Quantum machine learning is a research area that explores the interplay of ideas from quantum computing and machine learning to speed up the time it takes to train or evaluate a machine learning model. Variational quantum classifiers are among the most widely used quantum models for supervised machine learning and base their operation on learning free parameters through conventional optimization algorithms. In spite of their potential benefits, there is a design issue making quantum variational classifiers not fully operative: the existence of exponentially vanishing gradients, known as “barren plateau landscapes”. A barren plateau is a trainability problem that occurs in optimization algorithms for quantum machine learning when the problem-solving space turns flat as the algorithm is run. In that situation, the algorithm cannot find the downward slope in what appears to be a featureless landscape and there's no clear path to the optimum of the cost function. In this challenging scenario, evolutionary optimization techniques can be a viable choice in solving the problem because of their gradient independence. This paper introduces a comparative study involving different evolutionary algorithms to assess their suitability in acting as optimizer for a specific quantum machine learning method, known as variational quantum classifier.
Quantum machine learning is a research area that explores the interplay of ideas from quantum computing and machine learning to speed up the time it takes to train or evaluate a machine learning model. Variational quantum classifiers are among the most widely used quantum models for supervised machine learning and base their operation on learning free parameters through conventional optimization algorithms. In spite of their potential benefits, there is a design issue making quantum variational classifiers not fully operative: the existence of exponentially vanishing gradients, known as “barren plateau landscapes”. A barren plateau is a trainability problem that occurs in optimization algorithms for quantum machine learning when the problem-solving space turns flat as the algorithm is run. In that situation, the algorithm cannot find the downward slope in what appears to be a featureless landscape and there's no clear path to the optimum of the cost function. In this challenging scenario, evolutionary optimization techniques can be a viable choice in solving the problem because of their gradient independence. This paper introduces a comparative study involving different evolutionary algorithms to assess their suitability in acting as optimizer for a specific quantum machine learning method, known as variational quantum classifier.
Quantum supervised learning, utilizing variational circuits, stands out as a promising technology for NISQ devices due to its efficiency in hardware resource utilization during the creation of quantum feature maps and the implementation of hardware-efficient ansatz with trainable parameters. Despite these advantages, the training of quantum models encounters challenges, notably the barren plateau phenomenon, leading to stagnation in learning during optimization iterations. This study proposes an innovative approach: an evolutionary-enhanced ansatz-free supervised learning model. In contrast to parametrized circuits, our model employs circuits with variable topology that evolves through an elitist method, mitigating the barren plateau issue. Additionally, we introduce a novel concept, the superposition of multi-hot encodings, facilitating the treatment of multi-classification problems. Our framework successfully avoids barren plateaus, resulting in enhanced model accuracy. Comparative analysis with variational quantum classifiers from the technology's state-of-the-art reveal a substantial improvement in training efficiency and precision. Furthermore, we conduct tests on a challenging dataset class, traditionally problematic for conventional kernel machines, demonstrating a potential alternative path for achieving quantum advantage in supervised learning for NISQ era.
Variational Quantum Classifiers (VQCs) stand out as one of the most popular classification approaches within the Quantum Machine Learning (QML) community. They serve as an accessible entry point for newcomers to QML world, particularly those with a background in classical machine learning, due to their workflow similarity to conventional neural networks. In essence, VQCs boast a layered structure conducive to training via a classical optimizer. Recently, Evolutionary Algorithms (EAs) have emerged as a suitable methodology for this training goal. However, current quantum tools provide implementations for various gradient-based optimizers and just one EA tailored for VQC training. Moreover, it is important to note that access to this toolset remains largely confined to expert users. This highlights the necessity for platforms that facilitate the utilization of EAs for VQC training by non-experts in quantum or evolutionary computing. This work introduces a web application designed to implement quantum classifiers to be trained on user-supplied datasets by means of a set of EAs, which can be customized by users. The capabilities of the proposed web application are shown in the experimental session, where a VQC is trained to classify a well-known dataset using a variety of EAs.
