arXiv:2301.09363 · ACM Transactions on Quantum Computing
A Characterization of Quantum Generative Models
Carlos A. Riofrío, Oliver Mitevski, C. R. Jones, Florian Krellner, Aleksandar Vuckovic, Joseph A. Doetsch, Johannes Klepsch, Thomas Ehmer, André Luckow
Quantum generative modeling is a growing area of interest for industry-relevant applications. This work systematically compares a broad range of techniques to guide quantum computing practitioners when deciding which models and methods to use in their applications. We compare fundamentally different architectural ansatzes of parametric quantum circuits: (1) A continuous architecture, which produces continuous-valued data samples, and (2) a discrete architecture, which samples on a discrete grid. We also compare the performance of different data transformations: the min-max and the probability integral transforms. We use two popular training methods: (1) quantum circuit Born machines (QCBM), and (2) quantum generative adversarial networks (QGAN). We study their performance and tradeoffs as the number of model parameters increases, with a baseline comparison of similarly trained classical neural networks. The study is performed on six low-dimensional synthetic and two real financial data sets. Our two key findings are that: (1) For all data sets, our quantum models require similar or fewer parameters than their classical counterparts. In the extreme case, the quantum models require two orders of magnitude less parameters. (2) We empirically find that a variant of the discrete architecture, which learns the copula of the probability distribution, outperforms all other methods.
Quantum generative modeling is a growing area of interest for industry-relevant applications. This work systematically compares a broad range of techniques to guide quantum computing practitioners when deciding which models and methods to use in their applications. We compare fundamentally different architectural ansatzes of parametric quantum circuits: (1) A continuous architecture, which produces continuous-valued data samples, and (2) a discrete architecture, which samples on a discrete grid. We also compare the performance of different data transformations: the min-max and the probability integral transforms. We use two popular training methods: (1) quantum circuit Born machines (QCBM), and (2) quantum generative adversarial networks (QGAN). We study their performance and tradeoffs as the number of model parameters increases, with a baseline comparison of similarly trained classical neural networks. The study is performed on six low-dimensional synthetic and two real financial data sets. Our two key findings are that: (1) For all data sets, our quantum models require similar or fewer parameters than their classical counterparts. In the extreme case, the quantum models require two orders of magnitude less parameters. (2) We empirically find that a variant of the discrete architecture, which learns the copula of the probability distribution, outperforms all other methods.
This chapter develops quantum generative models as a distinct paradigm within quantum machine learning (QML), emphasizing distribution learning and sampling rather than prediction. It begins by showing how quantum states naturally define probability distributions through the Born rule, thereby making sampling an intrinsic physical process rather than an auxiliary computational step. On this foundation, this chapter presents three principal frameworks: Quantum circuit Born machines (QCBMs), quantum generative adversarial networks (QGANs), and quantum Boltzmann machines (QBMs). QCBMs are introduced as the most direct and conceptually transparent generative models based on parameterized quantum circuits; QGANs extend adversarial learning into the quantum domain through generator–discriminator competition; and QBMs connect generative modeling to quantum statistical mechanics through Gibbs states and quantum Hamiltonians. This chapter compares these models in terms of expressivity, training stability, and hardware feasibility, highlighting the recurring tension between representational richness and practical trainability. It also examines key application areas, including synthetic data generation, scientific distribution learning, quantum simulation, and hybrid classical-quantum pipelines, while emphasizing major open challenges such as noise, sampling overhead, limited hardware scalability, and the absence of clearly demonstrated quantum advantage. Taken together, this chapter establishes generative modeling as a central pillar of QML.
