ExSpliNet: An interpretable and expressive spline-based neural network

TL;DR

ExSpliNet is a novel neural network that combines spline functions, probabilistic trees, and Kolmogorov networks to achieve interpretability, expressiveness, and universal approximation capabilities, with demonstrated effectiveness on benchmarks.

Contribution

The paper introduces ExSpliNet, a new neural network architecture integrating splines and probabilistic trees, offering interpretability and strong approximation properties.

Findings

Effective on synthetic approximation tasks

Performs well on classical benchmarks

Offers a probabilistic interpretation of the model

Abstract

In this paper we present ExSpliNet, an interpretable and expressive neural network model. The model combines ideas of Kolmogorov neural networks, ensembles of probabilistic trees, and multivariate B-spline representations. We give a probabilistic interpretation of the model and show its universal approximation properties. We also discuss how it can be efficiently encoded by exploiting B-spline properties. Finally, we test the effectiveness of the proposed model on synthetic approximation problems and classical machine learning benchmark datasets.