Shallow Transits - Deep Learning: using deep learning to detect habitable planets
Monday 2 July, 15:10
Deep learning is currently taking the world of Artificial Intelligence by storm. Deep learning techniques already have proven success in varied fields, such as image processing, speech recognition and even drug discovery. Specifically, deep learning can provide new hope in needle-in-a-haystack problems, such as the detection of very faint signals in the presence of many kinds of noise. Detection of transiting terrestrial planets in the presence of stellar-activity red noise is one such problem. The non-linear nature of deep learning renders it completely different from traditional techniques (such as those based on the BLS). Such innovative approaches will be crucial in order to fully exploit the potential of future planet-detection space missions (TESS, PLATO). We hereby present an extremely short tutorial of what deep learning is, and how it can be applied to detect and analyze transiting terrestrial planets. We also introduce preliminary results of a feasibility study we have performed which demonstrate the immense capability of this novel and exciting approach.