Interpretable Human-Label-Free Deep Learning for Real-Bogus Classification with Uncertainty Quantification
arXiv:2607.05393v1 Announce Type: cross 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
The increasing volume of data from time-domain surveys in astronomy necessitates more efficient and automated classification methods to handle transient candidates.
This development allows for robust deep learning classification of astronomical phenomena without relying on expensive human-labeled data, addressing a major bottleneck in discovery pipelines.
The ability to train AI models for 'Real-Bogus' classification using simulated and contaminated data, along with uncertainty quantification, improves the reliability and independence of astronomical discovery.
- · Astronomical research institutions
- · Time-domain survey operators
- · AI/ML developers specializing in astrophysics
- · Manual data labeling services for astronomy
- · Legacy classification pipelines relying solely on human expertise
More efficient and accurate identification of new astronomical transients, leading to faster scientific discoveries.
Reduced operational costs for astronomy surveys due to less reliance on human intervention for initial data vetting.
The methodology could be adapted to other scientific fields facing similar challenges with high-volume, unlabeled or noisy data.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI