To access this work you must either be on the Smith College campus OR have valid Smith login credentials.

On Campus users: To access this work if you are on campus please Select the Download button.

Off Campus users: To access this work from off campus, please select the Off-Campus button and enter your Smith username and password when prompted.

Non-Smith users: You may request this item through Interlibrary Loan at your own library.

Publication Date

2024-5

First Advisor

Luca Capogna

Document Type

Honors Project

Degree Name

Bachelor of Arts

Department

Mathematics and Statistics

Keywords

neural networks, machine learning, system identification, inverse problem, Smith College Geothermal Project

Abstract

This thesis is intended to be a mathematical and practical primer to the concepts behind deep learning. It overviews the necessity and fundamentals of data-driven methods, particularly generalization and types of learning. It offers an introduction to neural networks and their relevance to system identification problems, including the theory of universal approximation. The early perceptron model is explained and a proof of the Perceptron Learning Algorithm is offered. Basic architecture of activation and loss functions are reviewed, as well as the method of learning network parameters via (stochastic) gradient descent and backpropagation. Then, specific literature relevant to the project is reviewed, with a special focus on [10] whose algorithm was the inspiration for the one developed in this thesis. Finally, the network used is described. The training data and its preprocessing, the network’s hyperparameters, and results of learning are discussed, and avenues for future work are suggested.

Rights

©2024 Abigail Bowering. Access limited to the Smith College community and other researchers while on campus. Smith College community members also may access from off-campus using a Smith College log-in. Other off-campus researchers may request a copy through Interlibrary Loan for personal use.

Comments

90 pages: color illustrations. Includes bibliographical references (pages 79-81).

Share

COinS