Harnessing the power of CNNs for unevenly-sampled light-curves using Markov transition field

• Code repository: github.com/Buguemar/PIIC19/

This work introduces a deep learning approach for identifying exoplanet candidates from unevenly sampled light curves without relying on handcrafted features, metadata, or fixed time-series formats. The method transforms variable-length, irregularly sampled light curves into fixed-size two-channel images using Markov Transition Fields (MTF), which are then classified with a convolutional neural network. Experiments on the Kepler Mission dataset show that the proposed approach achieves competitive performance compared with state-of-the-art methods while remaining simpler and faster. The results also demonstrate that MTF representations can serve as effective standalone data products for analyzing irregularly sampled transient light curves, offering a practical alternative for large-scale automated exoplanet detection.

Recommended citation: Bugueño, Margarita, et al. "Harnessing the power of CNNs for unevenly-sampled light-curves using Markov transition field." Astronomy and Computing 35 (2021): 100461.
Download Paper