קולוקוויום בחוג לגאופיזיקה: Stellar activity characterization using deep learning for extra-solar planet detection

Zoom: https://tau-ac-il.zoom.us/j/83800936221?pwd=dHQ5b0pYdWV3SzN1amNPanRQUnc4QT09

Abstract:

Advancements in photometry and spectroscopy allowed the detection of signals from Earth-like planets orbiting Sun-like stars. However, the persistent challenge lies in stellar activity, which can obscure planetary signals. My work envisions characterizing stars by registering the hyperparameters of the Gaussian process (GP) kernels representing stellar activity, using neural networks.

While GPs excel in characterizing stellar activity, conventional methods like Markov Chain Monte Carlo (MCMC) for the estimation of their hyperparameters are slow and computationally intensive.Therefore, we explore the flexibility and efficiency of neural networks for this task, demonstrating their feasibility for GP hyperparameter estimation. My lecture focuses on three main parts: estimating hyperparameters for the squared exponential, periodic, and quasi-periodic kernels.

Notably, neural networks, with their proven capacity to decipher complex and non-linear patterns, emerge as promising tools for characterizing stellar activity and detecting exoplanets. This is particularly crucial in challenging cases, such as identifying Earth-sized planets within the habitable zones of Sun-like stars.

​

מארגן האירוע: ד"ר רועי ברקן