Quantum Generative Diffusion Model for Real-World Time Series

Jack Waller, Filippo Caruso, Dimitrios Makris, Rajagopal Nilavalan, Xing Liang · 2026

Read Quantum Generative Diffusion Model for Real-World Time Series by Jack Waller, Filippo Caruso, Dimitrios Makris, Rajagopal Nilavalan, Xing Liang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-27561.txt.

Introduction

Waller leads a team building the first quantum generative diffusion model aimed at time series, and running it on real quantum hardware. Their design swaps the feed-forward parts of the denoising transformer for quantum neural networks, which yields a hybrid quantum transformer. Each replaced component carries roughly a thousand times fewer trainable parameters. In the 2026 quantum machine learning paper, financial series from two technology giants supply the test data, and the model trains and samples on a commercial quantum processor.

The synthetic data reproduce the real distributions more faithfully than the classical counterpart, cutting the Wasserstein distance by about forty-four percent across both datasets in the reported comparison. Augmenting a forecasting model with the generated data improves its error by up to seventy-one percent over training on real data alone in the same study. Quantum components match or beat classical ones with far fewer parameters.

Catalog record

CREATORJack Waller, Filippo Caruso, Dimitrios Makris, Rajagopal Nilavalan, Xing Liang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.27561

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Quantum Generative Diffusion Model for Real-World Time Series Jack Waller, Filippo Caruso, Dimitrios Makris, Rajagopal Nilavalan, Xing Liang Abstract Generative models have achieved remarkable success in data synthesis, though recent advances driven by increasing model scale have introduced challenges in computational cost and efficiency. Quantum machine learning offers a promising alternative, representing complex data distributions using compact, highly expressive models. Here, we propose QDiffusion-TS, the first quantum generative diffusion model for time series synthesis, and validate it on the IQM quantum processor. The framework extends a classical diffusion architecture by replacing feed-forward components within the denoising transformer with quantum neural networks, yielding a hybrid quantum transformer that reduces the number of trainable parameters in each replaced component by nearly three orders of magnitude. Evaluated on financial time series from Apple and Amazon, the model generates synthetic data that more accurately reproduces the real distributions, reducing Wasserstein distance by approximately 44% relative to its classical counterpart across both datasets. In a downstream forecasting task, augmentation with the generated data improves predictive performance by up to 71% in RMSE over a baseline trained solely on real data. These results show that quantum enhanced architectures can consistently match and frequently surpass classical performance with substantially fewer parameters, establishing a practical framework towards more efficient and scalable data-driven generative modelling. keywords quantum machine learning, diffusion models, time series generation, quantum neural networks, financial data, generative modelling 1 Introduction Generative machine learning models have experienced rapid advancement in recent years, achieving remarkable empirical performance across a wide range of data synthesis tasks 31Sengar ; 32Ooi ; 33Fahrner . Among these, image generation has progressed especially quickly, with variational autoencoders (VAEs) 34Huang ; 35Daniel , generative adversarial networks (GANs) 63Goodfellow ; 36Gulrajani ; 37Karras , and diffusion models 61Ho ; 62Dickstein ; 41Rombach ; 38Dhariwal ; 39Kingma ; 80Lu now capable of producing samples that are highly realistic. Whilst progress has been considerable, these gains have been driven in large part by the increasing scale of model capacity and training data 42Kaplan . State-of-the-art models, such as DALL·E 3 66OpenAI and Stable Diffusion 41Rombach , comprise billions of parameters 65Ramesh ; 64Esser ; 43Ramesh ; 44Saharia , raising concerns regarding computational cost, energy consumption, and further societal implications 45Wynsberghe ; 46Crawford . Beyond synthesising images, generative models have also been applied to time series data 48Brophy ; 49Li ; 50Yoon , adapting similar architectures to capture temporal dependencies and reproduce the statistical properties of the underlying data. While sampling from simple distributions is relatively straightforward, accurately modelling complex, high dimensional distributions remains challenging. Financial time series provide a particularly demanding setting 16Potluru , requiring generative models to produce synthetic sequences that replicate the distribution of log-returns, preserve stylised facts, and maintain the broader statistical structure of real market data. A range of generative approaches have been previously explored in this context. Denoising autoencoders have been used to transfer distributional characteristics onto generated data, producing