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

2026-07-06

AI Summary

This work addresses the challenges in classifying transient candidates from time-domain surveys, where reliance on costly human annotations leads to high label noise and poor generalization. To overcome these limitations, the authors propose a weakly supervised deep learning framework that operates without manual labels. Training data are constructed by injecting simulated transients into real non-transient images. The method employs an asymmetric co-teaching strategy with dual networks to handle severe class contamination and integrates a lightweight hybrid uncertainty quantification approach—combining Monte Carlo Dropout and deep ensembles—to produce well-calibrated predictions. Experiments demonstrate that the model maintains high performance under extreme label noise, accurately recovers light-curve classifications, and yields uncertainty estimates comparable to those of computationally expensive ensemble methods. Furthermore, it uncovers latent substructures within false-positive samples, significantly enhancing model interpretability and cross-survey transferability.

Abstract

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.

Problem

• Real-Bogus classification
• label-free learning
• uncertainty quantification
• class contamination
• time-domain surveys

Innovation

• label-free learning
• asymmetric co-teaching
• uncertainty quantification
• dual-network architecture
• transient injection