Machine Learning for Enhanced Lightcurve Classification

You are here

Project Description: 

Lightcurve classification is a crucial task in astronomy for identifying variable stars and detecting exoplanet transits. Traditional approaches rely on human expertise and classical algorithms (e.g. periodicity searches) to classify lightcurves, but these methods can be time-consuming and less effective with large datasets. Modern machine learning (ML) techniques have the potential to improve accuracy and automate the classification of lightcurves from large-scale surveys. This project will focus on the KELT-South telescope's commissioning dataset (a 46-day run in January– February 2010) to explore how ML can enhance classification accuracy and validate existing non-ML methods. By applying ML-based techniques and comparing them to traditional methods, the project aims to identify various types of variable stars (such as pulsating stars and eclipsing binaries) and potential exoplanet transit events more effectively.
Research Area: 
Astronomy
Project Level: 
Masters
This Project Is Offered At The Following Node(s): 
(UCT)
Special Requirements: 
1. Python Programming: Strong proficiency in Python for scientific computing. Experience with libraries such as NumPy and pandas for data manipulation, and matplotlib or other libraries for basic visualisation of lightcurves. 2. Astronomy Background: Basic understanding of stellar variability and exoplanet transits to make informed decisions on feature extraction and to interpret model results in a physically meaningful way. Prior exposure to time-series data analysis (e.g., Fourier transforms, period finding) would be helpful for comparing ML methods with classical approaches.

Supervisor

Dr
Rudi
Kuhn
E-mail Address: 
Affiliation: 
South African Astronomical Observatory (SAAO)

Co-Supervisor

Documents: 
PDF icon Machine Learning for Enhanced Lightcurve Classification