A Benchmarking Study of Kolmogorov-Arnold Networks on Tabular Data

TL;DR

This paper benchmarks Kolmogorov-Arnold Networks against MLPs on real-world tabular data, showing KANs often outperform in accuracy but at higher computational costs.

Contribution

It provides the first comprehensive comparison of KANs and MLPs on real-world tabular datasets, highlighting their relative strengths and computational trade-offs.

Findings

KANs achieve superior or comparable accuracy and F1 scores.

KANs perform particularly well on datasets with many instances.

KANs require more computational resources than MLPs.

Abstract

Kolmogorov-Arnold Networks (KANs) have very recently been introduced into the world of machine learning, quickly capturing the attention of the entire community. However, KANs have mostly been tested for approximating complex functions or processing synthetic data, while a test on real-world tabular datasets is currently lacking. In this paper, we present a benchmarking study comparing KANs and Multi-Layer Perceptrons (MLPs) on tabular datasets. The study evaluates task performance and training times. From the results obtained on the various datasets, KANs demonstrate superior or comparable accuracy and F1 scores, excelling particularly in datasets with numerous instances, suggesting robust handling of complex data. We also highlight that this performance improvement of KANs comes with a higher computational cost when compared to MLPs of comparable sizes.