Algorithms for Machine Learning and Data Analysis

We are developing machine learning algorithms to i) analyze complex sets of experimental data that arise in materials and catalysis design and to ii)  develop new sensing and control technologies. Specifically, we are combining techniques from space-time statistics and feature extraction to develop classification and deep learning models. 

The space-time nature of the feature information that we are considering leads to computationally challenging optimization models. We are addressing such problems by developing fast and scalable optimization algorithms based on parallel linear algebra, clustering, and projection techniques. These capabilities help us establish fundamental connections between chemistry, materials science, and systems engineering. 

This work is in collaboration with Profs. Nick Abbott, George Huber, and Jim Dumesic.