NCSA Welcomes 2022-23 Fellows September 23, 2022 Announcements Artificial IntelligenceData AnalyticsHealth SciencesHPC OperationsInstitutional PartnershipsModeling and SimulationSoftware and ApplicationsVisualization Share this page: Twitter Facebook LinkedIn Email By Andrew Helregel The National Center for Supercomputing Applications has named 12 new fellows for the 2022-23 academic school year. The NCSA Fellowship program is a competitive program for faculty and researchers at the University of Illinois Urbana-Champaign, which provides seed funding for new collaborations that include NCSA staff as integral contributors to the project. The new fellows and their projects are: A Machine Learning and Geospatial Approach to Targeting Humanitarian Assistance among Syrian Refugees in Lebanon Angela Lyons portrait NCSA Fellow: Angela Lyons (Agricultural and Consumer Economics) NCSA Collaborators: Yifang Zhang and Aiman Soliman About the Project: The Syrian refugee crisis is one of the largest mass displacement crises in recent times. For longer than a decade, Lebanon has hosted about 1.5 million refugees (20% of its population). The protracted displacement, coupled with the recent COVID-19 pandemic and Lebanon’s economic crisis, have made it more challenging for humanitarian agencies to identify the households most in need of assistance. This project proposes to apply machine learning and geospatial methods to newly released and proprietary data from key humanitarian agencies to help develop more effective and operationalizable targeting strategies that provide a reliable complementarity between proxy means testing and multidimensional methods. The insights from this project will have important implications for humanitarian organizations seeking to reduce the inclusion and exclusion errors, especially given increasing poverty and displacement and shrinking humanitarian funding. High Performance Computing for Magnetized Neutron Stars Antonios Tsokaros portrait NCSA Fellow: Antonios Tsokaros (Physics) NCSA Collaborator: Roland Haas About the Project: Neutron stars have magnetic fields which can reach levels that can distort the very nature of quantum vacuum. Although, numerical general relativistic magnetohydrodynamics has been investigated for many years, self-consistent modeling of these extreme compact objects is still missing. This project aims to study the combined Einstein-Euler-Maxwell system in a self-consistent way that will yield realistic models of magnetized neutron stars for the first time. This will be achieved by the further development of two independent general relativistic magnetohydrodynamics codes that will be used in tandem to perform state of the art numerical modeling in order to address the challenges of multi-messenger astronomy. Physics-Informed Neural Network Modeling of Fluid Flow in Microporous Materials Roman Makhnenko portrait NCSA Fellow: Roman Makhnenko (Civil and Environmental Engineering) NCSA Collaborators: Volodymyr Kindratenko, Seid Koric and Shirui Luo About the Project: Uncertainty quantification is important in geoenergy applications where the challenge is to make reliable predictions about the performance of complex physical systems based on sparse field and laboratory data. This project proposes the development of robust characterization methods at micro-scale to evaluate different effects on subsurface fluids flow can be combined with numerical models and artificial intelligence algorithms to predict the core and field scale properties. First-Principles and Machine Learning Modeling of Atomic Disorder and Optical Properties Andre Schleife portrait NCSA Fellow: Andre Schleife (Materials Science and Engineering) NCSA Collaborator: Santiago Nunez-Corrales About the Project: The team will address the important problem of describing the effect of disorder on optical and magneto-optical properties of materials by developing a quantitative computational approach that combines first-principles simulations and machine learning. This project aims to address this situation for atomic geometries and magnetic moments by using machine learning to render the computationally costly sampling of many configurations tractable. Target applications within this project include temperature-induced atomic disorder for the technologically important oxide semiconductor SnO2 and temperature-induced magnetic moment disorder for the antiferromagnetic insulator MnF2. The team will predict their optical and magneto-optical properties, respectively, for the disordered state as a function of temperature. If successful, the computational approach developed here has multiple immediate applications at the Illinois campus and beyond. Mixed Reality Platform for Immersive Physics-based Structural Digital Twins Mohamad Alipour portrait NCSA Fellow: Mohamad Alipour (Civil