Dual Averaging Lecture

[Kevin Waugh]

Hi all,

Tomorrow I will be giving the advanced topic lecture on Dual Averaging.  It can be thought of as subsuming the subgradient method and is due to Nesterov.  I promise that, unlike his accelerated methods, that there is actually something both understandable and insightful about it. As a bonus, I'll show you how to use it to train a classifier when you cannot fit your data set on a single machine (distributed optimization).

See you then,

 - Kevin