Interpretable Human-Label-Free Deep Learning for Real-Bogus Classification with Uncertainty Quantification - Research paper

Sep 21, 2026

Uploaded by Eden

Time-domain surveys produce vast transient candidate streams, making Real-Bogus classification essential for automated discovery while human-labeled training sets remain costly and community labels can be noisy and survey dependent. The work builds a weakly supervised framework trained without human-labeled data using physically motivated injected transients and bogus-dominated survey data. It uses an asymmetric co-teaching dual-network strategy under strong class contamination and delivers calibrated uncertainty quantification for robust downstream decisions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages20

Reading Time50 min

AccessFree to download

1. Introduction

Difference Image Analysis and bogus sources

Need for early transient-bogus discrimination

Abstract

Context, Aims, Methods, Results, Conclusions

Astronomy & Astrophysics manuscript no. main ©ESO 2026 July 7, 2026 .05393v1 [ astro-ph .IM] 6 Jul 2026 Interpretable Human-Label-Free Deep Learning for Real-Bogus Classification with Uncertainty Quantification Raphaël Bonnet-Guerrini 1, 2 , Bruno Sanchez2 , Dominique Fouchez2 , Benjamin Racine2 , Maya Guy3, 4 , Mariam Sabalbal5 , Manal Yassine2 , and Vincenzo Piuri 1 1 Università degli Studi di Milano, Department of Computer Science, Milan 20133, Italy e-mail: [raphael.bonnet-guerrini@unimi.it](raphael.bonnet-guerrini@unimi.it) ⋆ 2 Aix Marseille Univ, CNRS/IN2P3, CPPM, Marseille, France 3 Université Côte d’Azur, INRIA, CNRS, Laboratoire J.A.Dieudonné, Maasai team, Nice 06000, France 4 Université Côte d’Azur, Observatoire de la Côte d’Azur, CNRS, Laboratoire Lagrange, Nice 06000 France 5 Université de Liège, STAR Institute, Liège 4000, Belgium ABSTRACT Context. Time-domain surveys generate large numbers of transient candidates, making Real-Bogus classification a critical step in automated discovery pipelines. Obtaining reliable human-labeled training sets is costly, while community-provided labels can be noisy and survey-dependent. Aims. We aim to develop a Real-Bogus classification framework that can be trained without human-labeled data using physically motivated injected transients and bogus-dominated survey data, that remains robust under strong class contamination, and that provides well-calibrated and reliable uncertainty quantification suitable for downstream decision-making. Methods. We combine simulated transient injections with a heavily contaminated survey class and train a dual-network model using an asymmetric co-teaching strategy for classes with different label-noise levels. We evaluate performance on a labeled benchmark subset and analyze the learned representation with latent-space visualization tools. For uncertainty quantification (UQ), we compare standard approaches (MC dropout and deep ensembles) and propose a low-cost hybrid strategy that exploits the dual-network setting to improve calibration. We further extend the evaluation to the light-curve domain to assess recovery of light-curve classes. Results. The method achieves strong Real-Bogus performance on the labeled subset and remains stable under severe class contamination. It recovers transient light-curve classes with high fidelity, while single-source identification is fundamentally limited by intrinsic ambiguity in light-curve-derived labels. Our hybrid UQ approach achieves competitive calibration relative to substantially more expensive ensemble baselines. Latent-space analyses indicate that uncertainty aligns with the decision boundary and reveal that the model is able to identify subclasses within the bogus population. Conclusions. Our results show that injection-driven, weakly supervised training can enable scalable and consistent Real-Bogus classification without human-labeled training data while providing calibrated uncertainties. The methodology is well suited for transfer to forthcoming surveys by re-running the injection-based training pipeline. Key words. Astronomical instrumentation, methods and techniques-Methods: data analysis-Methods: statistical arXiv:2607 1. Introduction The new Vera C. Rubin Legacy Survey of Space and Time (LSST) (Abell et al. 2009 ; Ivezi et al. 2019) features a time-domain component that will detect transients throughout its 10-year run using Difference Image Analysis (DIA) LSST Science Collaboration et al. (2009); Ivezi et al. (2019); Liu et al. (2024) . In DIA, a template image is subtracted from a newly observed image to identify changes associated with new or variable sources (Alard & Lupton 1998 ; Alard 2000) . This method is crucial for detecting transient events in crowded fields or under varying observational conditions. Typically, supernovae occupy the short-duration region of the luminosity-timescale phase space and therefore benefit from high-cadence observations

FAQs

Why is Real-Bogus classification critical for time-domain surveys like LSST?

Real-Bogus discrimination is needed to suppress spurious DIA detections early so pipelines can prioritize true candidates before more expensive light-curve analysis.

How does the proposed method avoid relying on human-labeled training data?

It trains on physically motivated injected transients and bogus-dominated survey data using a weakly supervised dual-network framework rather than human-labeled examples.

What approach is used for uncertainty quantification, and how is it evaluated?

It compares standard UQ methods (MC dropout and deep ensembles) and introduces a low-cost hybrid strategy leveraging the dual-network setting; evaluation includes calibration comparisons to more expensive ensemble baselines.

Sep 21, 2026

Uploaded by Eden