Smiley face

(646) 73two 8628


Ben Usman

I'm a PhD student in the Image and Video Computing group at Boston University advised by Prof. Kate Saenko.

Right now I am working on applications of domain adaptation in vision. More specifically, we are trying to find better formulations of statistical distances that would lead to better convergence of adaptation procedures.

Before that, I worked on BSc in applied math and physics at MIPT and MSc in applied math and computer science at MIPT and Skoltech. Back then I gained some experience in natural language processing and numerical methods such as multigrid, tensor decompositions, fast multipole method, etc.

In addition to my primary research, I am also interested in information representations that could enable reasoning over structured data, and in applying Bayesian approaches to non-Bayesian models such as neural networks. I also really enjoy studying and finding connections in all sorts of math. I also enjoy learning about patters/rules in art such as film direction, music production or design of game mechanics. And also reading about interesting engineering solutions in programming languages and systems, like the way corouties are implemented in C, or examples of elegant metaprogramming, or distributed scheduling and graph execution in Dask. I also did some backend web development in Python a while ago.