Universality of neural networks describes the ability to approximate arbitrary function, and is a key ingredient to keep the method effective. The established models for universal quantum neural networks(QNN), however, require the preparation of multiple copies of the same quantum state to generate the nonlinearity, with the copy number increasing significantly for highly oscillating functions, resulting in a huge demand for a large-scale quantum processor. To address this problem, we propose a new QNN model that harbors universality without the need of multiple state-duplications, and is more likely to get implemented on near-term devices. To demonstrate the effectiveness, we compare our proposal with two popular QNN models in solving typical supervised learning problems. We find that our model requires significantly fewer qubits and it outperforms the other two in terms of accuracy and relative error.
Universality of neural networks describes the ability to approximate arbitrary function, and is a key ingredient to keep the method effective. The established models for universal quantum neural networks(QNN), however, require the preparation of multiple copies of the same quantum state to generate the nonlinearity, with the copy number increasing significantly for highly oscillating functions, resulting in a huge demand for a large-scale quantum processor. To address this problem, we propose a new QNN model that harbors universality without the need of multiple state-duplications, and is more likely to get implemented on near-term devices. To demonstrate the effectiveness, we compare our proposal with two popular QNN models in solving typical supervised learning problems. We find that our model requires significantly fewer qubits and it outperforms the other two in terms of accuracy and relative error.
Abstract Quantum machine learning, focusing on quantum neural networks (QNNs), remains a vastly uncharted field of study. Current QNN models primarily employ variational circuits on an ansatz or a quantum feature map, often requiring multiple entanglement layers. This methodology not only increases the computational cost of the circuit beyond what is practical on near-term quantum devices but also misleadingly labels these models as neural networks, given their divergence from the structure of a typical feed-forward neural network (FFNN). Moreover, the circuit depth and qubit needs of these models scale poorly with the number of data features, resulting in an efficiency challenge for real-world machine learning tasks. We introduce a bona fide QNN model, which seamlessly aligns with the versatility of a traditional FFNN in terms of its adaptable intermediate layers and nodes, absent from intermediate measurements such that our entire model is coherent. This model stands out with its reduced circuit depth and number of requisite CNOT gates, achieving a more than 50% reduction in both compared to prevailing QNN models. Furthermore, the qubit count in our model remains unaffected by the data’s feature quantity. We test our proposed model on various benchmarking datasets such as the breast cancer diagnostic (Wisconsin) and credit card fraud detection datasets. Our model achieved an accuracy of 91% on the breast cancer dataset and 85% on the credit card fraud detection dataset, outperforming existing QNN methods by 5–10% while requiring approximately 50% fewer quantum resources. These results showcase the advantageous efficacy of our approach, paving the way for the application of quantum neural networks to relevant real-world machine learning problems.
As one of the well-known methods to solve natural language processing (NLP) problems, the long short-term memory neural network (LSTM) has been designed to implement on quantum computers. However, existing method requires the number of qubits is equal to the dimension of the word vectors, and may result in a surge of the qubit number. To address this problem, we propose a duplication-free quantum long short-term memory neural network (DQLSTM). Specifically, our DQLSTM adopts the amplitude encoding method to store the classical information, and further can reduce the requirement for the number of qubits. Our numerical results show that our DQLSTM has similar effectiveness to the classical counterpart in a Chinese sentiment analysis task, and outperforms several previous proposals in the same task.
Today's quantum processors composed of fifty or more qubits have allowed us to enter a computational era where the output results are not easily simulatable on the world's biggest supercomputers. What we have not seen yet, however, is whether or not the complexity achievable with such quantum processors would ever be useful for any practical applications. In the quantum neural network approach, there has been an expectation that the Hilbert space can serve as a computational resource by providing a large feature space; however, it is not understood, nor illustrated in what way the Hilbert space provides that computational power. In this work we introduce a resource-efficient quantum neural network model with a focus on the role of the feature space the quantum processors can give. Then we apply it for classification tasks, showing a quantum advantage in terms of the physical resources needed to realize the neural network. To facilitate the Hilbert space as a large feature space, we utilize scale-free networks that can be generated in a discrete-time crystal model. The virtue of this approach is in both its theoretical simplicity and its practical applicability: our quantum neural network model is simple enough to identify the role of the Hilbert space as the feature space, and, furthermore, can be implemented with current technology. Our model does not require optimization of the quantum processor, and hence once the quantum processor has been set, it can be used for other different classification problems without any changes in the quantum processor itself.
