End-to-End QGAN-Based Image Synthesis via Neural Noise Encoding and Intensity Calibration

2026-03-19

AI Summary

This work proposes ReQGAN, a novel quantum generative adversarial network framework that enables end-to-end synthesis of full-resolution images using a single quantum circuit with D qubits to directly generate N = 2^D pixels. Addressing the limitations of existing QGANs—which rely on classical post-processing or patch-based strategies and thus fail to model global semantics—ReQGAN introduces a learnable neural noise encoder to overcome the rigidity of classical-quantum noise interfaces and incorporates a differentiable intensity calibration module to align quantum measurement outcomes with pixel intensity space. Experiments on MNIST and Fashion-MNIST demonstrate stable training and high-quality image generation under strict quantum resource constraints, validating the effectiveness of the proposed components.

Technology Category

• Machine Learning: Quantum Machine Learning
• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Quantum Generative Adversarial Networks (QGANs) offer a promising path for learning data distributions on near-term quantum devices. However, existing QGANs for image synthesis avoid direct full-image generation, relying on classical post-processing or patch-based methods. These approaches dilute the quantum generator's role and struggle to capture global image semantics. To address this, we propose ReQGAN, an end-to-end framework that synthesizes an entire N=2^D-pixel image using a single D-qubit quantum circuit. ReQGAN overcomes two fundamental bottlenecks hindering direct pixel generation: (1) the rigid classical-to-quantum noise interface and (2) the output mismatch between normalized quantum statistics and the desired pixel-intensity space. We introduce a learnable Neural Noise Encoder for adaptive state preparation and a differentiable Intensity Calibration module to map measurements to a stable, visually meaningful pixel domain. Experiments on MNIST and Fashion-MNIST demonstrate that ReQGAN achieves stable training and effective image synthesis under stringent qubit budgets, with ablation studies verifying the contribution of each component.

Problem

• Quantum Generative Adversarial Networks
• Image Synthesis
• End-to-End Generation
• Quantum Circuit
• Pixel Intensity

Innovation

• Quantum GAN
• End-to-End Image Synthesis
• Neural Noise Encoding
• Intensity Calibration
• Quantum Circuit