Thesis Defense - University of Houston
Skip to main content

Thesis Defense

In Partial Fulfillment of the Requirements for the Degree of Master of Science

Renuka Pampana

will defend her thesis

A New Approach To Domain Adaptation Applied To Supernova Photometric Classification


Supernovae play a vital role in measuring cosmological parameters. They are used as ‘standard candles’ for measuring extragalactic distances. There are other types of supernovae like supernova Ib and Ic that closely resemble supernova Ia (but are not as useful as supernova Ia). Large telescopic surveys captures light curve s of these supernovae events referred as photometric observations, which includes all the three types. Thus, accurate classification of supernovae from these photometric observations is desirable for proper calculation of cosmological parameters.

The existing method for classification of supernova photometric observations is based on spectroscopic method, which is very cumbersome and expensive. In future, with the increase in photometric surveys, myriad number of supernova photometric observations are expected. Thus, an efficient method for the classification of supernovae is required to replace existing methods. We also wanted to take advantage of existing dataset classified by spectroscopic method for the classification of upcoming photometric dataset. Since, these two dataset belongs to different domains, an adaptive mechanism across the domains is required. Thus, we propose a method to generate a predictive model using domain adaptation with active learning that will classify supernova (Ia, Ib, Ic) using spectroscopic data (aka source data) as a training set and photometric data (aka target data) as a testing set. Our method includes two concepts of machine learning: 1. Domain adaptation technique is used to transfer source domain information to target domain. 2. Active learning technique is used to rely on only few target domain l abels in a non-uniform distribution to build an effective model. The experiments and results show that our method outperforms various domain adaptation techniques with significant increase in classification accuracy.

Date: Tuesday, April 19, 2016
Time: 1:00 PM
Place: HBS 350
Advisor: Dr. Ricardo Vilalta

Faculty, students, and the general public are invited.