Scientists slash computations for deep learning

ronjor · Jun 3, 2017

June 1, 2017

Rice University computer scientists have adapted a widely used technique for rapid data lookup to slash the amount of computation—and thus energy and time—required for deep learning, a computationally intense form of machine learning.
Click to expand...

guest · Aug 1, 2018

This is your four-minute warning: Boffins train ImageNet-based AI classifier in just 240s
Helps to have 1k GPUs and a relaxed view of accuracy
August 01, 2018
https://www.theregister.co.uk/2018/08/01/imagenet_training_four_minutes/

Faster is always better in AI, although it comes at a price. As researchers strive to train their neural networks at breakneck speeds, the accuracy of their software falls.

Thus, teaching machine-learning code at high speed, with the resulting network able to make precise and correct decisions, is a vital goal.

To that end, a group of researchers at TenCent, one of the biggest tech conglomerates in China, and Hong Kong Baptist University have trained AlexNet – a popular neural network built on ImageNet, a dataset often used for image-classification systems – in just four minutes. The record previously was 11 minutes, held by UC Berkeley et al boffins, as far as we're aware.
Tricks of the trade
Thus, the TenCent and Hong Kong uni crew came up with a communications technique called “tensor fusion.” [...] multiple small size tensors are packaged together to reduce the amount of information that has to be transferred, thus reducing latency and increasing throughput. And increasing throughput reduces the amount of time taken to train the model.
Click to expand...

Log in or Sign up

Scientists slash computations for deep learning

ronjor Global Moderator

guest Guest

Log in or Sign up

Scientists slash computations for deep learning

ronjor Global Moderator

guest Guest

Useful Searches