skip to content

Department of Computer Science and Technology

A PhD student here investigating ways to tackle bias in machine learning systems has won the Young AI Researcher 2022 Award at the Perspektywy Women in Tech Summit in Warsaw, Poland.

Agnieszka Slowik, a final-year PhD student here, was selected to receive the award from over 300 nominees working on all aspects of artificial intelligence, including research, software engineering, entrepreneurship and policy making. She won for work accomplished during her studies.

Agnieszka's research during her PhD has been focused on out-of-distribution generalisation and robustness in Artificial Intelligence, under the supervision of Sean Holden and Mateja Jamnik from the Machine Learning and Artificial Intelligence Group

Agnieszka is interested in the performance of AI algorithms on data distributions that are different from those the algorithms were trained on. "This is one of the biggest challenges in modern Machine Learning," she says. She has presented her work at the leading Machine Learning conferences NeurIPS 2020, AAAI 2020 & 2022, and AISTATS 2021 & 2022.

In her latest paper, Agnieszka investigates the topic of generalisation to ensure fairness of AI models, especially on under-represented or challenging populations. The question of "whether fighting bias in machine learning systems is a data curation problem or an algorithmic problem has been the object of much discussion," she notes in the abstract of On the Relation between Distributionally Robust Optimization and Data Curation.

Her research found that the two approaches to addressing this problem – one algorithmic and one that involved reweighting the data samples in a training dataset – that they had an equivalent effect and therefore that either of them could be used.

And finding new ways to address bias in machine learning matters, she says, in order to make sure that machine learning systems learn from minority, as well as majority, populations within the data sets that they are trained on.

This is important in order to develop 'fair' machine learning tools that might be employed in online recruitment platforms – "for example, for technical jobs where female applicants are a minority and male applicants in the majority" – or in measuring reoffending rates across prisoners from a variety of race or demographic backgrounds.

"It's really interesting work and when I've finished my PhD, I'd like to continue research in the area of inclusive, robust AI," Aga says.

"I chose to pursue this research because, while I have also conducted fundamental research, I'm interested in work that is impactful and has to do with real life."

Agnieszka has also been involved with an important sub-population here in the Department. She has been actively involved in the women@CL initiative here, which supports women in computing research in their careers and encourages them to aspire to leadership positions, both in academia and industry. 

Agnieszka was first a Co-Chair for the Tech Events organised by women@CL, and later went on to be a Co-Chair of the Oxbridge Women in Computer Science Conference 2022.

Agnieszka is currently writing her PhD dissertation and looking for a job as a Machine Learning/Artificial Intelligence researcher. We wish her the best of luck!

Published by Rachel Gardner on Monday 18th July 2022