Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

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

Source

arXiv

arxiv.org/abs/2607.05393v1

Read the source

Source

  • Source name: arXiv
  • Source kind: paper
  • Authors / organization: Raphaël Bonnet-Guerrini, Bruno Sanchez, Dominique Fouchez, Benjamin Racine, Maya Guy, Mariam Sabalbal, Manal Yassine, Vincenzo Piuri
  • Venue: arXiv: astro-ph.IM
  • Published: 2026-07-06
  • Collected: 2026-07-07
  • URL: http://arxiv.org/abs/2607.05393v1
  • PDF: https://arxiv.org/pdf/2607.05393v1
  • DOI: UNKNOWN

Plain-Language Brief

Researchers developed a machine learning model that can classify transient candidates in time-domain surveys, known as Real-Bogus classification, without needing human-labeled data. The model is trained using simulated transient injections and contaminated survey data, and it provides calibrated uncertainty quantification. It performs well on a benchmark dataset and recovers light-curve classes with high fidelity, but single-source identification is limited by the ambiguity of light-curve-derived labels. The researchers propose a hybrid strategy for uncertainty quantification that improves calibration relative to more expensive ensemble baselines.

What the Document Says

  • The study aims to create a Real-Bogus classification framework without human-labeled data using injected transients and bogus-dominated survey data.
  • The authors combine simulated transient injections with contaminated survey class data and train a dual-network model using asymmetric co-teaching for classes with different label-noise levels.
  • The method achieves strong Real-Bogus performance on the labeled subset and remains stable under severe class contamination.
  • The researchers evaluate the performance on a benchmark subset and analyze the learned representation with latent-space visualization tools.
  • For uncertainty quantification (UQ), they compare MC dropout and deep ensembles and propose a low-cost hybrid strategy exploiting the dual-network setting.
  • They extend the evaluation to the light-curve domain to assess recovery of light-curve classes.
  • The method recovers transient light-curve classes with high fidelity, but single-source identification is limited by ambiguity in light-curve-derived labels.
  • The hybrid UQ approach achieves competitive calibration relative to more expensive ensemble baselines.
  • Latent-space analyses indicate that uncertainty aligns with the decision boundary and reveal subclasses within the bogus population.
  • The researchers conclude that injection-driven, weakly supervised training can enable scalable and consistent Real-Bogus classification without human-labeled training data while providing calibrated uncertainties.

Key Concepts Mentioned

  • Real-Bogus classification
  • machine learning model
  • injected transients
  • bogus-dominated survey data
  • asymmetric co-teaching
  • MC dropout
  • deep ensembles
  • uncertainty quantification (UQ)
  • light-curve classes
  • single-source identification

Claims Made by the Document

  • The proposed method can perform Real-Bogus classification without human-labeled data.
  • The method provides calibrated uncertainty quantification.

Evidence Used

  • Combining simulated transient injections with a contaminated survey class and training a dual-network model using asymmetric co-teaching for classes with different label-noise levels.
  • Achieving strong Real-Bogus performance on the labeled subset and remaining stable under severe class contamination.
  • Recovering transient light-curve classes with high fidelity.
  • The hybrid UQ approach achieving competitive calibration relative to more expensive ensemble baselines.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Time-domain surveys generate many transient candidates, making Real-Bogus classification a critical step in automated discovery pipelines. Reliable labels are costly, while community labels can be noisy and survey-dependent. We aim to develop a Real-Bogus classification framework that can be trained without human-labeled data using injected transients and bogus-dominated survey data, remains robust under strong class contamination, and provides calibrated uncertainty quantification. We combine simulated transient injections with a contaminated survey class and train a dual-network model using asymmetric co-teaching for classes with different label-noise levels. We evaluate performance on a benchmark subset and analyze the learned representation with latent-space visualization tools. For uncertainty quantification (UQ), we compare MC dropout and deep ensembles and propose a low-cost hybrid strategy that exploits the dual-network setting to improve calibration. We extend the evaluation to the light-curve domain to assess recovery of light-curve classes. 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 limited by ambiguity in light-curve-derived labels. Our hybrid UQ approach achieves competitive calibration relative to more expensive ensemble baselines. Latent-space analyses indicate that uncertainty aligns with the decision boundary and reveal subclasses within the bogus population. 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 method is suited for transfer to forthcoming surveys by re-running the injection-based training pipeline.