My main research interests are in optimization, neural networks and reinforcement learning. Currently, I focus on sketching-based compression methods, the understanding of neural networks through a convex perspective, and reinforcement and imitation learning algorithms for safety-critical applications. Part of my work was done in collaboration with the Autonomous Systems Lab at Stanford University. I also had the pleasure to be a visiting researcher at UC Berkeley under Laurent El Ghaoui, and at Facebook AI Research under Mohammad Ghavamzadeh.
My Google Scholar and Github pages.
Email: lacotte (at) stanford (dot) edu