Busy GPUs: Sampling and pipelining method speeds up deep learning on large graphs

Retrieved on: 2022-11-29 21:46:34

Tags for this article:

Artificial intelligence

Machine learning

Integrated circuits

GPGPU

Artificial neural networks

Graphics hardware

Emerging technologies

Coprocessors

Graphics processing unit

PyTorch

Deep learning

Central processing unit

Tim Kaler

Jie Chen

Click the tags to see associated articles and topics

Busy GPUs: Sampling and pipelining method speeds up deep learning on large graphs. View article details on HISWAI: https://news.mit.edu/2022/sampling-pipelining-method-speeds-deep-learning-large-graphs-1129

Excerpt

A machine-learning technique called SALIENT addresses key bottlenecks in computation with graph neural networks by optimizing usage of the ...

Article found on: news.mit.edu

View Original Article