Drawing Boundaries: Using individual evolved class boundaries for binary classification problems

This paper describes a technique which can be used with Genetic Programming (GP) to reduce implicit bias in binary classification tasks. Arbitrarily chosen class boundaries can introduce bias, but if individuals can choose their own boundaries, tailored to their function set, then their outputs are automatically scaled into a suitable range. These boundaries evolve over time as the individuals adapt to the data Our system calculates the Evolved Class Boundary (ECB) for each individual in every generation, with the twin aims of reducing training times and improving test fitness. The method is tested on three benchmark binary classification data sets from the medical domain. The results obtained suggest that the strategy can improve training, validation and test fitness, and can also result in smaller individuals as well as reduced training times. Our approach is compared with a standard benchmark GP system, as well as with over twenty other systems from the literature, many of which use highly tuned, non-EC methods, and is shown to yield superior results in many cases.