Graduate Seminar

Location:  CL 312

Speaker: Pornpiya Vikitset

MSc Student supervised by Andrei Volodin

Title: Similarity-based clustering of probability density functions

Abstract:

This presentation will focus on a method for clustering probability density functions (PDFs). This innovative technique utilizes a robust similarity measure to group PDFs, effectively addressing challenges such as varying cluster counts, unbalanced cluster sizes, and the presence of noisy data. Through numerical studies, we will demonstrate the efficacy of this algorithm in accurately clustering PDFs, even without a prior knowledge of the number of clusters. Furthermore, we will explore its potential applications in finance, specifically for grouping customers with similar portfolios or risk behaviors to facilitate targeted loan product offerings.