financial time series that differ sufficiently from purely stochastic simulations 14Silva . GAN-based methods have also been applied to generate realistic financial sequences, including frameworks designed to capture volatility structure 23Wiese , and cross-asset correlations 17Masi . Comparative studies across GANs, VAEs, and generative moment matching networks (GMMNs) suggest that although competitive performance can be achieved across architectures, VAE-based approaches tend to provide more stable and reliable results 5Dogariu . More recently, diffusion-based approaches have been introduced, including methods that operate on wavelet transformed representations of financial data for NASDAQ 4Takahashi and Japanese market datasets 13Tanaka . Score-based generative frameworks have further extended diffusion models by learning the data score function, enabling generation of data across varying temporal resolutions 11Wang . Quantum machine learning (QML) has emerged as an extension of classical machine learning, seeking to exploit the computational properties of quantum hardware through quantum algorithms to enhance model capabilities. Quantum neural networks (QNNs) employ parameterised quantum circuits (PQCs) as a learning framework 52Benedetti ; 54Mitarai ; 53Schuld , leveraging the exponential space state of quantum systems and entanglement between qubits to better represent complex data distributions. This enables expressive representations that can be achieved with significantly fewer trainable parameters than classical counterparts 55Abbas , as demonstrated across tasks such as graph classification 87Qiao . QNNs have been applied to generative learning in both fully quantum models, and hybrid settings where quantum components are embedded within otherwise classical models. A range of quantum generative architectures have been proposed, including Born machines that exploit shallow PQCs for generative tasks on near-term hardware 88Benedetti , and quantum generative adversarial networks (QGANs) 82Lloyd . Implementations of QGANs include models with both quantum generators and discriminators trained via quantum state fidelity objectives 83Braccia ; 84Braccia , as well as entangling QGAN frameworks demonstrated on real quantum hardware 57Niu . Subsequent works have explored variations in circuit design, including compact quantum generators designed to reduce circuit depth 58Chu . Hybrid QGAN frameworks have also been applied to learning and loading probability distributions, including applications in financial derivative pricing 86Zoufal . More recently, quantum Wasserstein GANs have been applied to financial time series generation, utilising QNN-based generators alongside classical or quantum discriminators 15Orlandi ; 22Dechant . This paper focuses on quantum diffusion models (QDMs), motivated by the success of classical diffusion architectures, which achieve state-of-the-art generative performance and have been shown to surpass GAN-based approaches in tasks such as image synthesis 38Dhariwal . QDMs were first proposed in 1Parigi , where quantum noise was leveraged to generate probability distributions that are intractable for classical processors, with the approach demonstrated on synthetic proof-of-concept distributions. In this setting, quantum features can be introduced into either the forward process or into the reverse denoising process through the integration of QNNs. Following the forward quantum noising direction, subsequent physics-inspired QDMs have harnessed the intrinsic noise of real quantum hardware as part of the diffusion process and applied it to image generation 85Parigi . Building on the original framework, the quantum denoising paradigm has been extended in a parallel line of work across a range of generative settings, including image synthesis on multiple datasets 6Kolle and architectures employing repeated circuit layers to enhance expressivity 29Zhang . Quantum latent diffusion models, in which classical autoencoders compress the data and QNNs operate within the reduced feature space, have also been explored 30Falco , including conditional variants for image generation tasks 8Cacioppo , and approaches combining vector quantisation with quantum autoencoders for quantum state generation 81Li . In addition, optimised quantum implicit neural networks (OQINN) have been proposed as alternative denoising models for image synthesis 60Zhang . Despite rapid progress in both quantum machine learning and generative models, to our knowledge no study has applied a QDM to the synthesis of time series data. Here we introduce QDiffusion-TS, that is a QDM for time series generation that extends a classical diffusion architecture designed for temporal data by incorporating QNNs into the denoising process. The model is applied to real financial time series, which exhibit non-stationarity, heavy-tailed distributions, and complex temporal dependencies that are broadly representative of the challenges encountered across time series domains. In this formulation, QNNs (Fig. 1 b) fully replace the feed-forward neural network (FFNN) components within the classical transformer, demonstrating how quantum circuits

Read the complete text