Machine learning methods for graph-structured data

Leslie O'Bray (Ph.D. Student)

Alongside the rise of machine learning methods has been an increase in graph-structured data, which captures rich and complex relationships in the data. Since many machine learning methods do not work out-of-the box with graph-structured data, specific research and focus is necessary to develop such methods to enable this data type to also benefit from the performance gains that machine learning has made in recent years. While the applicability of such methods spans different domains (e.g. social networks, telecommunications, etc.), bioinformatics alone offers a diverse set of use cases for these methods. For example, it can be used to predict the properties of chemical molecules, classify the functional group of enzymes, and predict whether a drug will bind with a target protein. However, most methods have been developed across various domains, which consequently can result in learning paradigms that are ill-suited for bioinformatics. One of the main such frameworks uses aggregations of node neighborhoods in order to perform classification, which is not always appropriate for bioinformatics datasets. The primary goal of my research will be to develop non-neighborhood based methods for graph classification in order to improve the applicability of these methods to biological datasets.

Primary Host:	Karsten Borgwardt (Max Planck Institute of Biochemistry)
Exchange Host:	Michael Bronstein (University of Oxford)
PhD Duration:	01 October 2019 - 31 October 2023
Exchange Duration:	15 January 2022 - 14 July 2022 - Ongoing