Recent advancements in machine learning have revolutionised research across various fields. Despite their success, conventional learning techniques are hindered by their significant computational resources and energy requirements. Prompted by recent experimental breakthroughs in quantum computing, variational quantum machine learning (QML) – machine learning integrated with variational quantum circuits (VQCs) – has emerged as a promising alternative. Nonetheless, the theoretical framework underpinning the advantages of variational QML is still rudimentary. Specifically, the training of VQCs faces several challenges, such as the barren plateau problem, where the gradient diminishes exponentially with an increasing qubits. A related issue arises in variational QML training, where the convergence rate is exponentially small. In this thesis, we present theoretically guaranteed solutions to these challenges. First, we construct innovative circuit architectures to address the vanishing gradient problem in deep VQCs. We propose quantum controlled-layer and quantum ResNet structures, demonstrating that the expected gradient norm's lower bound is unaffected by the increase in qubits and circuit depth. Next, we introduce an initialization strategy to mitigate the vanishing gradient issue in general deep quantum circuits. We prove that Gaussian-initialized parameters ensure the gradient norm's decay rate remains inversely polynomial despite the increase in qubit numbers and circuit depth. Finally, we propose a novel and effective theory for analysing the training of quantum neural networks with moderate depths. We prove that, under certain randomness conditions in the circuits and datasets, training converges linearly with a rate inversely proportional to the dataset size. Our approach surpasses previous results, achieving exponentially larger convergence rates with modest depth, or conversely, requiring exponentially less depth for equivalent rates.
This lecture provides an overview of quantum machine learning (QML), focusing on two prominent paradigms: Quantum Neural Networks (QNNs) and Quantum Support Vector Machines (QSVMs). We begin by introducing QNNs, exploring their structure as variational quantum circuits and the challenges associated with their training. We will cover key practical issues such as the barren plateau phenomenon, which can hinder eective gradient-based optimization. Next, the lecture will turn to Quantum Support Vector Machines, an algorithm that leverages quantum feature maps to encode classical data into a high-dimensional quantum Hilbert space. We will conclude by comparing the strengths and weaknesses of both QNNs and QSVMs in the current QML landscape, oering a realistic perspective on their potential and the research needed to realize their full capabilities.
Recent advances in Quantum Machine Learning (QML) bring together Quantum Computing and Machine Learning with the aim of enhancing the advantages of Machine Learning to solve problems efficiently. The properties of quantum systems (superposition, entanglement, etc.) to represent multiple states of information at once, QML can potentially offer significant advantages when solving complex high-dimensional (multidimensional) computational problems. This paper reviews the current state-of-art regarding QML in general through an organized compilation and assessment of selected studies recently published in scientific literature that were found via large electronic library catalogs based on peer-reviewed sources according to algorithms used, optimization approaches employed, and areas of applied study. This review provides comparative analyses of each selected studies for purposes of identifying major trends in QML research/practice as well as opportunities for methodological improvements and challenges still existing within QML. Finally, this review presents an overview of Quantum Computing and Quantum Data Encoding Principles utilized to translate Classical Data into Quantum States as a high-level evaluation of several popular Quantum Learning Algorithms (e.g., Quantum Neural Networks, Quantum Support Machines, and Variational Quantum Circuit). The survey describes the optimization strategies such as gradient based techniques, Bayesian Optimization and evolutionary techniques being used to improve the stability of the quantum machine learning (QML) training process. Then, it provides examples of application areas for such QML, where healthcare analytics, environmental modelling, manufacturing optimization and analysis of biologic data are included. The comparative review of the recent academic research has identified several limitations and challenges regarding the use of QML that researchers. Although QML is believed to be capable of solving complex computational problems, issues such as noise in hardware, limited numbers of qubits and scalability present significant challenges for researchers working with this type of machine learning application. Overall, it appears that this survey provides a broad overview of contemporary developments and future trends for developing high-performance quantum learning systems.
Quantum computing is rapidly advancing and offering new ways to improve artificial intelligence (AI). The most recent advancement is Quantum Machine Learning (QML), an exciting new field that involves the exploration of combining quantum computing principles with machine learning methods to solve complex problems that have high-dimensional problem-solving models that are impossible to address with classical models. The objective of this research was a descriptive-analytical examination of combining classical machine learning with quantum computing principles, specifically using Variational Quantum Classifiers (VQC) to train on a nonlinear XOR dataset, to assess the proposition of QML. The VQC was able to amongst quantum features, and learning how to classify using parameterized variational circuits to model nonlinear decision boundaries as models not achievable using classical linear classifiers. Training convergence was demonstrated with decreasing training and test losses, with the final model achieving test accuracy of nearly 0.85. The training curves and the decision boundary plots illustrated the model’s ability to learn the more complex decision boundaries. This study demonstrated QML's ability to coexist with classical machine learning studies with the ability to exploit quantum entanglement and quantum parallelism. This paper concluded that QML has great potential to challenge the concerns of next generation applications in finance, healthcare, and secure communications and to optimize computational power and scalability.