Learning Graphs from Data and Spectral Clustering with Applications to Finance
10am
Rm 4472, 4/F (via Lift 25/26), Academic Building, HKUST

HKUST Thesis Examination Committee

Prof. Daniel PALOMAR, ECE/HKUST (Thesis Supervisor)

Prof. Weichuan YU, ECE/HKUST

Dr. Kumar SANDEEP, ECE/HKUST
 

 

Abstract

In this bachelor's thesis we give a survey of methods to learn meaningful graphs from time series. Graph representations of data sets give us many possibilities to enhance the processing, analysis, and visualization of data. Specifically, they allow the use of new approaches to clustering data. Related to this, we have studied spectral clustering algorithms using different Laplacian matrices.

As a practical application, these techniques are used to find clusters in the stock market with a focus on comparing the clusters structure with the industry classification of companies. Finally, we have also tried to visualize the increase of systemic risk associated with economical crisis.

When
Time
10am
Where
Rm 4472, 4/F (via Lift 25/26), Academic Building, HKUST
Event Format
Language
English
Organizer
Department of Electronic & Computer Engineering
Post an event
Campus organizations are invited to add their events to the calendar.