A Computação Quântica e as Redes Adversárias Generativas

Abstract

Advances in quantum computing have opened up new possibilities for solving complex computational problems, with quantum machine learning (QML) standing out as an alternative to classical artificial intelligence methods. Quantum generative adversarial networks (QGANs), an adaptation of GANs for quantum devices, have evolved from synthetic tests to real-world applications. This work presents an alternative for generating 28x28 pixel images on quantum devices, using datasets such as MNIST and Fashion MNIST. A QGAN architecture capable of generating images with quality comparable to classical models, using fewer parameters, is presented. The obtained results demonstrate the potential of QGANs to revolutionize image generation and pave the way for new applications in artificial intelligence