2022 AI + Science Summer School
The first AI + Science Summer School was a huge success! Join the DSI newsletter for announcements about next year’s program.
Modern artificial intelligence and machine learning will fundamentally change scientific discovery. We are just beginning to understand the possibilities presented by an era of extraordinarily powerful computers coupled with advanced instruments capable of collecting enormous volumes of high-resolution experimental data. Off-the-shelf machine learning tools cannot fully extract the knowledge contained in these datasets, let alone generate new theories and propose future experiments.
The AI + Science Summer School was an August 2022 event jointly hosted by the Data Science Institute (DSI) and Institute for Mathematical and Statistical Innovation (IMSI) at the University of Chicago. It focused on four core themes at the heart of this emerging paradigm of scientific discovery: AI uncovering new laws of nature, AI guiding scientific measurement, physics-informed machine learning, and scientific discovery advancing AI frontiers.
Speakers and talks at the summer school included (click link to watch video):
- Christoph Ortner – “Atomic Cluster Expansion – A Framework For Modeling Equivariant Properties Of Atoms And Molecules“
- Camille Avestruz – “Machine Learning in Astrophysics/Cosmology“
- Romit Maulik – “Learning Nonlinear Dynamical Systems Using Scientific Machine Learning“
- Risi Kondor – “Equivariant Neural Networks For Physics And Chemistry“
- Panel (Eric Jonas, Romit Maulik, Christoph Ortner, Samanthan Riesenfeld; moderated by David Miller) – “Moving beyond the ML Hype – Research Frontiers in AI and Science“
- Samantha Riesenfeld – “Mathematical Models for Single-Cell Genomic Data: Challenges and Opportunities” (note: due to technical difficulties, video starts halfway through talk)
- Miles Stoudenmire – “Tensor Networks in Machine Learning Architectures“
- Roman Garnett – “Bayesian Active Learning For Experimental Design“
- Claudia Solis-Lemus – “The Challenges Of Machine Learning Models In Omics Data“
- Eric Jonas – “Sampling Methods For High-Dimensional Distributions“
The goal of the program is to introduce a new generation of diverse interdisciplinary graduate students and researchers to the emerging field of AI + Science. We also hope this program can build community and spur new research directions focused on AI-enabled scientific discovery across the physical and biological sciences.
The organizing committee for the AI + Science Summer School is Yuxin Chen, Aaron Dinner, Ian Foster, Eric Jonas, Yuehaw Khoo, Risi Kondor, David Miller, Brian Nord, Surinarayanan Vaikuntanathan, and Rebecca Willett.
Dr. Avestruz's research interests span astrophysics, cosmology, and computation. Dr. Avestruz uses simulations to make robust predictions and interpretations of observations of large-scale cosmic structure. Her primary focus is to understand the evolution of clusters of galaxies, the most massive gravitationally collapsed structures in our universe, comprised of hundreds to thousands of galaxies. Other aspects of her work prepare for the next decade of observations, which will produce unprecedented volumes of data. Dr. Avestruz incorporates big data methods, including machine learning, to extract gravitational lensing signatures that probe the mass distribution of massive galaxies and galaxy clusters.
Dr. Avestruz is passionate about making STEM accessible to those who have been historically excluded from the sciences and the academy. She has taught software and computation workshops to a variety of audiences ranging from undergraduate Women in Science and Engineering (WISE) groups to societies promoting the advancement of underrepresented minorities in STEM. She has also engaged in public outreach, believing that intellectual exchange is not and should not be isolated within the walls and members of higher education. This includes teaching labs to students from local middle and high schools and giving public talks at Astronomy on Tap and to underserved elderly communities.
Roman is Associate Professor at Washington University in St. Louis. His research focus is developing Bayesian methods for automating scientific discovery and validating these algorithmic developments in applications across the natural sciences and engineering. These efforts are supported by an NSF CAREER award. He is also the author of a recent comprehensive textbook on Bayesian optimization available at bayesoptbook.com.
Eric Jonas is a new professor in the Department of Computer Science at the University of Chicago. His research interests include biological signal acqusition, inverse problems, machine learning, heliophysics, neuroscience, and other exciting ways of exploiting scalable computation to understand the world. Previously he was at the Berkeley Center for Computational Imaging and RISELab at UC Berkeley EECS working with Ben Recht.
Risi Kondor is an Associate Professor in the Department of Computer Science, Statistics, and the Computational and Applied Mathematics Initiative at the University of Chicago. He joined the Flatiron Institute in 2019 as a Senior Research Scientist with the Center for Computational Mathematics. His research interests include computational harmonic analysis and machine learning. Kondor holds a Ph.D. in Computer Science from Columbia University, an MS in Knowledge Discovery and Data Mining from Carnegie Mellon University, and a BA in Mathematics from the University of Cambridge. He also holds a diploma in Computational Fluid Dynamics from the Von Karman Institute for Fluid Dynamics and a diploma in Physics from Eötvös Loránd University in Budapest.
I am an Assistant Computational Scientist at the Mathematics and Computer Science division (MCS) at Argonne National Laboratory. I am also jointly appointed as Research Assistant Professor in the Department of Applied Mathematics at the Illinois Institute of Technology, Chicago. Previously, I was the 2019 Margaret Butler Postdoctoral Fellow at Argonne National Laboratory and obtained my Ph.D. in Mechanical & Aerospace Engineering from Oklahoma State University. My interests are scientific machine learning, stochastic processes, high performance computing with applications to engineering, geoscience, plasma physics.
Christoph is a professor in the Mathematics Department at UBC. Christoph works in numerical analysis, applied analysis, scientific computing, and machine learning, with particular focus on applications to atomistic modelling, multi-scale modelling, and coarse-graining.
Samantha Riesenfeld is an Assistant Professor of Molecular Engineering and of Genetic Medicine, a member of the Committee on Immunology, an Associate Member of the Comprehensive Cancer Center, and co-director of the new Computational and Systems Immunology PhD track in Immunology and Molecular Engineering. She leads an interdisciplinary research program focused on developing and applying genomics-based machine learning approaches to investigate the cellular components, transcriptional circuitry, and dynamics underlying complex biological systems, with a special interest in inflammatory immune responses and solid tumor cancer.
I am an assistant professor at the Wisconsin Institute for Discovery and the Department of Plant Pathology at the University of Wisconsin-Madison. Originally from Mexico City, I did my Undergraduate degrees in Actuarial Sciences and Applied Mathematics at ITAM. Then, I did a MA in Mathematics and a PhD in Statistics at the University of Wisconsin-Madison. In my spare time, I enjoy swimming, running, biking, climbing and yoga!
Miles Stoudenmire joined the Flatiron Institute in 2017 as a research scientist at the Center for Computational Quantum Physics (CCQ). He earned a B.S. in physics and math from the Georgia Institute of Technology in 2005 and a Ph.D. in physics from the University of California Santa Barbara in 2010. He held postdoctoral and research scientist positions at the University of California Irvine and the Perimeter Institute for Theoretical Physics.
Miles’ speciality is using tensor networks, a technique for compressing and manipulating large collections of parameters encountered in quantum mechanics and in machine learning. A key area of his research is expanding the reach of tensor networks to include more detailed and realistic aspects of quantum systems, such as the use of chemically accurate basis sets and the inclusion of finite temperature effects.
Miles is also the lead developer of the ITensor software for tensor network calculations. The ITensor interface allows users to directly transcribe tensor network diagrams into efficient code, and is very useful for rapidly prototyping new algorithms.