Publication Date

Spring 2020

Degree Type

Master's Project

Degree Name

Master of Science (MS)


Computer Science

First Advisor

Robert Chun

Second Advisor

Thomas Austin

Third Advisor

Aaron Romanowsky


Machine Learning, Astronomy, Stellar Spectra, Classification, Multi-layer Perceptrons, K-fold Cross-Validation, Sampling, Hidden Layers, Neural Network


Lightyears beyond the Planet Earth there exist plenty of unknown and unexplored stars and Galaxies that need to be studied in order to support the Big Bang Theory and also make important astronomical discoveries in quest of knowing the unknown. Sophisticated devices and high-power computational resources are now deployed to make a positive effort towards data gathering and analysis. These devices produce massive amount of data from the astronomical surveys and the data is usually in terabytes or petabytes. It is exhaustive to process this data and determine the findings in short period of time. Many details can be missed out and can lead to increased errors. Machine Learning can thus be applied for automated intelligent data analysis and recognition in the field of astronomy to gather important information and recognize or classify star types. Celestial Spectral Classification is one such problem that needs to be addressed using Machine Learning and will help astronomers to know whether the classified star has particular physical or chemical properties. Machine Learning can help astronomers to determine the class of celestial spectra which in turn can help in determining various properties of the star and will make the classification process intelligent, automated and less cumbersome.