Detection of Exoplanet Systems In Kepler Light Curves Using Adaptive Neuro-Fuzzy System
R. M. Asif Amin, Abu Talha Khan, Zareen Tasnim Raisa, Nawar Chisty, Sumayra SamihaKhan, Mohammad Sultan Khaja, Rashedur Mohammad Rahman · 2018
The paper investigates the usability of fuzzy inference systems, namely ANFIS to assist in the detection of exoplanet transit events from time-series data of Flux intensity values from host stars. We use statistical properties of time curves, principal component analysis, and dynamic time warping (DTW) to extract features from time-series data. We propose the use of aggregated DTW distance to all positive training samples as a feature and show that it significantly increases performance if used along with statistical features and principal components. We propose that fuzzy methods can be used to classify exoplanet systems with a relatively low false positive rate and thus, can be used to filter out possible flux signatures for review by human experts.