Stochastic block BFGS: squeezing more curvature out of data

Robert Gower, Donald Goldfarb, Peter Richtarik

Research output: Contribution to conferencePaperpeer-review


We propose a novel limited-memory stochastic block BFGS update for incorporating enriched curvature information in stochastic approximation methods. In our method, the estimate of the inverse Hessian matrix that is maintained by it, is updated at each iteration using a sketch of the Hessian, i.e., a randomly generated compressed form of the Hessian. We propose several sketching strategies, present a new quasi-Newton method that uses stochastic block BFGS updates combined with the variance reduction approach SVRG to compute batch stochastic gradients, and prove linear convergence of the resulting method. Numerical tests on large-scale logistic regression problems reveal that our method is more robust and substantially outperforms current state-of-the-art methods.
Original languageEnglish
Publication statusPublished - 2016
Event33rd International Conference on Machine Learning: ICML 2016 - New York, United States
Duration: 19 Jun 201624 Jun 2016


Conference33rd International Conference on Machine Learning
Abbreviated titleICML 2016
CountryUnited States
CityNew York
Internet address

Fingerprint Dive into the research topics of 'Stochastic block BFGS: squeezing more curvature out of data'. Together they form a unique fingerprint.

Cite this