Quantum Generative Adversarial Network (qGAN) intends to incorporate all the core developing quantum computing ideas into a quantum-classical hybrid architecture. With an idea of constructing a qGAN, the whole process uses parameterized quantum circuits as generators and classical neural networks as discriminators to learn complex probability distributions from data, in high-dimensional again for examples in housing data or physics simulations. The generator makes use of quantum states to encode the data samples facilitating exploration optimally through phenomenona, such as entanglement and superposition, with, for instance, possible combinations of housing states replicating the underlying statistics of high-dimensional data sets, such the ones present in housing data or physics simulations, while for the discriminator estimating validity for samples generated by the generator. Herewith adversarial training, the parameters of the generator will tend to be adapted iteratively producing data that does not differ in any way from actual samples in obtaining models made of superposition and extension entanglement. The method combines quantum toolkits such as Qiskit with classical technologies like PyTorch to investigate whether there is any benefit with respect to quantum capabilities for generative modeling tasks. In spite of the other difficulties such as limited hardware and the complication of gradient computations, these qGANs are expected to improve generative models for applications from generating synthetic data to simulating high-energy physics.
Abstract Quantum generative adversarial networks (QGANs) have been studied in the context of quantum machine learning for several years, but there has not been yet a proposal for a fully QGAN with both, a quantum generator and discriminator. We introduce a fully QGAN intended for use with binary data. The architecture incorporates several features found in other classical and quantum machine learning models, which up to this point had not been used in conjunction. In particular, we incorporate classical noise reuploading in the generator, auxiliary qubits in the discriminator to enhance expressivity, and a direct connection between the generator and discriminator circuits, obviating the need to access the generator’s probability distribution. We show that, as separate components, the generator and discriminator perform as desired. We empirically demonstrate the expressive power of our model on both synthetic data as well as low energy states of an Ising model. Our demonstrations suggest that the model is not only capable of reproducing discrete training data, but also of potentially generalizing from it.
Abstract Generative modeling has seen a rising interest in both classical and quantum machine learning, and it represents a promising candidate to obtain a practical quantum advantage in the near term. In this study, we build over an existing framework for evaluating the generalization performance of generative models, and we establish the first quantitative comparative race towards practical quantum advantage (PQA) between classical and quantum generative models, namely Quantum Circuit Born Machines (QCBMs), Transformers (TFs), Recurrent Neural Networks (RNNs), Variational Autoencoders (VAEs), and Wasserstein Generative Adversarial Networks (WGANs). After defining four types of PQAs scenarios, we focus on what we refer to as potential PQA, aiming to compare quantum models with the best-known classical algorithms for the task at hand. We let the models race on a well-defined and application-relevant competition setting, where we illustrate and demonstrate our framework on 20 variables (qubits) generative modeling task. Our results suggest that QCBMs are more efficient in the data-limited regime than the other state-of-the-art classical generative models. Such a feature is highly desirable in a wide range of real-world applications where the available data is scarce.
Quantum generative models harness the probabilistic nature of quantum mechanics to represent and learn complex distributions, offering a novel framework for both classical and quantum data modeling. This chapter introduces foundational architectures such as quantum circuit Born machines, quantum generative adversarial networks, and variational autoencoders, emphasizing their expressivity, training methods, and applications. We explore use cases ranging from anomaly detection to spectral function sampling and discuss key challenges including trainability, noise, and evaluation metrics. The chapter concludes by outlining opportunities for quantum advantage in generative tasks and highlighting emerging strategies for scalable, noise-resilient model training on near-term quantum devices.
Generative quantum machine learning models are trained to deduce the probability distribution underlying a given dataset, and to produce new, synthetic samples from it. The majority of such models proposed in the literature, like the Quantum Circuit Born Machine (QCBM), fundamentally work on a binary level. Real-world data, however, is often numeric, requiring the models to translate between binary and continuous representations. We analyze how this transition influences the performance of quantum models and show that it requires the models to learn correlations that are solely an artifact of the way the data is encoded, and not related to the data itself. At the same time, structure of the original data, like continuity, can be obscured in the binary representation, hindering generalization. To mitigate these effects, we propose a strategy based on Gray-codes that can be implemented with essentially no overhead, conserves structures in the data, and avoids artificial correlations in situations in which the standard approach creates them. Considering datasets drawn from various low-dimensional probability distributions, we verify that, in most cases, QCBMs using the reflected Gray code learn faster and more accurately than those with standard binary code. This shows that, as complement to specifically tailored circuit ansätze or data pre-processing, binary encodings can be used to introduce inductive biases into generative machine learning models.