machine learning tasks can be posed as inference on probabilistic models, and 2) most probabilistic models can be described in graphical terms.

Fast inference is clearly useful for real-world applications. The promise of active inference can be expressed intuitively as introducing curiosity into machine learning. This allows for closed-loop learning, in which the machine is more autonomous, and learns faster because it can actively seek data to improve its understanding of the world. This has widespread application in web services such as recommendations, personalization, and diagnosis.