Hunting for “Oddballs” with Machine Learning: Detecting Anomalous Exoplanets Using a Deep-learned Low-dimensional Representation of Transit Spectra with Autoencoders

Abstract This study explores the application of autoencoder-based machine learning techniques for anomaly detection to identify exoplanet atmospheres with unconventional chemical signatures using a low-dimensional data representation. We use the Atmospheric Big Challenge database, a publicly available dataset with over 100,000 simulated exoplanet spectra, to construct an anomaly detection scenario by defining CO 2 -rich atmospheres as anomalies and CO 2 -poor atmospheres as the normal class. We benchmarked four different anomaly detection strategies: autoencoder reconstruction loss, one-class support vector machine, K -means clustering, and local outlier factor. Each method was evaluated in both the original spectral space and the autoencoder’s latent space using receiver operating characteristic curves and area-under-the-curve metrics. To test the performance of the different methods under realistic conditions, we introduced Gaussian noise levels ranging from 10 to 50 ppm. Our results indicate that anomaly detection is consistently more effective when performed within the latent space across all noise levels. Specifically, K -means clustering in the latent space emerged as a stable and high-performing method, being robust to noise levels up to 30 ppm and remaining viable even at 50 ppm. On the other hand, the performance of the anomaly detection methods applied directly in the raw spectral space degrades significantly with increasing level of noise. We also considered dimensionality reduction via principal component analysis and showed that its performance is inferior relative to the deep-learned latent representation. This suggests that autoencoder-driven dimensionality reduction offers a robust methodology for flagging chemically anomalous targets in large-scale surveys.