167-PFlops Deep Learning for Electron Microscopy: From Learning Physics to Atomic Manipulation
Authors:
- Robert M. PattonRPRobert M. Patton
- J. Travis JohnstonJJJ. Travis Johnston
- Steven R. YoungSYSteven R. Young
- Catherine D. SchumanCSCatherine D. Schuman
- Don D. MarchDMDon D. March
- Thomas E. PotokTPThomas E. Potok
- Derek C. RoseDRDerek C. Rose
- Seung-Hwan LimSLSeung-Hwan Lim
- Thomas P. KarnowskiTKThomas P. Karnowski
- Maxim A. ZiatdinovMZMaxim A. Ziatdinov
- Sergei V. KalininSKSergei V. Kalinin
- 1 November 2018
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 638-648
Abstract
An artificial intelligence system called MENNDL, which used 25,200 NVIDIA Volta GPUs on Oak Ridge National Laboratory's Summit machine, automatically designed an optimal deep learning network in order to extract structural information from raw atomic-resolution microscopy data. In a few hours, MENNDL creates and evaluates millions of networks using a scalable, parallel, asynchronous genetic algorithm augmented with a support vector machine to automatically find a superior deep learning network topology and hyper-parameter set than a human expert can find in months. For the application of electron microscopy, the system furthers the goal of improving our understanding of the electron-beam-matter interactions and real-time image-based feedback, which enables a huge step beyond human capacity towards nanofabricating materials automatically. MENNDL has been scaled to the 4,200 available nodes of Summit achieving a measured 152.5 PFlops, with an estimated sustained performance of 167 PFlops when the entire machine is available.Keywords
This publication has 19 references indexed in Scilit:
- Fire up the atom forgeNature, 2016
- Practical Efficiency of Asynchronous Stochastic Gradient Descent2016
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Direct Observation of Atomic Dynamics and Silicon Doping at a Topological Defect in GrapheneAngewandte Chemie, 2014
- Flexible metallic nanowires with self-adaptive contacts to semiconducting transition-metal dichalcogenide monolayersNature Nanotechnology, 2014
- Local crystallography analysis for atomically resolved scanning tunneling microscopy imagesNanotechnology, 2013
- Practical Recommendations for Gradient-Based Training of Deep Architectures2012
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998
- Support-vector networksMachine Learning, 1995
- A Stochastic Approximation MethodThe Annals of Mathematical Statistics, 1951