Journal article
Evaluating the generalization of complex-weight neural networks over simulated Lamb wave responses from hollow spheres
The Journal of the Acoustical Society of America, Vol.157(4), pp.2542-2555
04/01/2025
DOI: 10.1121/10.0036384
PMID: 40192330
Abstract
The advancement of complex-valued machine learning brings about new potential for the neglected phase information in acoustics research. A comparison between models that either ignore phase or represent complex numbers as pairs of reals and fully complex numbers yields results that indicate that complex-valued networks are as good as or better than the best fully real option, while having roughly half of the computer memory cost. This is performed using simulated partial wave responses (Lamb waves) for hollow spheres suspended in an infinite homogenous ocean. These spheres of varying thickness are classified based on material with fully connected networks applied to a frequency passband. For the hardest to classify subset of the generated data, there was comparable classification confidence and accuracy observed between the best-performing real network and the complex network. Perfect classification accuracy for unseen partial wave response data was achieved in some trained models, which suggests a disparity in minima in the gradient space and promotes further study into noise augmentation and convolution with simulated multipath channels.
Details
- Title: Subtitle
- Evaluating the generalization of complex-weight neural networks over simulated Lamb wave responses from hollow spheres
- Creators
- Timothy J Linhardt - University of IowaAnanya Sen Gupta - University of IowaIvars Kirsteins - Naval Undersea Warfare Center
- Resource Type
- Journal article
- Publication Details
- The Journal of the Acoustical Society of America, Vol.157(4), pp.2542-2555
- Publisher
- ACOUSTICAL SOC AMER AMER INST PHYSICS
- DOI
- 10.1121/10.0036384
- PMID
- 40192330
- ISSN
- 1520-8524
- eISSN
- 1520-8524
- Grant note
- U.S. Navy10.13039/100009896: N00174-20-1-0016 Naval Sea Systems Command, U.S. Navy
This research was supported by the Naval Sea Systems Command, U.S. Navy, under Grant No. N00174-20-1-0016.
- Language
- English
- Date published
- 04/01/2025
- Academic Unit
- Electrical and Computer Engineering
- Record Identifier
- 9984808527002771
Metrics
2 Record Views