14:00 – 15:00 – Valeria Vitelli (University of Oslo)
Title: Bayesian rank-based clustering of high-dimensional data in the Mallows rank model
Abstract:
Rank-based models can be used to estimate individual behaviours and preferences in several areas, such as marketing and politics. The Mallows model is a popular model for rankings, as it flexibly adapts to different types of preference data, and the previously proposed Bayesian Mallows Model (BMM) offers a computationally efficient framework for Bayesian inference also allowing capturing the users’ heterogeneity, via a finite mixture. However, the Mallows model does not scale, and also seems unrealistic, when the pool of items is large. Furthermore, BMM does not currently allow the use of additional user-related information (covariates), which can potentially lead to a better accuracy in individual predictions, by enhancing the understanding of the users’ personal profiles.
In this talk, I will introduce recent extensions of BMM in a number of directions: (i) embedding covariate information related to assessors, via a similarity function that a priori favours the aggregation of people into a cluster when their covariates are similar. (ii) combining a lower-dimensional version of BMM (lowBMM), that scales to large datasets, with a Bayesian mixture of Mallows models, and showcase its performance in the context of cancer genomics.
(iii) Finally, proposing a variational inferential setting to help with scalability.
This is joint work with Luca Coraggio, Emilie Eliseussen, Arnoldo Frigessi, Haakon Muggerud, and Ida Scheel.
Refreshments available between 15:00 – 15:30, Huxley Common Room (HXLY 549)