Hybrid Approach in Quantum Generative Adversarial Network
Date
2023
Authors
Student ID
Research Supervisor
Editor
Journal Title
Volume
Issue
Journal Title
Journal ISSN
Volume Title
Abstract
This project explores the application of quantum computing techniques in the domain of generative adversarial networks (GANs) for image synthesis, specifically targeting the well-known MNIST dataset of handwritten digits. Leveraging TensorFlow Quantum (TFQ), this research delves into hybrid quantum-classical architectures, combining the power of quantum circuits with classical deep learning models. This project aims to demonstrate the potential of quantum GAN in generating synthetic images. This study's methodology includes data encoding, quantum circuit design, and classical optimization. The experimental results showcase the ability to generate fake samples. Compared with the classical counterpart, the project shows an improvement in generating samples over multiple epochs. However, after running 100 epochs, the hybrid version that uses both quantum circuits and classical neural networks shows a decreasing point in improvement compared to the classical counterpart that was able to improve over time without a decreasing point of improvement. While the current development of this project shows the capability of generating fake samples, further research is necessary to increase its performance and overcome the related issues. This tells us the potential usage of real-world applications that can be done with the capability of quantum computing on machine learning in generative tasks.
Description
Keywords
Citation
Department Name
Electrical and Computer Engineering
Publisher
North South University