Computational tools towards automating the scientific method
Name
1252628873-MIT.pdf
Size
7.77 MB
Format
Adobe PDF
Checksum (MD5)
4f7982f4dd329bd04e8cffd0da786d33
Author(s)
Spanbauer, Span.
Advisor(s)
Ian W. Hunter.
Date Issued
2021
Publisher
Massachusetts Institute of Technology
Abstract
We present a collection of novel computational tools designed to contribute to the goal of large-scale scientific automation. Deep Involutive Neural MCMC and other inference compilation techniques present a promising path to accelerating inference in probabilistic programs. Neural Group Actions provide foundational methods for learning symmetric transformations useful for the development of statistical models and probabilistic algorithms. Coarse-Grained Nonlinear System Identification provides an exceptional new model class for nonlinear dynamic systems, enabling accurate model identification with minimal experimental data. Optimization plus Stochastic Interchange is a flexible new way to generate experimental stimuli, leading to optimally informative measurements during system identification. Extended Koopman Models advance a new method for the optimal control of nonlinear systems. When coupled with high-throughput laboratory automation, these and other computational tools made possible by recent developments in artificial intelligence promise to revolutionize the way we do science and engineering.
Description
Thesis: Ph. D., Massachusetts Institute of Technology, Department of Mechanical Engineering, February, 2021
Cataloged from the official PDF of thesis.
Includes bibliographical references (pages 123-134).
Subjects
Mechanical Engineering.
MIT Department
Massachusetts Institute of Technology. Department of Mechanical Engineering
Terms of Use
MIT theses may be protected by copyright. Please reuse MIT thesis content according to the MIT Libraries Permissions Policy, which is available through the URL provided.
Persistent DSpace Link
