Panagiotis (Panos) Toulis
Harvard University,
Department of Statistics
ptoulis at fas dot harvard dot edu

Homepage Short bio Publications Code Miscellaneous

This page provides more info about the code and simulations presented in (Toulis et. al., 2014) on implicit stochastic gradient descent (SGD) for large Generalized Linear Models (GLM). The implicit method is a modification of the typical SGD algorithm that has very attractive properties: (i) asymptotically it has similar guarantees for bias/variance and (ii) in small samples is more stable and more robust to misspecifications of the learning rate and/or outliers.

The following are the main parts of this code: