Restricted Research - Award List, Note/Discussion Page
Fiscal Year: 2023
2132 The University of Texas at Dallas (144020)
Principal Investigator: Hsu,Julia Wan-Ping
Total Amount of Contract, Award, or Gift (Annual before 2011): $ 285,673
Exceeds $250,000 (Is it flagged?): Yes
Start and End Dates: 1/1/23 - 12/31/23
Restricted Research: YES
Academic Discipline: Material Science Engineering
Department, Center, School, or Institute: ECS
Title of Contract, Award, or Gift: Merging Machine and Human Intelligence for Energy Materials Research
Name of Granting or Contracting Agency/Entity:
Simons Foundation
Program Title: none
Note:
I aim to understand which ML approach applies to which type of problems and learn in-depth the right approach for optimizing materials synthesis and processing. Leveraging human domain knowledge in the loop (as opposed to excluding the human through complete automation). The rationale for keeping the human in the loop is threefold: (i) from an ethical point of view, a human is necessary to avoid potentially hazardous unintentional errors and also to prevent intentional dual use of the technology. (ii) human expert input is necessary to build trust in the ML methods, so encourage adoption by a highly skilled labor force. (iii) in many cases when data is sparse, human domain knowledge can augment existing machine-learned trends, improving learning outcomes. Second, applying ML techniques to the physical sciences requires adapted methods for cases when data are expensive, sparse, noisy, and/or incomplete. This requires strategies for noise reduction, data augmentation, active learning, and sparse learning, among others.
Discussion: No discussion notes