VPNet: Variable Projection Networks

TL;DR

VPNet is a new neural network architecture based on variable projection that offers interpretable features, compact structure, and efficient learning, demonstrating promising results in signal classification tasks with low computational costs.

Contribution

The paper introduces VPNet, a novel model-driven neural network architecture leveraging variable projection for improved interpretability and efficiency in signal processing applications.

Findings

VPNet achieves fast learning and high accuracy.

Compared to traditional networks, VPNet has lower computational costs.

VPNet shows promising results in ECG signal classification.

Abstract

We introduce VPNet, a novel model-driven neural network architecture based on variable projection (VP). Applying VP operators to neural networks results in learnable features, interpretable parameters, and compact network structures. This paper discusses the motivation and mathematical background of VPNet and presents experiments. The VPNet approach was evaluated in the context of signal processing, where we classified a synthetic dataset and real electrocardiogram (ECG) signals. Compared to fully connected and one-dimensional convolutional networks, VPNet offers fast learning ability and good accuracy at a low computational cost of both training and inference. Based on these advantages and the promising results obtained, we anticipate a profound impact on the broader field of signal processing, in particular on classification, regression and clustering problems.