Leifur Leifsson
assistant professor of aerospace engineering
Multifidelity Modeling and Search Using Adaptive Field Prediction
Leifsson will create new computation tools that will enhance uncertainty analysis and design optimization under uncertainty in complex engineering systems, such as aerodynamics, electromagnetics and mechanical structures. The advanced methods will combine metamodeling techniques and machine learning, as well as novel adaptation techniques – offering rapid and reliable design of complex engineered systems, such as in transportation, energy harvesting, weather forecasting and communication.
Leifsson will also create a new undergraduate short course and organize a symposium on computational design – and will launch an online hub to make his advanced simulation-based design techniques available to other engineers across the country.
Matthew Panthani
assistant professor of chemical and biological engineering
Synthesis and Properties of Group IV Colloidal Quantum Wells
Panthani will develop new types of ultrathin semiconductor materials that may help improve the capabilities, efficiency, and costs of computing and telecommunication. His team will use novel materials synthesis techniques to create a single- to few-atom thick silicon-germanium alloys with controlled composition, structure and surface chemistry – and perform materials characterization to determine how structure and chemistry influence properties. The project also involves coordinating molecules to the surfaces of these two-dimensional semiconductors to create solutions that can be deposited onto substrates to make processing of electronic devices from these materials easier.
Panthani will also partner with high-school teachers who conduct research in his lab to incorporate their experiences into their curricula and implement a summer workshop designed to encourage underrepresented groups to pursue careers in STEM fields.
Soumik Sarkar
assistant professor of mechanical engineering
Robustifying Machine Learning for Cyber-Physical Systems
Sarkar will build computational techniques to detect and mitigate risk in using machine learning and artificial intelligence for cyber-physical systems, such as self-driving cars. His framework and algorithms will help deep learning models better address “edge cases” where the real-life situation isn’t represented well in the training data set – and to fend off adversarial attacks on machine learning based decision systems. Algorithms will be validated on experimental self-driving cars and robotics test beds.
Sarkar will also develop new curriculum, research experiences, and other outreach activities for high school students and teachers in the critical interdisciplinary area of system theory and data science.