The Center for Artificial Intelligence Innovation (CAII) at the National Center for Supercomputing Applications received $1 million in funding from NASA to support the Euclid space mission, which explores dark matter and dark energy throughout the universe.
By developing and integrating an open-sourced deep learning framework to process images captured by Euclid, CAII and Principal Investigator Xin Liu aim to accurately and efficiently identify blended galaxies or overlapping sources of information that make data analysis much more difficult.
“A significant challenge in Euclid data analysis is the presence of blended or overlapping sources, which leads to biased measurements in critical areas such as photometry, photometric redshift estimation, galaxy morphology and weak gravitational lensing,” Liu said. “Addressing this challenge is crucial for ensuring the accuracy of Euclid’s scientific outputs.”
Liu and a team of researchers will utilize an artificial intelligence tool known as Detection, Instance Segmentation and Classification with Deep Learning (DeepDISC), which leverages machine learning to transform how stars and galaxies are detected. Using DeepDISC within the Euclid mission will also allow researchers to quantify uncertainty in the analysis predictions.
DeepDISC will be essential for maximizing the scientific return of the Euclid mission to transform our understanding of the dark universe.
Xin Liu, Principal Investigator
This innovative framework will also be adaptable to other space exploration projects, including the Vera C. Rubin Observatory, which anticipates its first light later this year. Improving the accuracy and efficiency of deblending ground-based images and those taken in space emphasizes the interdisciplinary excellence of CAII, NCSA and NASA.
“Co-principal Investigators Director of CAII Vlad Kindratenko, Astronomy Professor Yue Shen and Computer Science Professor Yuxiong Wang provide critical expertise and leadership, strengthening the project’s interdisciplinary foundation,” Liu said. “Their contributions will ensure robust computational infrastructure, sophisticated data analysis techniques and advanced machine learning methodologies.”
“The computer vision and AI community has developed powerful foundation models for understanding the visual world through natural images,” Wang said. “Now is an exciting time to extend these capabilities toward unlocking the mysteries of the universe.”
ABOUT CAII
The Center for Artificial Intelligence at the National Center for Supercomputing Applications at the University of Illinois at Urbana-Champaign operates as a central nexus that spearheads AI research and application in academia and industry. The center empowers and supports advancements in AI by leveraging NCSA’s cutting-edge technology and expertise and facilitating collaboration across multiple disciplines, including agriculture, astrophysics, automotive, big data, and infrastructure. Furthering NCSA’s commitment to accelerating AI, the CAII is dedicated to building foundations that will pave the way for the next generation of innovators.