NovaQuest

The Project

A supernova is an explosive event marking the end of a massive star's life, scattering heavy elements and influencing star formation. With modern surveys like the Dark Energy Survey, vast supernova datasets are being generated. NovaQuest aims to compare different machine learning algorithms for supernova classification, focusing on the Convolutional Neural Network (CNN) for its effectiveness in utilizing flux characteristics. NovaQuest provides a preview of a future where real-time classification is going to revolutionize the research on supernovae. Target users, including astrophysicists, astronomers, researchers, and educational institutions, can gain from improved supernova classification, deepening our understanding of the universe.

View this project on YouTube

Team Comments

I chose this project due to my fascination with astronomy, particularly supernovae, and a recognized gap in research on applying diverse machine learning algorithms for classification. Leveraging my experience in exoplanet research, I aimed to merge astronomy with machine learning.

Balancing the dataset, dominated by one supernova type, posed an initial hurdle, which I tackled using Gaussian noise interpolation for a more balanced distribution. Fine-tuning the Convolutional Neural Network model required persistent experimentation and research to iteratively adjust parameters.

With more time, I'd explore additional machine learning algorithms for much effective supernova classification, integrate supplementary data like host galaxy properties and redshift data for further accuracy, and optimize model efficiency for real-time classification on larger datasets.