One of the most promising new directions in quantum computing and artificial intelligence convergence is the Quantum Neural Networks (QNNs) that can potentially revolutionize the manner in which intelligent systems are designed, trained, and implemented. Although traditional deep learning models have proven to be incredibly successful in fields like computer vision, natural language processing and autonomous systems, they are becoming limited by computational bottlenecks, energy use, and scalability constraints when operating on large high-dimensional data. The distinctive principles of quantum mechanics (superposition, entanglement, and quantum parallelism) may be applied to the QNNs to provide a distinct computational paradigm with the ability to execute exponentially complicated computations more effectively than his classical counterparts. This article gives an in-depth analysis of QNNs as the new frontier of artificial intelligence. We start with the description of the theoretical foundations of quantum computation and their realization in the form of neural networks. We then examine state-of-the-art QNN architectures, such as hybrid quantum-classical models that help overcome present quantum hardware limitations in relation to the capabilities of classical machine learning. Some of the upcoming applications are showcased including molecular simulation and drug discovery and financial modeling, optimization, cryptography and intelligent automation. Such applications reveal the revolutionary power of QNNs to tackle the issues that cannot be tackled by traditional systems. Nonetheless, there are also major obstacles on the way to a practical implementation of QNN. Problems that still remain a barrier to widespread adoption include quantum decoherence, noise in near-term quantum devices, and the inability to scale qubit architectures or to train large-scale quantum models. In order to overcome these obstacles, we speak about the current research work on the topic of error correction, variational algorithms, and hybrid system designs. Also we examine the ethical and societal ramifications of quantum-accelerated AI, such as security, data sovereignty, and technological equity globally.
Abstract The field of artificial neural networks is expected to strongly benefit from recent developments of quantum computers. In particular, quantum machine learning, a class of quantum algorithms which exploit qubits for creating trainable neural networks, will provide more power to solve problems such as pattern recognition, clustering and machine learning in general. The building block of feed-forward neural networks consists of one layer of neurons connected to an output neuron that is activated according to an arbitrary activation function. The corresponding learning algorithm goes under the name of Rosenblatt perceptron. Quantum perceptrons with specific activation functions are known, but a general method to realize arbitrary activation functions on a quantum computer is still lacking. Here, we fill this gap with a quantum algorithm which is capable to approximate any analytic activation functions to any given order of its power series. Unlike previous proposals providing irreversible measurement–based and simplified activation functions, here we show how to approximate any analytic function to any required accuracy without the need to measure the states encoding the information. Thanks to the generality of this construction, any feed-forward neural network may acquire the universal approximation properties according to Hornik’s theorem. Our results recast the science of artificial neural networks in the architecture of gate-model quantum computers.
Abstract The resurgence of self-supervised learning, whereby a deep learning model generates its own supervisory signal from the data, promises a scalable way to tackle the dramatically increasing size of real-world data sets without human annotation. However, the staggering computational complexity of these methods is such that for state-of-the-art performance, classical hardware requirements represent a significant bottleneck to further progress. Here we take the first steps to understanding whether quantum neural networks (QNNs) could meet the demand for more powerful architectures and test its effectiveness in proof-of-principle hybrid experiments. Interestingly, we observe a numerical advantage for the learning of visual representations using small-scale QNN over equivalently structured classical networks, even when the quantum circuits are sampled with only 100 shots. Furthermore, we apply our best quantum model to classify unseen images on the ibmq_paris quantum computer and find that current noisy devices can already achieve equal accuracy to the equivalent classical model on downstream tasks.