Analog Clipping Circuit Simulation with Recurrent Neural Networks
Loading...
Date
2021-03-29
Authors
ORCID
Advisor
Referee
Mark
Journal Title
Journal ISSN
Volume Title
Publisher
International Society for Science and Engineering, o.s.
Abstract
This article focuses on the practical use of recurrent neural networks for the simulation of analog audio circuits. Two virtual analog circuits were modeled using the Long Short-Term Memory neural networks. The neural network models presented in earlier literature were compared against newly proposed architectures, which used additional fully connected input layers. The signals processed by the neural network models of different complexity were compared to the ground truth data generated using the LTSpice software. It was found that the modifications done to the previously proposed neural network architectures can reduce the resulting prediction loss without significant increase in complexity.
This article focuses on the practical use of recurrent neural networks for the simulation of analog audio circuits. Two virtual analog circuits were modeled using the Long Short-Term Memory neural networks. The neural network models presented in earlier literature were compared against newly proposed architectures, which used additional fully connected input layers. The signals processed by the neural network models of different complexity were compared to the ground truth data generated using the LTSpice software. It was found that the modifications done to the previously proposed neural network architectures can reduce the resulting prediction loss without significant increase in complexity.
This article focuses on the practical use of recurrent neural networks for the simulation of analog audio circuits. Two virtual analog circuits were modeled using the Long Short-Term Memory neural networks. The neural network models presented in earlier literature were compared against newly proposed architectures, which used additional fully connected input layers. The signals processed by the neural network models of different complexity were compared to the ground truth data generated using the LTSpice software. It was found that the modifications done to the previously proposed neural network architectures can reduce the resulting prediction loss without significant increase in complexity.
Description
Keywords
Citation
Document type
Peer-reviewed
Document version
Published version
Date of access to the full text
Language of document
en
Study field
Comittee
Date of acceptance
Defence
Result of defence
Document licence
(C) 2021 Elektrorevue