Tim Moon
YOU?
Author Swipe
View article: Learning Interpretable Models Through Multi-Objective Neural Architecture Search
Learning Interpretable Models Through Multi-Objective Neural Architecture Search Open
Monumental advances in deep learning have led to unprecedented achievements across various domains. While the performance of deep neural networks is indubitable, the architectural design and interpretability of such models are nontrivial. …
View article: HARM3D+NUC: A New Method for Simulating the Post-merger Phase of Binary Neutron Star Mergers with GRMHD, Tabulated EOS, and Neutrino Leakage
HARM3D+NUC: A New Method for Simulating the Post-merger Phase of Binary Neutron Star Mergers with GRMHD, Tabulated EOS, and Neutrino Leakage Open
The first binary neutron star merger has already been detected in gravitational waves. The signal was accompanied by an electromagnetic counterpart including a kilonova component powered by the decay of radioactive nuclei, as well as a sho…
View article: Enabling rapid COVID-19 small molecule drug design through scalable deep learning of generative models
Enabling rapid COVID-19 small molecule drug design through scalable deep learning of generative models Open
We improved the quality and reduced the time to produce machine learned models for use in small molecule antiviral design. Our globally asynchronous multi-level parallel training approach strong scales to all of Sierra with up to 97.7% eff…
View article: Parallelizing Training of Deep Generative Models on Massive Scientific Datasets
Parallelizing Training of Deep Generative Models on Massive Scientific Datasets Open
Training deep neural networks on large scientific data is a challenging task that requires enormous compute power, especially if no pre-trained models exist to initialize the process. We present a novel tournament method to train tradition…
View article: Improving Strong-Scaling of CNN Training by Exploiting Finer-Grained Parallelism
Improving Strong-Scaling of CNN Training by Exploiting Finer-Grained Parallelism Open
Scaling CNN training is necessary to keep up with growing datasets and reduce training time. We also see an emerging need to handle datasets with very large samples, where memory requirements for training are large. Existing training frame…
View article: CEED ECP Milestone Report: Identify initial kernels, bake-off problems (benchmarks) and miniapps
CEED ECP Milestone Report: Identify initial kernels, bake-off problems (benchmarks) and miniapps Open
ECP Milestone Report Identify initial kernels, bake-off problems (benchmarks) and miniapps WBS 1.2.5.3.04, Milestone CEED-MS6
View article: ECP Milestone Report: Identify initial kernels, bake-off problems (benchmarks) and miniapps (CEED-MS6)
ECP Milestone Report: Identify initial kernels, bake-off problems (benchmarks) and miniapps (CEED-MS6) Open
The CEED co-design center is developing a number of kernels, bake-off / benchmark problems (BPs) and mini-applications (miniapps) that capture the unique requirements of high-order finite element algorithms and use them to influence standa…
View article: Communication Quantization for Data-Parallel Training of Deep Neural Networks
Communication Quantization for Data-Parallel Training of Deep Neural Networks Open
We study data-parallel training of deep neural networks on high-performance computing infrastructure. The key problem with scaling data-parallel training is avoiding severe communication/computation imbalance. We explore quantizing gradien…
View article: Accelerating eigenvector and pseudospectra computation using blocked multi-shift triangular solves
Accelerating eigenvector and pseudospectra computation using blocked multi-shift triangular solves Open
Multi-shift triangular solves are basic linear algebra calculations with applications in eigenvector and pseudospectra computation. We propose blocked algorithms that efficiently exploit Level 3 BLAS to perform multi-shift triangular solve…