High Dimensional Data

Опубликовано 12 августа 2016, 2:21

Match the applications to the theorems: (i) Find the variance of traffic volumes in a large network presented as streaming data. (ii) Estimate failure probabilities in a complex systems with many parts. (iii) Group customers into clusters based on what they bought. (a) Projecting high dimensional space to a random low dimensional space scales each vector's length by (roughly) the same factor. (b) A random walk in a high dimensional convex set converges rather fast. (c) Given data points, we can find their best-fit subspace fast. While the theorems are precise, the talk will deal with applications at a high level. Other theorems/applications may be discussed.