Quantum generative models offer a promising new direction in machine learning by leveraging quantum circuits to enhance data generation capabilities. In this study, we propose a hybrid quantum-classical image generation framework that integrates variational quantum circuits into a diffusion-based model. To improve training dynamics and generation quality, we introduce two novel noise strategies: intrinsic quantum-generated noise and a tailored noise scheduling mechanism. Our method is built upon a lightweight U-Net architecture, with the quantum layer embedded in the bottleneck module to isolate its effect. We evaluate our model on MNIST and MedMNIST datasets to examine its feasibility and performance. Notably, our results reveal that under limited data conditions (fewer than 100 training images), the quantum-enhanced model generates images with higher perceptual quality and distributional similarity than its classical counterpart using the same architecture. While the quantum model shows advantages on grayscale data such as MNIST, its performance is more nuanced on complex, color-rich datasets like PathMNIST. These findings highlight both the potential and current limitations of quantum generative models and lay the groundwork for future developments in low-resource and biomedical image generation.
Quantum generative models offer a promising new direction in machine learning by leveraging quantum circuits to enhance data generation capabilities. In this study, we propose a hybrid quantum-classical image generation framework that integrates variational quantum circuits into a diffusion-based model. To improve training dynamics and generation quality, we introduce two novel noise strategies: intrinsic quantum-generated noise and a tailored noise scheduling mechanism. Our method is built upon a lightweight U-Net architecture, with the quantum layer embedded in the bottleneck module to isolate its effect. We evaluate our model on MNIST and MedMNIST datasets to examine its feasibility and performance. Notably, our results reveal that under limited data conditions (fewer than 100 training images), the quantum-enhanced model generates images with higher perceptual quality and distributional similarity than its classical counterpart using the same architecture. While the quantum model shows advantages on grayscale data such as MNIST, its performance is more nuanced on complex, color-rich datasets like PathMNIST. These findings highlight both the potential and current limitations of quantum generative models and lay the groundwork for future developments in low-resource and biomedical image generation.
Quantum generative modeling is a rapidly evolving discipline at the intersection of quantum computing and machine learning. Contemporary quantum machine learning is generally limited to toy examples or heavily restricted datasets with few elements. This is not only due to the current limitations of available quantum hardware but also due to the absence of inductive biases arising from application-agnostic designs. Current quantum solutions must resort to tricks to scale down high-resolution images, such as relying heavily on dimensionality reduction or utilizing multiple quantum models for low-resolution image patches. Building on recent developments in classical image loading to quantum computers, we circumvent these limitations and train quantum Wasserstein GANs on the established classical MNIST and Fashion-MNIST datasets. Using the complete datasets, our system generates full-resolution images across all ten classes and establishes a new state-of-the-art performance with a single end-to-end quantum generator without tricks. As a proof-of-principle, we also demonstrate that our approach can be extended to color images, exemplified on the Street View House Numbers dataset. We analyze how the choice of variational circuit architecture introduces inductive biases, which crucially unlock this performance. Furthermore, enhanced noise input techniques enable highly diverse image generation while maintaining quality. Finally, we show promising results even under quantum shot noise conditions.
Quantum Generative Adversarial Networks (QGANs) combine quantum circuit based generative models with classical discriminators to exploit quantum advantages in generative modeling. This work investigates a hybrid QGAN architecture with a quantum-enhanced generator and a classical discriminator, applied to the Modified National Institute of Standards and Technology (MNIST) dataset. We implement and benchmark three different quantum circuit ansätze - namely a basic entangler circuit, a simple two-design circuit and a strongly entangling circuit, with each QGAN model being constructed using six layers of the respective ansätze. To manage the 28x28 image complexity, the generator employs a patch wise scheme, dividing each image into small patches generated by separate sub-circuits. The performance of each ansatz is evaluated in terms of training stability and output image quality. Our results demonstrate that the choice of quantum circuit ansatz significantly impacts the QGAN’s ability to learn the distribution of handwritten digits. In particular, the strongly entangling ansatz achieves the best performance at the cost of stability and image fidelity among the three yielding recognizable digit images. While no quantum advantage was demonstrated, this study provides comparative analysis and insights into designing effective quantum generators for hybrid QGANs on image data, serving as a benchmark for future developments in quantum-enhanced generative models.
Quantum Generative Adversarial Networks (QGANs), as a rising paradigm in Quantum Machine Learning, have shown promising potential in image generation and processing. However, their output quality remains suboptimal, and existing research is largely limited to small-scale, proof-of-concept studies. In this work, we propose a hybrid quantum-classical GAN architecture, where the generator integrates parameterized quantum circuits (PQCs) and classical neural networks. This integration significantly enhances the visual quality of generated images. Our model leverages the complementary strengths of quantum and classical components and outperforms existing methods (Tsang et al. in IEEE Trans. Quantum Eng. 4:1–19, 2023 ; Gulrajani et al. in Proceedings of the 31st international conference on neural information processing systems. NIPS’17, Red Hook, pp. 5769–5779, 2017 ), particularly in terms of image fidelity. Experiments conducted on the MNIST family of datasets show that our hybrid approach achieves a 20.26% average reduction in Fréchet Inception Distance. Furthermore, it improves the Structural Similarity Index Measure, Cosine Similarity, and Peak Signal-to-Noise Ratio by 26.04%, 2.22%, and 7.62%, respectively. These results highlight the effectiveness of combining quantum computing with machine learning, and underscore the potential of hybrid quantum-classical models in advancing generative tasks.
