Presenter: Tiziano ZINGALES
Abstract:
Artificial Intelligence has become an increasingly popular tool in the study of exoplanets. Generative Adversarial Networks (GANs) are a class of machine learning algorithms able to learn from any dataset. Here I show two possible applications for Deep Convolutional GANs, to help us characterizing exoplanetary atmospheres and detecting new non-transiting planets.
Atmospheric retrievals: they usually involve computationally intensive Bayesian sampling methods. Large parameter spaces and increasingly complex atmospheric models create a computational bottleneck forcing a trade-off between statistical sampling accuracy and model complexity. It is especially true for upcoming JWST and ARIEL observations. I introduce ExoGAN, the Exoplanet Generative Adversarial Network, a new deep learning algorithm able to recognise molecular features, atmospheric trace-gas abundances and planetary parameters using unsupervised learning. Once trained, ExoGAN is widely applicable to a large number of instruments and planetary types. The ExoGAN retrievals constitute a significant speed improvement over traditional retrievals and can be used either as a final atmospheric analysis or provide prior constraints to subsequent retrieval.
Detection of non-transiting planets: their phase curve modularity can give us some hint of their properties. DCGANs can also be trained to detect and study non-transiting planets, starting from their orbital phase curves.