Reducing false positives in strong lens detection through effective augmentation and ensemble learning

Monthly Notices of the Royal Astronomical Society, vol. 538, pp. 1081–1095

Abstract

ABSTRACT This research studies the impact of high-quality training data sets on the performance of Convolutional Neural Networks (CNNs) in detecting strong gravitational lenses. We stress the importance of data diversity and representativeness, demonstrating how variations in sample populations influence CNN performance. In addition to the quality of training data, our results highlight the effectiveness of various techniques, such as data augmentation and ensemble learning, in reducing false positives while maintaining model completeness at an acceptable level. This enhances the robustness of gravitational lens detection models and advancing capabilities in this field. Our experiments, employing variations of DenseNet and EfficientNet, achieved a best false positive rate (FP rate) of $10^{-4}$, while successfully identifying over 88 per cent of genuine gravitational lenses in the test data set. This represents an 11-fold reduction in the FP rate compared to the original training data set. Notably, this substantial enhancement in the FP rate is accompanied by only a 2.3 per cent decrease in the number of true positive samples. Validated on the Kilo Degree Survey data set, our findings offer insights applicable to ongoing missions, like Euclid.

Authors 6

1. Samira Rezaei corresponding Aachen Leiden Institute of Advanced Computer Science Leiden Observatory

   Leiden University · Leiden Observatory

   Affiliation as printed

   Leiden Institute of Advanced Computer Science (LIACS), Leiden University , NL-2333 CC Leiden ,

   Leiden Observatory, Leiden University , NL-2333 CC Leiden ,

   Leiden Institute of Advanced Computer Science (LIACS), Leiden University , 2333 CC Leiden ,

   Leiden Observatory, Leiden University , 2333 CC Leiden ,

2. Amirmohammad Chegeni Aachen Leiden Observatory

   Leiden University · University of Padua · Istituto Nazionale di Fisica Nucleare, Sezione di Padova · Leiden Observatory

   Affiliation as printed

   Dipartimento di Fisica e Astronomia ‘G. Galilei’, Universit‘a di Padova , Via Marzolo 8, I-35131 Padova ,

   INFN-Padova , Via Marzolo 8, I-35131 Padova ,

   Leiden Observatory, Leiden University , NL-2333 CC Leiden ,

   Dipartimento di Fisica e Astronomia ‘G. Galilei’, Universit‘a di Padova , Via Marzolo 8, 35131 Padova ,

   INFN-Padova , Via Marzolo 8, 35131 Padova ,

   Leiden Observatory, Leiden University , 2333 CC Leiden ,

3. B.C Nagam

   University of Groningen

   Affiliation as printed

   Kapteyn Astronomical Institute, University of Groningen , Postbus 800, NL-9700 AV Groningen ,

   Kapteyn Astronomical Institute, University of Groningen , Postbus 800, NL9700 AV Groningen ,

4. J. P. McKean

   University of Groningen · South African Radio Astronomy Observatory · University of Pretoria

   Affiliation as printed

   Department of Physics, University of Pretoria , Lynnwood Road, Hatfield, Pretoria, 0083 ,

   Kapteyn Astronomical Institute, University of Groningen , Postbus 800, NL-9700 AV Groningen ,

   South African Radio Astronomy Observatory (SARAO) , P.O. Box 443, Krugersdorp 1740 ,

   Kapteyn Astronomical Institute, University of Groningen , Postbus 800, NL9700 AV Groningen ,

5. Mitra Baratchi Aachen Leiden Institute of Advanced Computer Science

   Leiden University

   Affiliation as printed

   Leiden Institute of Advanced Computer Science (LIACS), Leiden University , NL-2333 CC Leiden ,

   Leiden Institute of Advanced Computer Science (LIACS), Leiden University , 2333 CC Leiden ,

6. Léon V. E. Koopmans

   University of Groningen

   Affiliation as printed

   Kapteyn Astronomical Institute, University of Groningen , Postbus 800, NL-9700 AV Groningen ,

   Kapteyn Astronomical Institute, University of Groningen , Postbus 800, NL9700 AV Groningen ,