English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Training deep neural networks on large scientific data is a challenging taskthat requires enormous compute power, especially if no pre-trained models existto initialize the process. We present a novel tournament method to traintraditional as well as generative adversarial networks built on LBANN, ascalable deep learning framework optimized for HPC systems. LBANN combinesmultiple levels of parallelism and exploits some of the worlds largestsupercomputers. We demonstrate our framework by creating a complex predictivemodel based on multi-variate data from high-energy-density physics containinghundreds of millions of images and hundreds of millions of scalar valuesderived from tens of millions of simulations of inertial confinement fusion.Our approach combines an HPC workflow and extends LBANN with optimized dataingestion and the new tournament-style training algorithm to produce a scalableneural network architecture using a CORAL-class supercomputer. Experimentalresults show that 64 trainers (1024 GPUs) achieve a speedup of 70.2 over asingle trainer (16 GPUs) baseline, and an effective 109% parallel efficiency.

Parallelizing Training of Deep Generative Models on Massive Scientific Datasets