DELOS: Contrastive Deep Learning for Low-SNR Blind Transit Searches in Kepler Photometry
Researchers have developed a deep-learning framework called DELOS that can detect shallow transits in Kepler photometry. The framework uses contrastive scoring to assign a transit-likeness score to each folded light curve, allowing for the identification of intermediate-to-long-period signals without relying on pre-detected threshold-crossing events. In experiments, DELOS achieved high accuracy and improved detection performance compared to existing methods. It also accelerat
Researchers have developed a deep-learning framework called DELOS that can detect shallow transits in Kepler photometry. The framework uses contrastive scoring to assign a transit-likeness score to each folded light curve, allowing for the identification of intermediate-to-long-period signals without relying on pre-detected threshold-crossing events. In experiments, DELOS achieved high accuracy and improved detection performance compared to existing methods. It also accelerates the search process by factors of 3-5 and 74-80 compared with Box-fitting Least Squares (BLS) and Transit Least Squares (TLS), respectively.
---
Why it matters: This matters because it provides a more efficient and sensitive method for detecting low-Signal-to-Noise Ratios (SNR) transit signals, which is crucial for identifying longer-period terrestrial planets in Kepler data. The framework's improved performance and speed can help researchers accelerate their search for these planets.
Source: https://arxiv.org/abs/2605.29428
This article was originally published at: https://arxiv.org/abs/2605.29428