Ensemble Model Class

At the National Synchrotron Light Source II (NSLS-II) researchers are able to study materials in order to gain information about next generation energy applications, as well as the electronics and computers of the future. As these studies progress, there are rapid amounts of data being collected that need to be analyzed. At the moment, the speed of modern synchrotron data collection has outperformed data analysis techniques for years. A new approach to this problem, machine learning, seeks to automate data analysis preventing researchers from having to hand pick through large quantities of data. The goal of this project is to identify the situations where human intervention is required and understand if using smaller amounts of data for analysis are representative of real results. This project will also develop methods to construct new data points given limited information in order to optimize the beamlines in real time. This will benefit situations where it is unknown whether it is necessary to continue measuring samples to get a clearer reading. It will be important to utilize reinforced learning methods as they have shown promise for machine optimization uncovering the most valuable data. This project will evaluate various methods to create mathematical models which analyze data and determine which are best fit for applications at NSLS-II. I now am proficient in machine learning techniques such as supervised, unsupervised, reinforced learning, and their implementations in Python. I have gained skills in software development, big data analysis, mathematical modeling, and how to apply all of these newly learned techniques to real world problems.