An Introduction to Random Optimization Problems

Abstract: Optimization lies at the heart of many computation tasks, such as training a neural network. These optimization problems arise from stochastic data and are often high-dimensional and non-convex. In this talk, I will overview recently developed algorithms for one random optimization problem, the mixed p-spin glass, and how they generalize to other problems. If time permits I will discuss recent lower bounds suggesting that these algorithms are optimal.

Computer scienceMathematicsPhysics

Audience: researchers in the topic

MIT Simple Person's Applied Mathematics Seminar