Abstract: Model reduction methods have grown from the computational science community, with a focus on reducing high-dimensional models that arise from physics-based modeling, whereas machine learning has grown from the computer science community, with a focus on creating expressive models from black-box data streams. Yet recent years have seen an increased blending of the two perspectives and a recognition of the associated opportunities. This talk presents our work in operator inference, where we learn effective reduced-order operators directly from data. The physical governing equations define the form of the model we should seek to learn. Thus, rather than learn a generic approximation with weak enforcement of the physics, we learn low-dimensional operators whose structure is defined by the physics. This perspective provides new opportunities to learn from data through the lens of physics-based models and contributes to the foundations of Scientific Machine Learning, yielding a new class of flexible data-driven methods that support high-consequence decision-making under uncertainty for physical systems.

Bio: Karen E. Willcox is Director of the Oden Institute for Computational Engineering and Sciences, Associate Vice President for Research, and Professor of Aerospace Engineering and Engineering Mechanics at the University of Texas at Austin. She is also External Professor at the Santa Fe Institute. Before joining the Oden Institute in 2018, she spent 17 years as a professor at the Massachusetts Institute of Technology, where she served as the founding Co-Director of the MIT Center for Computational Engineering and the Associate Head of the MIT Department of Aeronautics and Astronautics. Prior to joining the MIT faculty, she worked at Boeing Phantom Works with the Blended-Wing-Body aircraft design group. She is a Fellow of the Society for Industrial and Applied Mathematics (SIAM) and Fellow of the American Institute of Aeronautics and Astronautics (AIAA).

Abstract: Rich functionalities of quantum and strongly correlated materials emerge from the interplay between the electronic, orbital, lattice, and spin degrees of freedom that often lead to complex structural and electronic phenomena spanning atomic to mesoscopic scales. In many cases, these phenomena are associated with translational symmetry breaking, local frozen disorder, or strongly correlated disorder. However, the relevant mechanisms and roles of individual subsystems often remain unknown. Over the last decade, Scanning Transmission Electron Microscopy has emerged as a powerful quantitative probe of materials structure on the atomic level, providing high veracity information on local chemical bonding, composition, and symmetry breaking distortions. We aim to harness the power of machine learning methods to build a comprehensive picture of the chemistry and physics of quantum materials from these observations. In this presentation, I will illustrate the application of rotationally-invariant variational autoencoders (rVAE) towards the effective exploration of the chemical evolution of the system based on local structural changes, effectively discovering molecular building blocks and chemical reactions pathways in unsupervised manner. I will further illustrate the applications of the Bayesian inference methods towards inferring the mesoscopic and atomistic physics of materials, and illustrate the pathways towards incorporation of physical models as priors within Bayesian optimization towards effective sampling of experimental parameter spaces. Ultimately, we seek to answer the questions such as whether frozen atomic disorder drives the emergence of the local structural distortions or average shift of the Fermi level induces structural reconstruction that in turn drive cation distribution, whether the nucleation spot of phase transition can be predicted based on observations before the transition, and what is the driving forces controlling the emergence of unique functionalities in quantum materials. This research is supported by the by the U.S. Department of Energy, Basic Energy Sciences, Materials Sciences and Engineering Division and the Center for Nanophase Materials Sciences, which is sponsored at Oak Ridge National Laboratory by the Scientific User Facilities Division, BES DOE.

