EU-Nets: Enhanced, Explainable and Parsimonious U-Nets

TL;DR

EU-Nets are novel U-Net variants that improve explainability, uncertainty estimation, and performance while maintaining a parsimonious model size, applicable across various architectures.

Contribution

Introduction of EU-Nets with the Equivalent Convolutional Kernel and collaboration gradient for enhanced interpretability and uncertainty estimation.

Findings

Average accuracy improved by 1.389%

Variance reduced by 0.83%

Fewer than 0.1M parameters required

Abstract

In this study, we propose MHEX+, a framework adaptable to any U-Net architecture. Built upon MHEX+, we introduce novel U-Net variants, EU-Nets, which enhance explainability and uncertainty estimation, addressing the limitations of traditional U-Net models while improving performance and stability. A key innovation is the Equivalent Convolutional Kernel, which unifies consecutive convolutional layers, boosting interpretability. For uncertainty estimation, we propose the collaboration gradient approach, measuring gradient consistency across decoder layers. Notably, EU-Nets achieve an average accuracy improvement of 1.389\% and a variance reduction of 0.83\% across all networks and datasets in our experiments, requiring fewer than 0.1M parameters.