Timezone: »

WoodFisher: Efficient Second-Order Approximation for Neural Network Compression
Sidak Pal Singh · Dan Alistarh

Tue Dec 08 09:00 AM -- 11:00 AM (PST) @ Poster Session 1 #286

Second-order information, in the form of Hessian- or Inverse-Hessian-vector products, is a fundamental tool for solving optimization problems. Recently, there has been significant interest in utilizing this information in the context of deep neural networks; however, relatively little is known about the quality of existing approximations in this context. Our work considers this question, examines the accuracy of existing approaches, and proposes a method called WoodFisher to compute a faithful and efficient estimate of the inverse Hessian.

Our main application is to neural network compression, where we build on the classic Optimal Brain Damage/Surgeon framework. We demonstrate that WoodFisher significantly outperforms popular state-of-the-art methods for one-shot pruning. Further, even when iterative, gradual pruning is allowed, our method results in a gain in test accuracy over the state-of-the-art approaches for popular image classification datasets such as ImageNet ILSVRC. Further, we show how our method can be extended to take into account first-order information, and illustrate its ability to automatically set layer-wise pruning thresholds, or perform compression in the limited-data regime.

Author Information

Sidak Pal Singh (ETH Zürich)
Dan Alistarh (IST Austria & Neural Magic Inc.)

More from the Same Authors