Timothy Chu (CMU)
Tuesday, April 6, 2021
Virtual via Bluejeans - 11:00 am
Title: Manhattan Distances, Kernels, and Metric Transforms
Abstract: Take n points in any dimension, compute the Manhattan distance between each pair of points, and take the cube root of each distance. The resulting distances are guaranteed to be a Manhattan distance! Can you prove why?
In this talk, we prove this result and find a full classification of functions that transform Manhattan distances to Manhattan distances. This work involves group symmetry, matrix eigenvalues, and machine learning kernels. No special background will be needed to enjoy this talk!
Joint work with Gary Miller, Shyam Narayanan, and Mark Sellke.
----------------------------------
Videos of recent talks are available at: http://arc.gatech.edu/node/121
Click here to subscribe to the seminar email list: arc-colloq@Klauscc.gatech.edu
]]>