Exoplanet Transit Candidate Identification in TESS Full-Frame Images via a Transformer-Based Algorithm

TL;DR

The paper tackles the challenge of identifying exoplanet transit candidates in TESS full-frame images without relying on phase folding or periodicity assumptions. It introduces a Transformer based architecture with CNN embeddings that processes flux, centroid, and background time series to detect transit signals directly from complete light curves. The approach achieves 214 candidate planetary system identifications across TESS sectors 1–26, including 122 multi-transit, 88 single-transit, and 4 multi-planet systems, highlighting the method's ability to detect nonperiodic and long period signals. Background and centroid information substantially improve performance, though Earth sized planets remain difficult to detect, pointing to avenues for future enhancements and applicability to upcoming missions.

Abstract

The Transiting Exoplanet Survey Satellite (TESS) is surveying a large fraction of the sky, generating a vast database of photometric time series data that requires thorough analysis to identify exoplanetary transit signals. Automated learning approaches have been successfully applied to identify transit signals. However, most existing methods focus on the classification and validation of candidates, while few efforts have explored new techniques for the search of candidates. To search for new exoplanet transit candidates, we propose an approach to identify exoplanet transit signals without the need for phase folding or assuming periodicity in the transit signals, such as those observed in multi-transit light curves. To achieve this, we implement a new neural network inspired by Transformers to directly process Full Frame Image (FFI) light curves to detect exoplanet transits. Transformers, originally developed for natural language processing, have recently demonstrated significant success in capturing long-range dependencies compared to previous approaches focused on sequential data. This ability allows us to employ multi-head self-attention to identify exoplanet transit signals directly from the complete light curves, combined with background and centroid time series, without requiring prior transit parameters. The network is trained to learn characteristics of the transit signal, like the dip shape, which helps distinguish planetary transits from other variability sources. Our model successfully identified 214 new planetary system candidates, including 122 multi-transit light curves, 88 single-transit and 4 multi-planet systems from TESS sectors 1-26 with a radius > 0.27 $R_{\mathrm{Jupiter}}$, demonstrating its ability to detect transits regardless of their periodicity.

Figures (14)

• Figure 1: (a) the transformer encoder, which processes time series inputs using positional encodings combined with input embeddings, and computes feature representations through a self-attention mechanism. (b) explains the self-attention mechanism, where the $Q$, $K$ matrices are used to calculate attention scores. These scores are then applied to the $V$ to produce the self-attention feature map.
• Figure 2: Schematic of the proposed architecture. The input includes flux $x_i$, centroid $c_i$, and background $b_i$ time series, which are concatenated into input embeddings $x'_i$ which are processed using convolutional embeddings. The tokens, along with positional encodings, are passed through a MSA mechanism within a transformer encoder. The features embedding produced by the transformer encoder are then passed through to average pooling, followed by a feed-forward MLP head to predict the class. The predicted output $\hat{y}$ is evaluated using a loss function $H(\hat{y}, y)$ to classify the input into one of two classes(0 or 1).
• Figure 3: CNN embedding, where the kernels slide across the input time series, transforming each local window into an embedding vector.
• Figure 4: General diagram for the identification of new exoplanet candidates based on model predictions. The blue boxes represent the inputs to the NN and the outputs after inference (candidates and transit vetting). The grey boxes are related to the NN, which includes the creation of the input representation, used to train the model and subsequently perform inference to find candidates. After inference, the resulting candidates are evaluated through transit vetting to identify potential exoplanets for follow-up observations.
• Figure 5: Distribution of single-planet multi-transit light curve candidate radii measured in Jupiter raidii.