# Machine Learning in Astronomy

### Venue: A54, Postgraduate Statistics Centre, Lancaster University.

### Date: Thursday 30-05-2019, 1 - 6pm.

**Please note: **The event is free to attend, but attendees must register for a ticket in advance.

**David van Dyk (Imperial College)**

**Chris Arridge (Lancaster University)**

**Florent Leclercq (Imperial College)**

**Ingo Waldmann (University College London)**

The Royal Statistical Society section on Statistical Computing, and the Lancashire and Cumbria local group, are organising a one-day workshop on Machine Learning in Astronomy. This workshop is focused on the growing area of research in Astrostatistics and the application and Machine Learning techniques to answer scientific questions in Astronomy and Cosmology. In recent decades there has been an enormous increase in the volume and complexity of recorded astronomical data. To answer the many important scientific questions posed by the astronomical community, there is a need to develop efficient and objective scientific tools to exploit multifaceted astronomical data sets and to link these to astrophysical theory. Ongoing work in this area has already led to new statistical methods and machine learning techniques for classifying galaxies, discovering new pulsars and detecting of exoplanets.

This workshop is a half-day event which aims to bring together academics and students interested in the research challenges that lie at the interface between Astronomy and Data Analysis. There will be a poster session and wine reception sponsored by the RSS local group at the end of the workshop.

**Please note: **The event is free to attend, but attendees must register for a ticket in advance.

This event is jointly organised by the Royal Statistical Society Statistical Computing Section, and the Royal Statistical Society Lancashire and Cumbria local group.

### Schedule

13:00 - 13:15 - Opening and introductions

13:15 - 14:00 - **David van Dyk: **Data-Driven and Science-Driven Bayesian Methods in Astronomy and Solar Physics.

14:00 - 14:45 - **Chris Arridge: **Research Challenges in Solar System Plasmas.

14:45 - 15:15 - Coffee break

15:15 - 16:00 - **Florent Leclercq: **Bayesian inference with black-box cosmological models.

16:00 - 16:45 - **Ingo Waldmann:** Deep learning exoplanets and the solar system.

16:45 - 18:00 - Wine reception and poster session

### Abstracts

**Data-Driven and Science-Driven Bayesian Methods in Astronomy and Solar Physics**

**David A van Dyk, Imperial College Londo**n

In recent years, technological advances have dramatically increased the quality and quantity of data available to astronomers. Newly launched or soon-to-be launched space-based telescopes are tailored to data-collection challenges associated with specific scientific goals. These instruments provide massive new surveys resulting in new catalogs containing terabytes of data, high resolution spectrography and imaging across the electromagnetic spectrum, and incredibly detailed movies of dynamic and explosive processes in the solar atmosphere. The spectrum of new instruments is helping scientists make impressive strides in our understanding of the physical universe, but at the same time generating massive data-analytic and data-mining challenges for scientists who study the resulting data. In this talk I will illustrate and discuss the interplay of data science, machine learning, Bayesian statistics, data-driven methods, and science-driven methods in the context of several problems in astrophysics, ranging from studying the expansion history of the universe, to fitting models for stellar evolution, and mapping the physical characteristics of the solar corona.

**Research Challenges in Solar System Plasmas**

**Chris Arridge, Lancaster University**

The solar system is filled with a conducting supersonic fluid known as the solar wind. The Earth’s magnetic field deflects this plasma and carves a cavity out of the wind known as the magnetosphere. Other celestial bodies in our solar system have magnetospheres, including bodies that don’t possess large-scale ordered magnetic fields, such as Venus and comets. The governing physical laws that apply to “planetary magnetospheres” are Maxwell’s equations, along with conservation equations for energy, mass and momentum. Gravity is rarely important. The plasma, energetic particles and fields also couple with other elements of planetary systems, including planetary atmospheres, surfaces, and interiors, neutral gas clouds, ring systems, and electromagnetic radiation, across scales ranging from 10s to millions of km. These scales cross kinetic and fluid regimes in the system. In this lecture I will discuss a range of problems in the study of planetary magnetospheres, focusing on inference and inversion problems in these systems. In particular I will emphasise finite wave speed and causal decoupling, time-history effects, the constraints of the governing physical laws, and observer bias.

**Deep learning exoplanets and the solar system**

**Ingo Waldmann, University College London**

The field of exoplanetary spectroscopy is as fast moving as it is new. Analysing currently available observations of exoplanetary atmospheres often invoke large and correlated parameter spaces that can be difficult to map or constrain. This is true for both: the data analysis of observations as well as the theoretical modelling of their atmospheres. Issues of low signal-to-noise data and large, non-linear parameter spaces are nothing new and commonly found in many fields of engineering and the physical sciences. Recent years have seen vast improvements in statistical data analysis and machine learning that have revolutionised fields as diverse as telecommunication, pattern recognition, medical physics and cosmology. In many aspects, data mining and non-linearity challenges encountered in other data intensive fields are directly transferable to the field of extrasolar planets as well as planetary sciences.

In this talk, I will given an overview of the current state of machine learning in extrasolar planets and showcase two recent examples of solving long-standing problems with machine learning: 1) The modelling of exoplanetary atmospheres using a new deep learning framework, ExoGAN (Tzingales & Waldmann, 2018, AJ); and 2) The mapping of Saturn’s storm regions using our new hyper-spectral image classification code, PlanetNET (Waldmann & Griffith, 2019, Nature Astronomy). As we firmly move into the era of ‘big data’ for both planetary (e.g. Juno) and exoplanetary sciences (e.g. JWST, Ariel), intelligent algorithms will play an important part in facilitating the analysis of these rich data sets in the future.

**Bayesian inference with black-box cosmological models**

**Florent Leclercq, Imperial College London**

Large-scale astronomical surveys carry opportunities for testing physical theories about the origin and evolution of the Universe. Advancing the research frontier requires solving challenging and unique statistical problems, to unlock the information content of massive and complex data streams. In this talk, I will present recent methodological advances, aiming at fitting cosmological data with "black-box" numerical models. I will discuss two different solutions, depending on the scenario: Bayesian optimisation and Taylor-expansion of the simulator.