Skip to yearly menu bar Skip to main content


Poster

Error Feedback Fixes SignSGD and other Gradient Compression Schemes

Sai Praneeth Reddy Karimireddy · Quentin Rebjock · Sebastian Stich · Martin Jaggi

Pacific Ballroom #96

Keywords: [ Convex Optimization ] [ Large Scale Learning and Big Data ] [ Non-convex Optimization ] [ Optimization ]


Abstract:

Sign-based algorithms (e.g. signSGD) have been proposed as a biased gradient compression technique to alleviate the communication bottleneck in training large neural networks across multiple workers. We show simple convex counter-examples where signSGD does not converge to the optimum. Further, even when it does converge, signSGD may generalize poorly when compared with SGD. These issues arise because of the biased nature of the sign compression operator.

We then show that using error-feedback, i.e. incorporating the error made by the compression operator into the next step, overcomes these issues. We prove that our algorithm (EF-SGD) with arbitrary compression operator achieves the same rate of convergence as SGD without any additional assumptions. Thus EF-SGD achieves gradient compression for free. Our experiments thoroughly substantiate the theory.

Live content is unavailable. Log in and register to view live content