TITLE: Application of Optimization and Machine Learning for New Materials Development
ABSTRACT: Material discovery traditionally has taken the iterative trial and error approach in order to develop new materials with desired properties. More recently, there has been a broad initiative to use statistical, optimization, and machine learning methods to improve the process of new material development by making the process less costly and time consuming. The role of using these methods is to (1) help select appropriate set of experiments to be performed, (2) develop more effective ways of learning from the experiments, and (3) guide the process towards better performing materials.
The proposed project has two parts. The first part consists of collaborative effort by experimental material scientists, computational material scientists, and our group in order to solve specific materials development problems. The goal is to develop an integrated computational-statistical-experimental methodology for the fabrication of new materials for specific purposes. The more specific task of the collaboration is to fabricate a new material for the removal of persistent organic pollutants from drinking water. The focus of this part of the proposed project is to understand the overall problem from a decision theoretic point of view, develop an appropriate formulation of the problem, and finally to provide solutions to guide experimentation.
The second part of the project views the material development process more broadly as a sequential decision making problem and considers making new contributions to the existing methods to solve this more general problem.
COMMITTEE: Advisor/Chair
Pirooz Vakili, SE/ME; Emily Ryan, MSE/ME; Ioannis Paschalidis, SE/ECE/BME; Jillian Goldfarb, BEE, Cornell University |