and Environmental Engineering) NCSA Collaborator: Robert Sisneros About the Project: The backlog of cyber-physical system technology development in the domain of civil infrastructure systems is evidenced by the grossly inadequate state of our national infrastructure. The vast majority of the engineering operations in this domain are still highly manual and even if computerized, they lack the necessary linkage between the physical structure and their computer simulations and sensing data. This project proposes to test the hypothesis that immersive mixed reality technology can be leveraged to create agile physics-based digital replicas of physical structures wherein sensing and simulation are tightly coupled, superimposed, and visualized with the physical asset. PALYIM – An Automated Image Analysis Platform for Fossil Pollen Classification Surangi Punyasena portrait NCSA Fellow: Surangi Punyasena (School of Integrative Biology) NCSA Collaborator: Sandeep Puthanveetil Satheesan About the Project: This project aims to develop an intelligent web-accessible palynology image analysis platform (PALYIM) integrating high-throughput high-resolution imaging and computer vision to tackle the long-standing problem of fossil pollen identification. The proposed image analysis system will automate and streamline curation of modern reference and fossil specimen images and image analysis workflows. The result will be an intelligent image database that incorporates phylogenetically meaningful characterizations of pollen morphology, provides testable hypotheses of taxonomic affinity, and is open and available to researchers without experience in programming or machine learning. Insurtech Innovation and University-Industry Collaboration Zhiyu “Frank” Quan portrait NCSA Fellow: Zhiyu (Frank) Quan (Mathematics) NCSA Collaborator: Volodymyr Kindratenko About the Project: Social media is now ranked as one of the top categories of alternative data currently used by insurance companies to stay competitive. The ability to apply natural language processing to online content and other unstructured documents will unleash new opportunities to investigate business risks and policyholder behaviors. In addition, foot traffic data captured by various sources can help insurance monitor policy holders’ behavior. There are various state-of-art techniques to extract useful information from the foot traffic data, including spatial and temporal analysis, and geospatial analysis. With the support of the Insurtech company and insurance company, the team would be able to ensemble proprietary datasets and perform research to improve the loss propensity models. Synergistic Integration of AI with MR Spectroscopic Imaging to Unravel Molecular Fingerprints of Brain Function and Neurodegenerative Diseases Zhi-Pei Liang portrait Brad Sutton portrait NCSA Fellows: Zhi-Pei Liang (Electrical and Computer Engineering) and Brad Sutton (Bioengineering, ECE) NCSA Collaborators: Volodymyr Kindratenko, Maria Jaromin and Colleen Bushell About the Project: Brain mapping is one of the most exciting frontiers of contemporary science. Prof. Liang’s group breakthroughs in magnetic resonance spectroscopic imaging (MRSI) have opened up new opportunities for brain mapping, especially for metabolic imaging to unravel the structural, functional and molecular fingerprints of brain function and neurodegenerative disorders. However, these high-dimensional data sets acquired also pose new significant challenges for image processing and analysis to efficiently and effectively extract important biological information (e.g., metabolic and morphological alterations due to aging). The primary objective of this project is to develop special AI tools and brain atlases of structures and molecular distributions urgently needed for AI-enabled processing and analysis of high-dimensional MRSI data. Early Detection and Prediction of Parkinsonism Powered by Multi-Modal Few-Shot Learning Yuxiong Wang portrait NCSA Fellows: Yuxiong Wang (Computer Science), Christopher M. Zallek (OSF) and George Heintz (Health Care Engineering Systems Center) NCSA Collaborator: Volodymyr Kindratenko About the Project: Neurological disorders are among the most frequent causes of morbidity and mortality in the US, the most common being Parkinson’s and Alzheimer’s. The insidious and heterogeneous onset of neurodegenerative diseases challenges the abilities of the primary care systems to appropriately diagnose and manage these diseases. A data-efficient AI is urgently needed for healthcare and its patients. The team proposes an AI supported system that leverages various examination modalities and tracks complementary symptoms of neurological patients and reports findings to the neurologists and will focus on discriminating several indicators that are associated with Parkinsonism.