Guest
Data Fest Online 2020 Math Optimization Track In this talk we discuss how to compute an integral (or find an expected value of a function) in a high dimensional space. To this end, we first discuss an importance sampling technique which stands for a Monte-Carlo type approximation to the integral by changing the probability measure. Secondly, we describe various importance sampling methods from the (convex and non-convex) optimization perspective, and explore the importance sampling extensions and limitations. Later on, we consider applications of the importance sampling to Beyesian statistics, stochastic optimization, and optimal control. At the end of the talk, I explain how the importance sampling helps to predict, detect, and mitigate energy system’s blackouts. Register and get access to the tracks: Join the community:
