Ïã¸ÛÁùºÏ²Ê

XClose

Ïã¸ÛÁùºÏ²Ê's Centre for Data Intensive Science

Home
Menu

Alicja Polanska

In an increasingly interdisciplinary world, developing the computer science, data analysis and artificial intelligence skills necessary for future large-scale scientific and industrial problems

Alicja Polanska

1 January 2022

Project title:ÌýMachine learning assisted model comparison with normalizing flows

Research Group: MSSL

Supervisor(s): Prof Jason Mcewen

±õ²Ô³Ù°ù´Ç»å³Ü³¦³Ù¾±´Ç²Ô:Ìý

Before coming to Ïã¸ÛÁùºÏ²Ê, I obtained an MSci Physics with Theoretical Physics degree from Imperial College London. My interest in cosmology and statistics was sparked by a summer project I carried out at the Imperial Astrophysics Group as part of the IROP programme. After graduation, I further developed my interest in data science and machine learning through a project in bioinformatics. I'm now working with Prof Jason McEwen at the Mullard Space Laboratory AstroInfo group. My PhD will focus on geometric deep learning and its applications to cosmology.Ìý


Project description:ÌýÌý

My research focuses on developing machine learning and statistical methods for astrophysics and beyond. Recently, I have been working on the learned harmonic mean estimator with normalizing flows: a robust, scalable and flexible estimator of the marginal likelihood for model comparison. The method is implemented in the open-source ‘harmonic’ Python package (), of which I am one of the main developers. Recently, we have applied our method in a practical, cosmological context and are currently working on further applications within astrophysics.

As part of my first-year group project, under the guidance of the data science team at the Guardian, my colleagues and I leveraged Natural Language Processing for coreference resolution to construct a custom pipeline for quote attribution in Guardian articles. Working on the project was an invaluable experience that allowed me to explore a new field in machine learning, as well as learn about how data science can be used in industry.

Before coming to Ïã¸ÛÁùºÏ²Ê, I obtained an MSci Physics with Theoretical Physics degree from Imperial College London. My interest in cosmology and data science was sparked by a summer project I carried out at the Imperial Astrophysics Group as part of the IROP programme.

First year group project:ÌýCoreference resolution for quote attribution at the Guardian

Placement:Ìý


Ìý

Publications:

RPS Widget Placeholder

Ìý
Ìý

Ìý