An Evaluation of Training Strategies in QuGAN

Recently, researchers have attempted to enhance Generative Adversarial Network (GAN) for content generation using hybrid quantum-classical approaches. However, quantum GAN (QuGAN) is a novel field of study, and it remains unclear to understand the behaviors due to the presence of two competing modul...

Full description

Saved in:
Bibliographic Details
Published in2023 IEEE International Conference on Quantum Computing and Engineering (QCE) Vol. 2; pp. 336 - 337
Main Authors Ngo, Tuan A., Luu, Nhan T., Thang, Truong Cong
Format Conference Proceeding
LanguageEnglish
Published IEEE 17.09.2023
Subjects
Behavioral sciences
Content generation
Generative adversarial networks
Generators
Hybrid power systems
Hybrid quantum-classical system
Quantum computing
QuGAN
Training
Training strategies
Online AccessGet full text

Cover

More Information
Summary:Recently, researchers have attempted to enhance Generative Adversarial Network (GAN) for content generation using hybrid quantum-classical approaches. However, quantum GAN (QuGAN) is a novel field of study, and it remains unclear to understand the behaviors due to the presence of two competing modules, namely generator and discriminator. In this study, through empirical evaluation, we show that the choice of training strategies for generator and discriminator strongly affects the performance of QuGAN.
DOI:10.1109/QCE57702.2023.10270