Researchers have developed a novel deep learning framework for classifying astronomical transients as real or bogus without requiring human-labeled data. This method utilizes injected simulated transients and a contaminated survey dataset, employing asymmetric co-teaching to handle varying label noise levels. The framework also incorporates a hybrid uncertainty quantification strategy, combining MC Dropout and deep ensembles, to provide calibrated confidence in its classifications. This approach aims to enable scalable and consistent Real-Bogus classification for time-domain surveys, even with noisy or limited labels. AI
IMPACT Enables scalable and consistent astronomical transient classification without human-labeled data, improving efficiency in time-domain surveys.
RANK_REASON The cluster contains a research paper published on arXiv detailing a new deep learning methodology.
- alphaXiv
- arXiv
- CatalyzeX Code Finder for Papers
- DagsHub
- Deep Ensembles
- Deep learning
- Gotit.pub
- MC Dropout
- Raphael Bonnet Guerrini
- Uncertainty quantification
- Zwicky Transient Facility
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