Bio: Sergei V. Kalinin is corporate fellow at the Center for Nanophase Materials Sciences (CNMS) at Oak Ridge National Laboratory. His research interests include atom by atom fabrication, application of machine learning and artificial intelligence in atomically resolved and mesoscopic imaging to guide the development of advanced materials for energy and information technologies, as well as coupling between electromechanical, electrical, and transport phenomena on the nanoscale. He received his Ph.D. from the University of Pennsylvania in 2002, followed by a Wigner fellowship at ORNL (2002-2004). He is a recipient of the Blavatnik National Awards for Young Scientists (2018); RMS medal for Scanning Probe Microscopy (2015); Presidential Early Career Award for Scientists and Engineers (PECASE) (2009); IEEE-UFFC Ferroelectrics Young Investigator Award (2010); Burton medal of Microscopy Society of America (2010); ISIF Young Investigator Award (2009); American Vacuum Society Peter Mark Memorial Award (2008); R&D100 Awards (2008 and 2010); Ross Coffin Award (2003); Robert L. Coble Award of American Ceramics Society (2009); and a number of other distinctions. He has published more than 500 peer-reviewed journal papers, edited 4 books, and holds more than 10 patents. He has organized numerous symposia (including symposia on Scanning Probe Microscopy on Materials Research Society Fall meeting in 2004, 2007, and 2009) and workshops (including International workshop series on PFM and Nanoferroelectrics), and acted as consultant for companies such as Intel and several Scanning Probe Microscopy manufacturers. He is also a member of editorial boards for several international journals, including Nanotechnology, Journal of Applied Physics/Applied Physics Letters, and recently established Nature Partner Journal Computational Materials.

Abstract: Many tasks in fluid mechanics, such as design optimization and control, are challenging because fluids are nonlinear and exhibit a large range of scales in both space and time. This range of scales necessitates exceedingly high-dimensional measurements and computational discretization to resolve all relevant features, resulting in vast data sets and time-intensive computations. Indeed, fluid dynamics is one of the original big data fields, and many high-performance computing architectures, experimental measurement techniques, and advanced data processing and visualization algorithms were driven by decades of research in fluid mechanics. Machine learning constitutes a growing set of powerful techniques to extract patterns and build models from this data, complementing the existing theoretical, numerical, and experimental efforts in fluid mechanics. In this talk, we will explore current goals and opportunities for machine learning in fluid mechanics, and we will highlight a number of recent technical advances. Because fluid dynamics is central to transportation, health, and defense systems, we will emphasize the importance of machine learning solutions that are interpretable, explainable, generalizable, and that respect known physics.

Bio: Dr. Steven L. Brunton is the James B. Morrison Professor of Mechanical Engineering at the University of Washington. He is also Adjunct Associate Professor of Applied Mathematics and of Computer Science, and he is a Data Science Fellow at the eScience Institute. Steve received the B.S. in mathematics from Caltech in 2006 and the Ph.D. in mechanical and aerospace engineering from Princeton in 2012. His research combines machine learning with dynamical systems to model and control systems in fluid dynamics, biolocomotion, optics, energy systems, and manufacturing. He is a co-author of three textbooks, received the Army and Air Force Young Investigator Program awards, the Presidential Early Career Award for Scientists and Engineers (PECASE), and he was awarded the University of Washington College of Engineering junior faculty and teaching awards.

Abstract: While deep learning has shown tremendous success in many domains, it remains a grand challenge to incorporate physical principles to such models for applications in physical sciences. In this talk, I will discuss (1) Turbulent-Flow Net: a hybrid approach for predicting turbulent flow by marrying well-established computational fluid dynamics techniques with deep learning (2) Equivariant Net: a systematic approach to improve generalization of spatiotemporal models by incorporating symmetries into deep neural networks. I will demonstrate the advantage of our approaches to a variety of physical systems including fluid and traffic dynamics.

Bio: Dr. Rose Yu is an assistant professor at the University of California San Diego, Department of Computer Science and Engineering. She earned her Ph.D. in Computer Sciences at the University of Southern California in 2017. She was subsequently a Postdoctoral Fellow at the California Institute of Technology. She was an assistant professor at Northeastern University prior to her appointment at UCSD. Her research focuses on advancing machine learning techniques for large-scale spatiotemporal data analysis, with applications to sustainability, health, and physical sciences. A particular emphasis of her research is on physics-guided AI which aims to integrate first-principles with data-driven models. Among her awards, she has won Google Faculty Research Award, Adobe Data Science Research Award, NSF CRII Award, Best Dissertation Award in USC, and was nominated as one of the 'MIT Rising Stars in EECS.'

