PhD Studies (thesis 1.4MB)
Biliana Alexandrova-Kabadjova joined the
Group at University of Essex
in 2002 as a PhD student.
She is registered under the
Centre For Computational Finance and Economic Agents.
She was supervised jointly by
Andreas Krause (School of Management, University of Bath) and
Edward Tsang (Department of Computer Science, University of Essex).
Benefited from joint supervision, Biliana was able to tackle a very complex
computational finance project which demands non-trivial knowledge in both economics and computation.
Biliana passed her viva on 27 July 2007. She was examined by
Professor Peter Allen
(External Examiner, Head of Complex Systems Management Centre, Cranfield University, world authority in modelling)
and Maria Fasli (Internal Examiner, author of
Agent Technology for E-Commerce -- a thorough and rigorous text in this field).
Research and Contributions
Building on existing work, Biliana has built the most advanced model
(in terms of complexity and realism) of credit card payment in the world.
She has developed a framework under which one can learn regulatory strategies and
individual agent can learn business strategies.
Learning is based on Population based Incremental Learning (PBIL),
a learning mechanism of which Qingfu Zhang
at University of Essex is a world expert.
With this model, Biliana was able to predict market shares and demands.
This is useful for wind-tunnel testing Government policies.
Banks can also use her model to learn pricing strategies with the
view to maximize their market share or profit.
Given clear objectives, Governments can use
Biliana's framework to evolve policy decisions
(evolved policies are hopefully more robust than designed policies).
Biliana is now working for the Mexican Central Bank.
Maintained by Edward Tsang;
updated 4 November 2007