Purpose This study investigates the integration of quantum-enhanced machine learning (QML) into medical image analysis, focusing on classification and feature extraction tasks. It aims to evaluate whether hybrid quantum-classical models can overcome the limitations of traditional machine learning in handling high-dimensional, clinically relevant imaging data. Design/methodology/approach Hybrid architectures combining classical preprocessing with quantum components, specifically variational quantum circuits (VQCs) and quantum kernel methods, were developed and implemented in simulated quantum environments using Qiskit and PennyLane on Google Colab. Publicly available datasets (ChestX-ray14, Brain Tumor MRI and CT Hemorrhage) were preprocessed and encoded into quantum states. Model performance was evaluated against classical baselines (MLP and CNN) using accuracy, precision, recall, F1-score, convergence speed and trainable parameters. Findings Quantum-enhanced models achieved competitive or superior performance compared to classical models with significantly fewer trainable parameters and faster convergence. The Quantum Kernel SVM demonstrated the highest precision (89.1%) and F1-score (88.5%) using only 8 parameters, while VQC models converged up to 2× faster than classical baselines. These results highlight the potential of QML to improve efficiency, robustness and scalability in diagnostic imaging workflows, particularly in resource-constrained or data-limited settings. Research limitations/implications The study relies on simulated quantum environments, which do not capture hardware noise or qubit limitations. Future work will extend to real quantum devices, multimodal data integration and interpretability strategies. Practical implications Quantum kernel methods and lightweight VQC architectures enable faster, resource-efficient diagnostic modeling, making them suitable for clinical environments with limited computational capacity or time-sensitive decision-making needs. Social implications Improving the efficiency and accessibility of medical image analysis has the potential to support faster diagnosis and enhance healthcare delivery, particularly in underserved or resource-limited regions. Originality/value This study provides a practical evaluation of hybrid quantum-classical models applied to multiple real medical imaging datasets in a reproducible simulated environment. It offers an early-stage framework demonstrating how QML techniques can be integrated into existing machine learning pipelines for medical image analysis.
Quantum machine learning (QML) has received increasing attention due to its potential to outperform classical machine learning methods in problems such as classification and identification tasks. A subclass of QML methods is quantum generative adversarial networks (QGANs) which have been studied as a quantum counterpart of classical GANs widely used in image manipulation and generation tasks. The existing work on QGANs is still limited to small-scale proof-of-concept examples based on images with significant down-scaling. Here we integrate classical and quantum techniques to propose a new hybrid quantum-classical GAN framework. We demonstrate its superior learning capabilities over existing quantum techniques by generating 28×28 pixels grey-scale images without dimensionality reduction or classical pre/post-processing on multiple classes of the standard MNIST and Fashion MNIST datasets, which achieves comparable results to classical frameworks with three orders of magnitude less trainable generator parameters. To gain further insight into the working of our hybrid approach, we systematically explore the impact of its parameter space by varying the number of qubits, the size of image patches, the number of layers in the generator, the shape of the patches and the choice of prior distribution. Our results show that increasing the quantum generator size generally improves the learning capability of the network. The developed framework provides a foundation for future design of QGANs with optimal parameter set tailored for complex image generation tasks
Generative AI is one of the modern humankind's most transformative fields, paving a path for machines to generate realistic information in many forms (e.g., images/text/audio/video/scientific data). Diffusion models are the system that recently showed how to learn to generate them step by step by reversing stochastic noise processes through normalisation layers. Such models outperformed previous techniques including Generative Adversarial Networks (GANs) in terms of stability, diversity and controllability. While, despite those successes, classical diffusion models have major limitations such as expensive computations, slow sampling times or great training pipeline requirements and some challenges when modelling ultra-high-dimensional probability distributions. Emerging computational paradigms to accelerate and enhance generative intelligence systems have come when data complexity is becoming more challenging in domains like healthcare, finance, cybersecurity, climate science, industrial automation. One of the main solutions being offered by quantum computing, which takes advantage of quantum mechanical principles exploits superposition, entanglement and quantum parallelism to process information in a fundamentally new way. QEDMs, or Quantum-Enhanced Diffusion Models, specify a hybrid architecture that integrates classical diffusion architectures with quantum circuits, quantum kernels or variation quantum neural networks to improve learning efficiency and quality of the generated samples. Quantum processors can be included in these models as part of their noise estimates, latent representation learning, optimization routines or probabilistic sampling stages. This integration can lead to less-complex models, faster convergence, better feature extraction and more informative multimodal outputs. In addition, one could potential zing QEDMs for simulating molecular data, material discovery, optimizing synthetic dataset and privacy conserving data generation via quantum randomness. In this paper, we review the theory, architecture, implementation schemes, applications and future scope of Quantum-Enhanced Diffusion Models for Generative Intelligence and Data Synthesis. It contrasts demonising score matching, latent-space transitions, and stochastic reverse diffusion processes with quantum resources. This comparative analysis illustrates the anticipated enhancements in computational scalability, accuracy, and energy efficiency in contrast to existing methods. It also tackles real-world issues, including finite quit counts, DE coherence and noise in near-term quantum hardware, integration overhead and algorithmic complexity. Last but not least the paper suggests future research directions with respect to fault-tolerant quantum systems, quantum cloud delivery frameworks, federated generative intelligence and autonomous creative systems. Combining the strengths of probabilistic deep learning, but leveraging the transformative power of quantum computation, quantum-enhanced generative models could become a foundational technology for the next generation of artificial intelligence.