Document Type
Article
Publication Date
8-10-2026
Abstract
We apply machine learning methods to demonstrate radar range superresolution using a denoising autoencoder trained without supervision. Focusing on the estimation of a single physical parameter, the separation between two scatterers in the subwavelength regime, we constrain the network to a one-dimensional bottleneck layer with its size matched to the parameter dimensionality. We find that the bottleneck layer forms a reproducible, monotonic mapping with the true separation, showing that the network learns a low-dimensional representation directly aligned with the underlying physical parameter. We further show that this representation preserves the Fisher information of the signal, indicating that the network recovers a physically meaningful representation for parameter estimation. We investigate the behavior of the bottleneck layer for the following types of pulses: a traditional sinc pulse, a bandlimited triangle-type pulse, and a theoretically near-optimal pulse created from a spherical Bessel function basis. Our results demonstrate that neural networks can discover physically meaningful compressed representations without explicit supervision, with performance ultimately governed by the information content of the signal.
Recommended Citation
R. Czupryniak, A. Chakraborty, A. N. Jordan, and J. C. Howell, Demonstrating superresolution in radar range estimation using a denoising autoencoder, Phys. Rev. Res. 8, 033157 (2026). https://doi.org/10.1103/w6m6-n6rk
Peer Reviewed
1
Copyright
The authors
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Comments
This article was originally published in Physical Review Research, volume 8, in 2026. https://doi.org/10.1103/w6m6-n6rk