ExSpliNet: An interpretable and expressive spline-based neural network
This work addresses the need for more interpretable neural network models in machine learning, though it appears incremental by integrating existing concepts.
The authors tackled the challenge of creating an interpretable and expressive neural network by introducing ExSpliNet, which combines Kolmogorov networks, probabilistic trees, and B-splines, and demonstrated its effectiveness on synthetic and benchmark datasets.
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.