Introduction

Abstract

Suppose one is confronted with a medical classification problem. What trustworthy technique should one then use to solve it? Support vector machines (SVMs) are known to be a smart choice. But how can one make a personal, more flexible implementation of the learning engine that makes them run that well? And how does one open the black box behind their predicted diagnosis and explain the reasoning to the otherwise reluctant fellow physicians? Alternatively, one could choose to develop a more versatile evolutionary algorithm(EA) to tackle the classification task towards a potentially more understandable logic of discrimination.