Using Machine Learning for Exoplanet Classification

With aims of discovering potential candidates and using that new found information to analyze the composition of the world as we know now, many efforts have been made to conduct research efficiently and accurately. With multiple methods for exoplanet detection such as shadow searching, the data produced from these methods still require interpretation to reach a conclusion (e.g. is there a dip in the light curve?), which is why machine learning has recently come into the scene of astronomy with its potential to be trained for image classification tasks, requiring only a couple of seconds to a couple of minutes to complete their task. This paper used convolutional neural networks (CNN), a type of machine learning specifically designed to classify images, and achieved an area-under-curve coverage of 0.91.