I am an Astronomer at the European Southern Observatory (ESO), stationed at the Headquarters in Garching bei München (Germany). Before joining ESO, I was at the Space Telescope Science Institute in Baltimore (USA) and at the Scuola Normale Superiore in Pisa (Italy), from which I received my PhD in Physics in October 1998.
I share my (work) time between research and functional duties. My main line of research is currently on Cepheid stars as a high-precision tool to determine cosmic distances, and specifically how they can be used to measure the current rate of expansion of the Universe (Hubble constant) to the accuracy required to pinpoint the nature of Dark Energy and Cosmic Acceleration.
On the functional side, I am the Head of ESO’s Back-end Operations Department, which is charged with the handling of the science data from the La Silla Paranal Observatory, and, a few years down the line, of ESO’s next big science machine, the Extremely Large Telescope (ELT). As a department, we are responsible for the operations and development of the ESO Science Archive, both in terms of data content and user services. The Science Archive, with its open and FAIR access to the data, is a major booster to ESO’s scientific output: more than 35% of the refereed papers that make use of data from ESO’s Paranal Observatory contain archival data, with an upwards trend that has continued for several years. In addition, we lead the development of the ESO data processing tools and infrastructure, which provide the means to ESO science operations and the astronomical community to calibrate the data and extract the science signal.Profile Picture – adass-xxxi-2021/question_uploads/Martino02_SXmwTGD.png Affiliation –
ESO - European Southern ObservatoryPosition –
Head, Back-end Operations DepartmentHomepage –
Machine learning has been widely applied to clearly defined problems of astronomy and astrophysics. However, deep learning and its conceptual differences to classical machine learning have been largely overlooked in these fields. The broad hypothesis behind our work is that letting the abundant real astrophysical data speak for itself, with minimal supervision and no labels, can reveal interesting patterns which may facilitate discovery of novel physical relationships.
We train an encoder-decoder architecture on the self-supervised auxiliary task of reconstruction to allow it to learn general representations without bias towards any specific task. By exerting weak disentanglement at the information bottleneck of the network, we implicitly enforce interpretability in the learned features.
So far we have achieved interesting results in two avenues: firstly, our "AstroMachines" have learned to infer physical parameters such as radial velocity and effective temperature, just by watching a large number of stellar spectra and without being asked to do so. Secondly and more recently, we have observed semantic source separation abilities in the same architecture, and have reinforced it to "randomize out" telluric lines in stellar spectra, again in a non-supervised fashion.