Abstract: Images from scanning transmission electron microscopes (STEM) are used to routinely to quantify the atomic scale structure, composition, chemistry, bonding, electron/phonon distribution and optical properties of nanostructures, interfaces and defects in many materials systems. However, quantitative and reproducible observations for many materials of current technological importance is limited by electron beam damage. The aim for broadening STEM applications to a wider range of samples and processes is therefore now to focus on more efficient use of the dose that is supplied to the sample. In practice, this is achieved by modelling and minimizing the dose, dose rate and dose overlap for any image. These minimized physical interaction conditions result in two main categories for achieving dose fractionation and optimizing the data content per unit dose – reducing the number of pixels being sampled in scanning mode (STEM), or increasing the speed of individual images in projection mode (TEM). For the case of the STEM, inpainting /machine learning methods allow data to be automatically recorded in a faster compressed form with less material damage. For the TEM, a similar increase in speed and damage reduction can be achieved by implementing compressive sensing/machine learning – this reduces the dose per image. In this presentation, the basic approach to understanding the physical interaction between the electron beam and the sample and how this leads to the integration of sub-sampling/inpainting/compressive sensing and machine learning into the imaging hardware will be described and the potential for future developments will be discussed.

Bio: Dr Nigel Browning is currently the Chair of Electron Microscopy in the School of Engineering and the School of Physical Sciences at the University of Liverpool. He received his undergraduate degree in Physics from the University of Reading, UK and his PhD in Physics from the University of Cambridge, UK. After completing his PhD in 1992, he joined the Solid State Division at Oak Ridge National Laboratory (ORNL) as a postdoctoral research associate before taking a faculty position in the Department of Physics at the University of Illinois at Chicago (UIC) in 1995. In 2002, he moved to the Department of Chemical Engineering and Materials Science at the University of California-Davis (UCD) and also held a joint appointment in the National Center for Electron Microscopy (NCEM) at Lawrence Berkeley National Laboratory (LBNL). In 2005 he moved the joint appointment from LBNL to Lawrence Livermore National Laboratory (LLNL) to become project leader for the Dynamic Transmission Electron Microscope (DTEM). In 2009, he also joined the Department of Molecular and Cellular Biology at UCD to focus on the development of the DTEM to study live biological structures. From 2011-2017 he was a Laboratory Fellow at the Pacific Northwest National Laboratory and the Director of the Chemical Imaging Initiative, a $42M program designed to change the paradigm in operando imaging methods for imaging energy materials and processes. He has over 25 years of experience in the development of new methods in electron microscopy for high spatial, temporal and spectroscopic resolution analysis of engineering and biological structures. His research has been supported by DOE, NSF, NIH, DOD and by industry, leading to research projects for over 30 graduate students and 35 postdoctoral research fellows. He is a Fellow of the American Association for the Advancement of Science (AAAS) and the Microscopy Society of America (MSA). He received the Burton Award from the Microscopy Society of America in 2002 and the Coble Award from the American Ceramic Society in 2003 for the development of atomic resolution methods in scanning transmission electron microscopy (STEM). With his collaborators at LLNL he also received R&D 100 and Nano 50 Awards in 2008, and a Microscopy Today Innovation Award in 2010 for the development of the dynamic transmission electron microscope (DTEM). He has over 350 publications (h-index=74) and has given over 300 invited presentations on the development and application of advanced TEM methods.

