Purdue University research team included in NSF $15M AI award
2021-09-28
As scientific data sets become progressively larger, algorithms to process this data quickly become proportionally more complex. In response, artificial intelligence (AI) has emerged as a solution to efficiently analyze these massive data sets. New processor types, such as graphics processing units (GPUs) and field-programmable gate arrays (FPGAs), allow AI algorithms to be greatly accelerated. The combination of AI and new processor types is leading to a revolution in the way data is analyzed, minimizing the time needed to perform the most advanced of analysis and allowing challenges brought about by the omnipresent onslaught of data to be addressed.
To harness these new developments and leverage them for the advancement of science, the newly-created $15 million NSF HDR Institute of Accelerated AI Algorithms for Data-Driven Discovery (A3D3) aims to incorporate AI algorithms with new processors such that they can analyze unprecedented data sets.
A team of researchers at Purdue University were selected as part of the A3D3 award. The team will be led by Dr. Mia Liu of the Department of Physics and Astronomy. The team includes Dr. Maria Dadarlat of Weldon School of Biomedical Engineering and Dr. Pan Li of the Department of Computer Science.
“This grant will fuel collaborative work between computer science and domain scientists,” says Liu. “For example, Pan and I have been working on solving some of the very challenging problems in particle physics with graph neutral networks that he is an expert on. We have presented our results in recent conferences (Semi-supervised machine learning for pileup per particle identification at the LHC with Graph Neural Networks) and have a few papers in the pipeline. Our collaborative work has inspired theoretical machine learning (ML) research that Pan’s group is doing follow-up studies on. This multidisciplinary institute is built upon the successful fast machine learning community where we work effectively and efficiently with computer scientists and engineers to develop methods and tools for applying ML in scientific instruments and data processing. With this institute, we would like to expand its impact by working with other domain experts such as neuroscientists (Maria's field) and astrophysicists, as well as improving the toolkit with CS/ECE experts such as Song Han (MIT), Deming Chen (UIUC).”
To take full advantage of fast AI, A3D3 targets fundamental problems in three fields of science: high energy physics, multi-messenger astrophysics, and systems neuroscience. A3D3 works closely within these domains to develop customized AI solutions to process large datasets in real-time, significantly enhancing their discovery potential. The ultimate goal of A3D3 is to construct the institutional knowledge essential for real-time applications of AI in any scientific field. A3D3 will empower scientists with new tools to deal with the coming data deluge through dedicated outreach efforts.
At the Large Hadron Collider (LHC), the challenge of processing data is daunting. With future aggregate data rates exceeding 1 petabit per second, the data rates at the LHC exceed all other devices in the world. The aim of A3D3 is to build a series of tools that will enable the processing of all of this information in real-time using AI. Through the use of AI, A3D3 aims to perform advanced analyses, such as anomaly detection, and particle reconstruction on all collisions happening 40 million times per second!
Within the field of multi-messenger astrophysics, A3D3 is working to integrate AI to promptly and computationally efficiently process the data from telescopes, neutrino detectors, and gravitational-wave detectors in order to quickly identify astronomical events corresponding to the most violent phenomena in the Cosmos. The ability to identify and further distribute these events as astronomical alerts enables the entire transient astronomy to cross-correlate observations and understand astrophysical phenomena across multiple different forces.
In systems neuroscience, A3D3 is working to discover the computations that brain-wide neural networks perform to process sensory and motor information during behavior. To do so, A3D3 will develop and implement high-throughput and low-latency AI algorithms to process, organize, and analyze massive neural datasets in real time. These real-time analyses will enable new approaches to probing brain function such as causal, closed-loop manipulations. Applying powerful AI methods to systems neuroscience will significantly advance our ability to analyze and interpret neural activity and its relationship to behavior.
Other than Purdue University, institutions involved with the A3D3 are as follows:
- University of Washington
- University of Illinois at Urbana-Champaign
- Duke University
- Massachusetts Institute of Technology
- University of Minnesota
- University of California San Diego
- Wisconsin IceCube Particle Astrophysics Center
- California Institute of Technology
NSF Contact: Maria (Fernanda) Pembleton
UW Point of Contact: James Urton jurton@uw.edu | 206-543-2580
Purdue University Contacts: Mia Liu, Maria Dadarlat, and Pan Li