Michael Chang is a 2nd year Ph.D. student at U.C. Berkeley advised by Sergey Levine and Tom Griffiths. At MIT he worked in Josh Tenenbaum’s lab on representation learning and intuitive physics. He has also interned with Jürgen Schmidhuber at IDSIA and Honglak Lee at the University of Michigan. He is currently interested in building learners that exploit compositionality in their internal representations as well as their internal computations in the hope that such compositionality would enable better extrapolation and knowledge transfer to future problems the learner encounters. Please see http://mbchang.github.io/ for more details.