Data Analysis and Machine Learning group
Developing algorithms and foundations for data analysis and machine learning.
Oak Ridge National Laboratory’s Data Analysis and Machine Learning Group harnesses AI, high-performance computing, and advanced data science to revolutionize how scientific experiments and simulations are conducted, with a particular focus on real-time analysis and steering of complex experiments at major research facilities. Through innovative algorithms and computational frameworks that span from edge devices to exascale supercomputers, the group aims to accelerate scientific discovery while making computational and experimental facilities more efficient and accessible, ultimately advancing the understanding of materials, climate, and other critical scientific domains.
The group’s work spans three main areas: experimental facility optimization, algorithms for large-scale scientific simulations, and foundational machine learning methods.
One focus is revolutionizing scattering experiments at facilities such as the Spallation Neutron Source. The group has developed an innovative edge-to-exascale workflow that combines real-time data processing with AI-driven predictive modeling. This system enables researchers to make adaptive, data-driven decisions during experiments, significantly reducing data processing time from hours to minutes and improving experimental accuracy. The group’s work with the Frontier supercomputer demonstrates how advanced computing can make complex scientific facilities more efficient and accessible.
Privacy and security are also key priorities, as evidenced by the group’s development of privacy-preserving federated learning systems that enable secure collaboration across multiple research facilities. This work helps protect sensitive data while allowing researchers to benefit from combined computational resources and datasets.
The real-world impact of this work extends beyond basic science. The advances in computational and experimental steering could accelerate materials research, leading to breakthroughs in energy storage, electronics, and other critical technologies. Our scalable data assimilation and AI-driven predictive methods in climate and weather modeling could improve weather predictions and our understanding of climate.
Looking ahead, the Data Analytics and Machine Learning Group envisions a future in which AI-driven automation and real-time analysis become standard features at major computational and experimental research facilities, thereby democratizing access to advanced scientific tools while dramatically increasing the pace of scientific discovery. This work lays the foundation for a new era of data-driven scientific exploration that is both more efficient and more accessible to the broader scientific community.
Expertise
- Sparse optimization
- Uncertainty quantification
- Machine learning
- Data analysis
- Algorithms for interconnected scientific ecosystems
