Binary and Multinomial Classification through Evolutionary Symbolic Regression

TL;DR

This paper introduces three evolutionary symbolic regression algorithms for classification tasks, demonstrating their competitiveness against leading machine learning methods across numerous datasets and enabling automatic method selection via hyperparameter optimization.

Contribution

The paper presents novel evolutionary symbolic regression algorithms for classification and a method for automatic selection of the best algorithm for a given dataset.

Findings

Algorithms are competitive with XGBoost, LightGBM, and neural networks.

Demonstrated effectiveness across 162 datasets.

Automatic method selection improves classification performance.

Abstract

We present three evolutionary symbolic regression-based classification algorithms for binary and multinomial datasets: GPLearnClf, CartesianClf, and ClaSyCo. Tested over 162 datasets and compared to three state-of-the-art machine learning algorithms -- XGBoost, LightGBM, and a deep neural network -- we find our algorithms to be competitive. Further, we demonstrate how to find the best method for one's dataset automatically, through the use of a state-of-the-art hyperparameter optimizer.