🤖 AI Summary
This study addresses the challenge of detecting single-epoch exoplanet transits, which are inherently non-periodic and thus evade conventional transit detection methods reliant on periodic signals. The authors propose a self-supervised world model based on a Transformer architecture that learns the intrinsic stellar light-curve behavior through masked prediction. By identifying anomalous transit-like signals via prediction residuals—without requiring phase folding—the method enables effective detection of singular transit events. This approach achieves, for the first time, reliable identification of single transits, supports zero-shot transfer across tasks, and incorporates conformal prediction to enhance result reliability. Evaluated on Kepler data, the model attains an AUC of 0.938, recovers 32% of injected single transits with a depth of 1000 ppm, uncovers 179 new blind-search candidates, and demonstrates 100% zero-shot recovery of known TESS planets.
📝 Abstract
I present EXOVEIL, a transit detection system that learns what a star's brightness should look like and flags when reality disagrees. Unlike existing systems that require phase-folded input, EXOVEIL operates on raw flux time series and can detect planets that transit only once.A Transformer world model, trained on 16,499 Kepler light curves with transit-masked self-supervised learning, predicts expected stellar flux. A matched-filter detector with variance weighting extracts transit signals from the prediction residuals. A learned classifier (XGBoost) separates planets from false positives, achieving AUC 0.938 on Kepler DR25. Applied to single-transit injection-recovery, EXOVEIL recovers 32% of transits at 1000 ppm depth a task where all classification-based systems score 0% by construction. A blind search of 3,737 Kepler stars yields 179 new transit-like signals not present in the DR25 TCE catalogue, including 46 monotransit candidates. Applied withoutretraining to 47 confirmed TESS planets in the PLATO LOPS2 field, EXOVEIL achieves 100% recovery, demonstrating zero-shot cross-mission transfer. At PLATO's 25-second cadence, detection reaches 100 ppm -- approaching the Earth-analog regime. I provide the first application of conformal prediction to transit detection (95.9% empirical coverage) and release the system as pip install exoveil with pretrained weights and a candidate catalogue.