Publications
Thesis
This thesis develops prototype-based pattern classifiers whose prototypes are optimized using differential evolution, a global, population-based optimization method. This work is the origin of my background in computational intelligence, pattern recognition, and optimization methods that now inform my interest in computational and agent-based approaches to economics and public policy.
Peer-Reviewed Publications
This paper, based on my master’s thesis research, combines ideas from Learning Vector Quantization (LVQ) with a differential-evolution-based automatic clustering approach to design prototype-based classifiers — determining both the optimal number of prototypes per class and their positions in the data space.
Author name as published; my legal name is Luiz Soares de Andrade Filho (see the thesis above, published under that name).