Interspeech 2014

Overview

We show empirically that in SGD training of deep neural networks, one can, at no or nearly no loss of accuracy, quantize the gradients aggressively—to but one bit per value—if the quantization error is carried…

Publisher page information

Source page title

1-Bit Stochastic Gradient Descent and Application to Data-Parallel Distributed Training of Speech DNNs - Microsoft Research

Site name (reported)

Microsoft Research

Declared page language

en-US

Observed source URL

https://www.microsoft.com/en-us/research/publication/1-bit-stochastic-gradient-descent-and-application-to-data-parallel-distributed-training-of-speech-dnns/ ↗

Publisher canonical URL

https://www.microsoft.com/en-us/research/publication/1-bit-stochastic-gradient-descent-and-application-to-data-parallel-distributed-training-of-speech-dnns/ ↗

Source metadata observed

Oct 11, 2026

Recorded from the source page’s published HTML metadata. Publisher page language can differ from this directory’s English interface.

Source links & provenance

Discovery references

• GitHub · Zhen-Dong/Awesome-Quantization-Papers ↗

References explain how this resource entered the collection. They do not represent an endorsement.