Venue: GO Jones LG1
The field of transiting extrasolar planets and especially the study of their atmospheres is one of the youngest and most dynamic subjects in current astrophysics. Permanently at the edge of technical feasibility, we are successfully discovering and characterising smaller and smaller planets. To study exoplanetary atmospheres, we typically require a 10e-4 to 10e-5 level of accuracy in flux. Achieving such a precision has become the central challenge to exoplanetary research and is often impeded by systematic (nongaussian) noise from either the instrument, stellar activity or both. This is particularly true for current instrumentation and detectors that do not feature high stabilities in time nor well characterised response functions. In recent times, much has been debated on how to calibrate and de-trend one's data leading to various controversies in the field over the use of instrument information as effective means of de-correlation.
A promising solution to the current controversies is the use of non-parametric or blind de-trending. Assuming no prior nor auxiliary information of the data and instrument at all, I will show in this seminar that we can nonetheless rid our data from instrument and stellar systematics using statistical machine-learning techniques. Such an unbiased approach to data analysis does not only achieve a higher degree of objectivity but can also be shown to be more versatile than conventional techniques. In this seminar, I will give an overview of exoplanetary spectroscopy, the problems we are currently facing and discuss the use of unsupervised machine-learning techniques as promising alternatives to the controversy prone parametric data de-trending approach.