Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Machine learning for the early classification of broad-lined Ic supernovae

Laura Cotter, Antonio Martin Carrillo, Joseph Fisher, Gabriel Finneran, Gregory Corcoran, Jennifer Lebron

TL;DR

This work tackles the challenge of rapidly classifying rare SNe Ic-BL using a novel early-epoch feature space built from magnitude rates computed from three initial photometric points. It evaluates multiple ML classifiers within a carefully balanced training scheme and validates the approach on SN Ic-BL and SN Ia populations, with Random Forest emerging as the most robust option. Results show promise in identifying a meaningful fraction of Ic-BL events (up to about 13% under certain conditions) but highlight generalisation challenges due to the small Ic-BL sample and label noise in current pipelines. The study emphasizes the practical impact of integrating this method with real-time survey data (e.g., LSST) to enable timely follow-up and improve the quality and completeness of Ic-BL datasets, thereby advancing SN-GRB connection studies.

Abstract

Science is currently at an age where there is more data than we know how to deal with. Machine learning (ML) is an emerging tool that is useful in drawing valuable science out of incomprehensibly large datasets, identifying complex trends in data that are otherwise overlooked. Moreover, ML can potentially enhance the quality and quantity of scientific data as it is collected. This paper explores how a new ML method can improve the rate of classification of rare Ic-BL supernovae (SNe). New parameters called magnitude rates were introduced to train ML models to identify SNe Ic-BL in large datasets. The same methodology was applied to a population of SN Ia transients to see if the methodology could be reproducible with another SN class. Three magnitudes, three time differences, two magnitude rates and the second derivative of these rates were calculated using the first three available photometric data points in a single filter. Initial investigations show that the Random Forest algorithm provides a strong foundation for the early classifications SNe Ic-BL and SNe Ia. Testing this model again on an unseen dataset shows that the model can identify upward of 13% of the total true SN Ic-BL population, significantly improving upon current methods. By implementing a dedicated observation campaign using this model, the number of SN Ic-BL classified and the quality of early-time data collected each year will see considerable growth in the near future.

Machine learning for the early classification of broad-lined Ic supernovae

TL;DR

This work tackles the challenge of rapidly classifying rare SNe Ic-BL using a novel early-epoch feature space built from magnitude rates computed from three initial photometric points. It evaluates multiple ML classifiers within a carefully balanced training scheme and validates the approach on SN Ic-BL and SN Ia populations, with Random Forest emerging as the most robust option. Results show promise in identifying a meaningful fraction of Ic-BL events (up to about 13% under certain conditions) but highlight generalisation challenges due to the small Ic-BL sample and label noise in current pipelines. The study emphasizes the practical impact of integrating this method with real-time survey data (e.g., LSST) to enable timely follow-up and improve the quality and completeness of Ic-BL datasets, thereby advancing SN-GRB connection studies.

Abstract

Science is currently at an age where there is more data than we know how to deal with. Machine learning (ML) is an emerging tool that is useful in drawing valuable science out of incomprehensibly large datasets, identifying complex trends in data that are otherwise overlooked. Moreover, ML can potentially enhance the quality and quantity of scientific data as it is collected. This paper explores how a new ML method can improve the rate of classification of rare Ic-BL supernovae (SNe). New parameters called magnitude rates were introduced to train ML models to identify SNe Ic-BL in large datasets. The same methodology was applied to a population of SN Ia transients to see if the methodology could be reproducible with another SN class. Three magnitudes, three time differences, two magnitude rates and the second derivative of these rates were calculated using the first three available photometric data points in a single filter. Initial investigations show that the Random Forest algorithm provides a strong foundation for the early classifications SNe Ic-BL and SNe Ia. Testing this model again on an unseen dataset shows that the model can identify upward of 13% of the total true SN Ic-BL population, significantly improving upon current methods. By implementing a dedicated observation campaign using this model, the number of SN Ic-BL classified and the quality of early-time data collected each year will see considerable growth in the near future.
Paper Structure (21 sections, 8 equations, 5 figures, 9 tables)

This paper contains 21 sections, 8 equations, 5 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Motivation and tests on ALeRCE's classifier
  3. ALeRCE Data Processing
  4. Reliability of ALeRCE's SNIbc classification
  5. Monitoring of ALeRCE's ML classification
  6. Exploration of a new parameter space for Machine Learning supernova classification
  7. Approaching New Machine Learning Methods
  8. Light curve fitting
  9. Comparing Magnitude Rates of SN classes
  10. Supernova sample
  11. Methodology
  12. Using scikit-learn
  13. Machine Learning Algorithms
  14. Machine Learning Implementation
  15. Addressing Class Imbalance
  16. ...and 6 more sections

Figures (5)

  • Figure 1: Fits for the median magnitude rising for different SN classes rates 5 days before peak.
  • Figure 2: Results from the training and validation sample after 500 runs using the 50-50 distribution of yes and no, with each run containing 476 transients.
  • Figure 3: Results from the real-life testing scenario from the 500 runs using a 50-50 distribution. The real-life dataset consisted of 100 transients in the SN Ia scenario and 65 transients in the SN Ic-BL scenario.
  • Figure 4: Results from the real-life dataset testing scenario from the 500 runs using a 70-30 distribution. The real-life dataset consisted of 100 transients in the SN Ia scenario and 65 transients in the SN Ic-BL scenario.
  • Figure 5: F1 scores from Random Forest models vs imbalance in dataset.