Machine learning masters massive data sets: Algorithm breaks the exabyte barrier

https://techxplore.com/news/2023-09-machine-masters-massive-algorithm-exabyte.html

"dividing into manageable batches that don't choke computer hardware... efficient for laptops through supercomputers... out-of-memory' implementation of non-negative matrix factorization method allowing factorizing larger data sets, processes 1 at a time, cycling them in and out of memory... processed 340-terabyte dense matrix and 11-exabyte sparse matrix using 25k GPUs... cancer research, satellite imagery, social media networks, national security science and earthquake research"

Related:

The catch-22s of reservoir computing: Researchers find overlooked weakness in powerful machine learning tool
https://phys.org/news/2023-09-catch-22s-reservoir-overlooked-weakness-powerful.html

A Hebbian memory that achieves human-like results on sequential processing tasks
https://techxplore.com/news/2023-10-hebbian-memory-human-like-results-sequential.html

A machine learning–based tool to model phase-change memory materials
https://techxplore.com/news/2023-11-machine-learningbased-tool-phase-change-memory.html

How symmetry can come to the aid of machine learning
https://techxplore.com/news/2024-02-symmetry-aid-machine.html