Aravind Rajeswaran (Rajesh)

Aravind Rajeswaran 

Machine Learning PhD student,
University of Washington Seattle



Short Bio

I am a PhD student at UW working on machine learning, optimization, and deep RL with Sham Kakade, Emo Todorov, and Sergey Levine. I'm also interested in aspects of computer vision, and learning predictive models of dynamical systems – also called intuitive physics and system ID. Previously, I was a student at IIT Madras with Balaraman Ravindran, and received the best undergraduate thesis award. Before switching focus to AI, I worked in statistical physics of complex networks with Sitabhra Sinha.

My research interests are broad and include many areas of applied math like: optimization, (deep/machine) learning, optimal control, statistical physics, and dynamical systems. Current research directions are:

  • Policy optimization using natural gradients and Newton-CG

  • Reconciling policy search with trajectory optimization (specifically, iLQG and DDP)

  • Learning intuitive models of physics from video data