Multinode Evolutionary Neural Networks for Deep Learning (MENNDL) Wins R&D100 Award

MENNDL, developed by CSMD researchers Robert Patton (PI), Thomas Karnowski, Seung-Hwan Lim, Thomas Potok, Derek Rose and Steven Young, was announced as a winner at this year’s R&D 100 Awards ceremony.  The award is presented by R&D Magazine to recognize the top 100 revolutionary technologies of the past year.


An artificial intelligence system called MENNDL, which used 18,000 NVIDIA Volta GPUs on ORNL’s Summit, automatically designed an optimal deep learning network in order to extract structural information from raw atomic-resolution microscopy data.

MENNDL has been scaled to the 3,000 available nodes of Summit achieving a measured 98.6 PFlops, with an estimated sustained performance of 151 PFlops no the entire machine. 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. In fact, MENNDL evaluated 2 million networks on 3,000 nodes of Summit in 4 just hours. 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.

This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research, Robinson Pino, program manager, under contract number DE-AC05-00OR22725.