Network Comparison via Subgraph counts

Speaker: Dr Luis Ospina-Forero (The University of Manchester)

Venue: MANDEC Lecture Theatre, 3rd Floor, University Dental Hospital, Higher Cambridge Street, Manchester, M15 6FH

Abstract: There are many techniques for network comparison, from simply comparing network summary statistics to sophisticated but computationally costly alignment-based approaches. Yet it remains challenging to accurately cluster networks that are of a different size and density, but hypothesized to be structurally similar or which share similar core properties. In this talk I will give a brief description of several network comparison methods based on subgraph counts. I will show that although most network comparison methods based on subgraph counts aim to capture the same type of structural similarity, some methods focus more on fine grain similarities and other methods capture core similarities better despite having networks with different numbers of nodes and edges. 
I will end this talk by showing some ongoing case studies were we have been exploring the usefulness of these network comparison methods. Mainly their use in the forecast of stock market prices by means of correlation networks; their use in hypothesis testing for the assessment of fit of random graph models to protein-protein interaction networks; and finally to test the impact of road network structures to house prices.

Bio: Luis Ospina-Forero is a researcher in Data Science in the Alliance Manchester Business school. Prior to his position at Manchester University, he was a Data Scientist at The Alan Turing Institute (London). His research interests relate to Network Inference, Network Comparison and the application of Network Analysis methodologies to practical and meaningful problems in finance and society. Luis holds a PhD degree in Statistics from the University of Oxford and is a statistician by training from the National University of Colombia.