LSTM-QGAN: Scalable NISQ Generative Adversarial Network

TL;DR

LSTM-QGAN introduces a scalable quantum generative adversarial network architecture that eliminates PCA preprocessing and leverages QLSTM, significantly improving performance, scalability, and resource efficiency for practical quantum data generation.

Contribution

It proposes a novel LSTM-QGAN architecture that removes PCA and incorporates QLSTM, addressing scalability and effectiveness issues in existing QGANs.

Findings

Enhanced visual data quality and lower Frechet Inception Distance scores.

Achieved 5x reductions in qubit counts and single-qubit gates.

Achieved 12x reduction in two-qubit gates.

Abstract

Current quantum generative adversarial networks (QGANs) still struggle with practical-sized data. First, many QGANs use principal component analysis (PCA) for dimension reduction, which, as our studies reveal, can diminish the QGAN's effectiveness. Second, methods that segment inputs into smaller patches processed by multiple generators face scalability issues. In this work, we propose LSTM-QGAN, a QGAN architecture that eliminates PCA preprocessing and integrates quantum long short-term memory (QLSTM) to ensure scalable performance. Our experiments show that LSTM-QGAN significantly enhances both performance and scalability over state-of-the-art QGAN models, with visual data improvements, reduced Frechet Inception Distance scores, and reductions of 5x in qubit counts, 5x in single-qubit gates, and 12x in two-qubit gates.