Steinmetz Neural Networks for Complex-Valued Data

TL;DR

This paper introduces Steinmetz Neural Networks, a novel architecture for complex-valued data processing that enhances interpretability and robustness through multi-view learning and analytic signal representations.

Contribution

It proposes a new neural network architecture for complex data, incorporating a consistency penalty for analytic signals and demonstrating improved generalization and noise robustness.

Findings

Lower generalization gap upper bound compared to traditional Steinmetz networks

Enhanced performance on benchmark datasets

Increased robustness to additive noise

Abstract

We introduce a new approach to processing complex-valued data using DNNs consisting of parallel real-valued subnetworks with coupled outputs. Our proposed class of architectures, referred to as Steinmetz Neural Networks, incorporates multi-view learning to construct more interpretable representations in the latent space. Moreover, we present the Analytic Neural Network, which incorporates a consistency penalty that encourages analytic signal representations in the latent space of the Steinmetz neural network. This penalty enforces a deterministic and orthogonal relationship between the real and imaginary components. Using an information-theoretic construction, we demonstrate that the generalization gap upper bound posited by the analytic neural network is lower than that of the general class of Steinmetz neural networks. Our numerical experiments depict the improved performance and robustness to additive noise, afforded by our proposed networks on benchmark datasets and synthetic examples.