Fujitsu Laboratories Ltd. has developed technology to improve the speed of deep learning software. This tech has now achieved the world’s highest speed when the time required for machine learning. It was measured using the AI Bridging Cloud Infrastructure (ABCI) system, deployed by Fujitsu Limited for the National Institute of Advanced Industrial Science and Technology (AIST).

Deep learning software is an aspect of Artificial Intelligence (AI) that is concerned with how computers learn. Computers approach learning how human beings use to obtain certain knowledge instead of what humans program it to do. Speech recognition, sound, text, and image recognition are some examples of deep learning software.

ResNet-50(1) is a deep neural network for image recognition. It’s generally used as a benchmark to measure deep learning processing speed, comparing training times using image data from the ImageNet Large Scale Visual Recognition Challenge 2012 (ILSVRC2012), a contest of image recognition accuracy.

Based on the technology Fujitsu Laboratories has cultivated over its HPC development, the company has now developed a technology to expand computation volume per GPU without compromising training accuracy.  This newly developed technology was applied to open source deep learning software. It took 2,048 GPUs in the ABCI system and measured for this benchmark.

Fujitsu Laboratories confirmed that it beats the previous speed record by more than 30 seconds(2), completing the training in 74.7 seconds, the world’s highest speed(2).

(1) ResNet-50 A high performance image recognition deep neural network developed by Microsoft. Laboratories’ investigation.
(2) Beat the previous speed record by more than 30 seconds, completing the training in 74.7 seconds, the world’s highest speed as of March 26, 2019, as confirmed by Fujitsu