Particles, programs, and machines that have ideas: applying statistical physics to build the next generation of machine learning technology in an MIT startup company

Abstract: The current generation of machine learning learns input-output relationships, but the internal structure of the learned function is a black box. The next generation of machine learning will be composed of modules that can be learned rather independently, edited, shared, and automatically re-combined to explain new data in order to reason by analogy. At Ben Vigoda’s startup we are doing the research to develop this technology while at the same time developing a successful business to make smarter AI assistants.

Ben Vigoda will talk about his experience over the past 15 years since graduating with his PhD from MIT, creating startups that pursue hard science challenges in machine learning while also creating successful businesses, and having fun doing it.

Jan/23 Wed 1:30PM-2:30PM 6-120

Ben Vigoda - Physics Alum