OASIS: On-the-fly ASynchronous Image Stream classification -- Transformer-based models for the early detection of strongly lensed supernovae
Abstract
Time-domain surveys often produce asynchronous image sequences across irregular observation times and photometric bands. In many applications, predictions are needed after each new observation, particularly when transient sources must be identified early to trigger follow-up. Strongly lensed supernovae (LSNe), for which follow-up is particularly time-critical, provide a useful example. We formulate this setting as On-the-fly ASynchronous Image Stream (OASIS) classification and explore reusable transformer architectures for causal, on-the-fly classification. We demonstrate these architectures on the lensed-transient classification problem studied by Bag et al. (2026a,b), distinguishing LSNe Ia from unlensed astrophysical transients. We compare them with the recurrent baseline adopted by Bag et al. (2026b), which combines a ConvLSTM branch for image processing with a separate LSTM branch for encoding observation times. The baseline maps asynchronous observations onto a fixed band-by-observation grid, filling unobserved bands with zeros. Using a common training setup, we compare two transformer-based approaches to modeling interactions among spatio-temporal tokens: factorized attention, which separates spatial and temporal interactions, and joint attention, which models them together. On the demonstration dataset, factorized architectures achieve mean losses of $0.095−−0.100$, compared with $0.128$ for the recurrent baseline and $0.107$ or higher for joint architectures. A selective readout further reduces the best loss to $0.089$ for a factorized model. These results suggest that factorized attention suits this dataset better, without implying universal architectural superiority. We release the architectures as reusable foundations for time-domain astronomy and other applications involving evolving spatial features that require on-the-fly predictions.
Related
- Equation of State Classification of Core-Collapse Supernovae with a Network of Gravitational-Wave Detectors
- A Fast and Scalable Transformer Pipeline for Binary Black Hole Detection
- Anomaly detection in Fink. I. Discovery, follow-up, and classification of unusual sources
- Photometric classification and out-of-taxonomy detection of ZTF transients with FLARE
- First Detection of the Van Hoof Effect in First-overtone RR Lyrae Stars: A New Kinematic Classification of Shock Waves
BibTeX
@misc{bag2026oasis,
title = {{OASIS: On-the-fly ASynchronous Image Stream classification -- Transformer-based models for the early detection of strongly lensed supernovae}},
author = {Satadru Bag and Raoul Canameras and Sherry H. Suyu and Stefan Schuldt and Stefan Taubenberger and Irham Taufik Andika and Alejandra Melo},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.17168},
eprint = {2610.17168},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.17168},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - OASIS: On-the-fly ASynchronous Image Stream classification -- Transformer-based models for the early detection of strongly lensed supernovae
AU - Satadru Bag
AU - Raoul Canameras
AU - Sherry H. Suyu
AU - Stefan Schuldt
AU - Stefan Taubenberger
AU - Irham Taufik Andika
AU - Alejandra Melo
PY - 2026
DA - 2026/10/12
PB - arXiv
DO - 10.48550/arxiv.2610.17168
UR - https://arxiv.org/abs/2610.17168
AB - Time-domain surveys often produce asynchronous image sequences across irregular observation times and photometric bands. In many applications, predictions are needed after each new observation, particularly when transient sources must be identified early to trigger follow-up. Strongly lensed supernovae (LSNe), for which follow-up is particularly time-critical, provide a useful example. We formulate this setting as On-the-fly ASynchronous Image Stream (OASIS) classification and explore reusable transformer architectures for causal, on-the-fly classification. We demonstrate these architectures on the lensed-transient classification problem studied by Bag et al. (2026a,b), distinguishing LSNe Ia from unlensed astrophysical transients. We compare them with the recurrent baseline adopted by Bag et al. (2026b), which combines a ConvLSTM branch for image processing with a separate LSTM branch for encoding observation times. The baseline maps asynchronous observations onto a fixed band-by-observation grid, filling unobserved bands with zeros. Using a common training setup, we compare two transformer-based approaches to modeling interactions among spatio-temporal tokens: factorized attention, which separates spatial and temporal interactions, and joint attention, which models them together. On the demonstration dataset, factorized architectures achieve mean losses of $0.095−−0.100$, compared with $0.128$ for the recurrent baseline and $0.107$ or higher for joint architectures. A selective readout further reduces the best loss to $0.089$ for a factorized model. These results suggest that factorized attention suits this dataset better, without implying universal architectural superiority. We release the architectures as reusable foundations for time-domain astronomy and other applications involving evolving spatial features that require on-the-fly predictions.
ER -