Abstract: Given rapid data growth due to advances in sensor technologies, there is a tremendous opportunity to systematically advance modeling in these domains by using machine learning methods. However, the “black box” use of ML often leads to serious false discoveries in scientific applications. Because the hypothesis space of scientific applications is often complex and exponentially large, a pure data-driven search using limited observation data can easily select a highly complex model that is neither generalizable nor physically interpretable, resulting in the discovery of spurious patterns. In this talk, we present a physics-guided machine learning approach that combines advanced machine learning models and physics-based models for predicting water temperature and streamflow in river networks. We build customized deep learning models to capture complex spatial and temporal patterns amongst river segments and also transfer knowledge to guide ML models to learn the physics of streamflow and thermodynamics. Additionally, we propose a new loss function that balances the performance over different river segments. We demonstrate the effectiveness of the proposed method in predicting temperature and streamflow in a subset of the Delaware River Basin. The proposed method has also been shown to produce better performance when generalized to different seasons or river segments with different streamflow ranges.

Bio: Xiaowei Jia is an Assistant Professor in the Department of Computer Science at the University of Pittsburgh. He received his Ph.D. degree from the University of Minnesota in 2020 under the supervision of Prof. Vipin Kumar. Prior to that, he got his M.S. degree from State University of New York at Buffalo in 2015 and his B.S. degree from University of Science and Technology of China in 2012. His research interests include physics-guided data science, spatio-temporal data mining, deep learning, and remote sensing. His work has been published in major data mining and AI journals (e.g., TKDE) and conferences (e.g., SIGKDD, ICDM, SDM, and CIKM), as well top-tier journals in hydrology (e.g., WRR) and agronomy (e.g., Agricultural Economics). Xiaowei was the recipient of UMN Doctoral Dissertation Fellowship (2019), the Best Conference Paper Award in ASONAM 16, and the Best Student Paper Award in BIBE 14.

Abstract: The discovery of gravitational waves by the Advanced Laser Interferometer Gravitational-wave Observatory (LIGO) has ushered in a new era in astrophysics. Successful detection of gravitational wave signals requires ground-based interferometers like LIGO and Virgo to be exquisitely isolated from environmental and instrumental noise. Albeit the highly sophisticated design of the detectors which carefully mitigates effects of most types of noise, they are still susceptible to non-Gaussian noise transients called glitches. As they can mask or mimic real gravitational wave signals and given their high rate of occurrence, proper characterization and classification of glitches is necessary. In this talk, we will first discuss the current state of the art in classifying LIGO glitches, which is largely based on fully-supervised deep transfer learning, and we will present recent results, based on tensor analysis, that show promise especially in cases where there is limited supervision. Subsequently, we will focus on the problem of characterizing those glitches by tracing their behavior throughout the numerous auxiliary channels that monitor different parts of the detector. We will present recent tensor-based analysis that aims at isolating a (small) number of auxiliary channels that are associated with different glitch types, and we will conclude by discussing the significance of achieving this goal, towards improving the quality of gravitational wave detection.

Bio: Evangelos Papalexakis is an Assistant Professor of the CSE Dept. at University of California Riverside. He received his PhD degree at the School of Computer Science at Carnegie Mellon University (CMU). Prior to CMU, he obtained his Diploma and MSc in Electronic & Computer Engineering at the Technical University of Crete, in Greece. Broadly, his research interests span the fields of Data Science, Machine Learning, Artificial Intelligence, and Signal Processing. His research involves designing interpretable models and scalable algorithms for extracting knowledge from large multi-aspect datasets, with specific emphasis on tensor factorization models, and applying those algorithms to a variety of real world problems, including detection of misinformation on the Web, explainable AI, and gravitational wave detection. His work has appeared in top-tier scientific venues and has attracted a number of distinctions, including best student paper award at PAKDD’14 and SDM’16. He was a finalist for the Microsoft PhD Fellowship and the Facebook PhD Fellowship. Besides his academic experience, he has industrial research experience working at Microsoft Research Silicon Valley during the summers of 2013 and 2014 and Google Research during the summer of 2015. He has been named as one of the "2016 KDD Rising Stars" by Microsoft Academic Search, and his doctoral dissertation received the 2017 SIGKDD Doctoral Dissertation Award (runner up).