Monday, May 16, 2016 - 3:10pm - 4:00pm
David Woodruff (IBM Research Division)
We consider an illustrative problem in distributed machine learning - computing a low rank approximation in the arbitrary partition model. In this model each of s servers holds an n x d matrix A^i, where each entry is an O(log nd)-bit word, and we let A = A^1 + ... + A^s. We would like each server to output the same k x d matrix V, so that V is an approximation to the top k principal components of A, in the sense that projecting onto V provides a (1+eps)-approximate low rank approximation.
Tuesday, April 21, 2015 - 11:30am - 12:30pm
Viktoria Averina (Boston Scientific), Jeff Sachs (Merck & Co., Inc.)
Viktoria Averina: Algorithm Development for Medical Devices

Implantable cardiac devices such as pacemakers and defibrillators can track and analyze patient’s physiological data. I will showcase key steps of research and development of device-based respiratory monitoring: how one measures patient’s respiratory function, why one would want to track it, and what challenges come up during the implementation.
Subscribe to RSS - communication