AI Algorithms for Science at the Edge

Accelerating scientific discovery through data reduction and processing

The ways scientific data are compressed and aggregated have important implications for how scientific research is done. This project, titled “Real-Time Data Reduction Codesign at the Extreme Edge for Science,” aims to accelerate scientific discovery by bringing intelligent ML-based data reduction and processing as close as possible to the data source.

We concentrate on powerful, specialized compute hardware at the extreme edge such as FPGAs, ASICs, and systems on-chip, which form the initial processing layers of many modern experiments.

Our goals are to:

• Develop reliable AI algorithms for science at the edge;
• Develop codesign tools to build efficient implementations of those algorithms in hardware; and
• Enable rapid exploration for domain scientists and system designers with an accessible tool flow.

To demonstrate our newly developed techniques, we are applying them to two example use cases:

• Real-time trigger systems at the CERN Large Hadron Collider, an example which offers a complex, geometry constrained progressive data reduction flow, and
• Processing data streams in transmission electron microscopy, which involves fast feature extraction that must be capable of performing across a multitude of samples and experiments.

Project Team

Outcomes

About