Almost all phenomena in science and engineering are inherently multiscale, and many require exploration across orders of magnitude in both space and time. Solving such problems at the finest scales is computationally prohibitive, whereas the affordable coarser scales do not provide the required fidelity to answer relevant questions.

Understanding such behavior requires coupling across scales adaptively—a task commonly performed using computational science technique.

CASC is leading the development of new scientific machine learning (SciML) technology to explore such scientific behavior across scales through intelligent automation and decision support for complex systems. The cross-cutting nature of such technology makes it widely applicable to many scientific domains. Combined with the modern HPC resources, SciML will provide a new paradigm for large scientific simulations.

The early adoption of CASC’s SciML technology was demonstrated on the “Pilot 2” project for studying cancer initiation mechanisms.

Grid of images generated by AttGAN

Attribute-Guided Adversarial Training

More research accepted at AAAI 2021 enables DNNs to be robust against a wide range of naturally occurring perturbations.

screen shot of part of the poster PDF

Advancing ML for Mission-Critical Applications

As our booth and poster at the 2020 LLNL Computing Virtual Expo explained, our work is a partnership between ML, HPC, and Lab programs.

Harsh Bhatia

Meet a Data Viz Expert

Harsh Bhatia, PhD, was a Lawrence Graduate Scholar and an LLNL postdoctoral researcher before joining CASC full time in 2017.

Rushil Anirudh in front of a flowering tree

Meet an ML Expert

Rushil Anirudh, PhD, reconstructs computed tomography images using convolutional neural networks from partially scanned objects.