Abstract We reduce measurement errors in a quantum computer using machine learning techniques. We exploit a simple yet versatile neural network to classify multi-qubit quantum states, which is trained using experimental data. This flexible approach allows the incorporation of any number of features of the data with minimal modifications to the underlying network architecture. We experimentally illustrate this approach in the readout of trapped-ion qubits using additional spatial and temporal features in the data. Using this neural network classifier, we efficiently treat qubit readout crosstalk, resulting in a 30% improvement in detection error over the conventional threshold method. Our approach does not depend on the specific details of the system and can be readily generalized to other quantum computing platforms.
Abstract We reduce measurement errors in a quantum computer using machine learning techniques. We exploit a simple yet versatile neural network to classify multi-qubit quantum states, which is trained using experimental data. This flexible approach allows the incorporation of any number of features of the data with minimal modifications to the underlying network architecture. We experimentally illustrate this approach in the readout of trapped-ion qubits using additional spatial and temporal features in the data. Using this neural network classifier, we efficiently treat qubit readout crosstalk, resulting in a 30% improvement in detection error over the conventional threshold method. Our approach does not depend on the specific details of the system and can be readily generalized to other quantum computing platforms.
We demonstrate qubit-state measurements assisted by a supervised convolutional neural network (CNN) in a neutral-atom quantum processor. We present two CNN architectures for analyzing neutral-atom qubit readout data: a compact five-layer single-qubit CNN architecture and a six-layer multiqubit CNN architecture. We benchmark both architectures against a conventional Gaussian-threshold analysis method. In a sparse array (9-$\text{\ensuremath{\mu}}\mathrm{m}$ atom separation) which experiences negligible crosstalk, we have observed up to 32% and 56% error reduction for the multiqubit and single-qubit architectures, respectively, as compared to the benchmark. In a tightly spaced array (5-$\text{\ensuremath{\mu}}\mathrm{m}$ atom separation), which suffers from readout crosstalk, we have observed up to 43% and 32% error reduction in the multiqubit and single-qubit CNN architectures, respectively, as compared to the benchmark. By examining the correlation between the predicted states of neighboring qubits, we have found that the multiqubit CNN architecture reduces the crosstalk correlation by up to 78.5%. This work demonstrates a proof of concept for a CNN network to be implemented as a real-time readout-processing method on a neutral-atom quantum computer, enabling faster readout time and improved fidelity.
Quantum computers require high-fidelity measurement of many qubits to achieve a quantum advantage. Traditional approaches suffer from readout crosstalk for a neutral-atom quantum processor with a tightly spaced array. Although classical machine learning algorithms based on convolutional neural networks can improve fidelity, they are computationally expensive, making it difficult to scale them to large qubit counts. We present two simpler and scalable machine learning algorithms that realize matched filters for the readout problem. One is a local model that focuses on a single qubit, and the other uses information from neighboring qubits in the array to prevent crosstalk among the qubits. We demonstrate error reductions of up to 32% and 43% for the site and array models, respectively, compared to a conventional Gaussian threshold approach. Additionally, our array model uses two orders of magnitude fewer trainable parameters and four orders of magnitude fewer multiplications and nonlinear function evaluations than a recent convolutional neural network approach, with only a minor (3.5%) increase in error across different readout times. Another strength of our approach is its physical interpretability: the learned filter can be visualized to provide insights into experimental imperfections. We also show that a convolutional neural network model for improved can be pruned to have 70x and 4000x fewer parameters, respectively, while maintaining similar errors. Our work shows that simple machine learning approaches can achieve high-fidelity qubit measurements while remaining scalable to systems with larger qubit counts.
In recent years, quantum simulators have been the focus of intense research due to their potential in unraveling the inner workings of complex quantum systems. The exponential scaling of the Hilbert space of quantum systems limits the capabilities of classical approaches. Quantum simulators, on the other hand, are suited to efficiently emulate these quantum systems for in-depth research. Out of the various platforms for quantum simulators, trapped-ions have been prized for their long coherence times, exceptional initialization and detection fidelities and their innate full-connectivity. In the development of the trapped-ion apparatus as a quantum simulator, numerous optimization problems emerge. In this thesis, we examine, develop and refine the strategies available to us for addressing such problems. A class of techniques of particular interest to us is the set of machine learning algorithms. This can be attributed to their capabilities at identifying correlations in massive data sets. \nSpecifically, we describe how artificial neural networks can be employed to assist in the programming of the trapped-ion system as an arbitrary spin model quantum simulator. In addition, we present an augmentation to the trapped-ion architecture, utilizing optical tweezers, that enables the programmable manipulation of phonon modes. We describe the protocols developed to effectively program the phonon modes and propose an application of the scheme for enhancing the performance of multi-species systems. Finally, we explore the use of machine learning algorithms to perform state readout of the trapped-ion system. The work in this thesis extends the utility of the trapped-ion system for performing quantum information processing experiments.
Measurement of qubits plays a key role in quantum computation. In superconducting multiqubit quantum processors, a multiplexed readout scheme is widely used. In such a scheme, measurement of a qubit state may be influenced by the state of neighboring qubits, due to various crosstalk effects, which will degrade the readout fidelity. To reduce the impact of crosstalk, we model the digital signal processing system used in measurements as a shallow neural network and train it to become a state discriminator. Applying our method to a six-qubit superconducting quantum chip, we see an overall improved readout performance compared with a contemporary qubit-state discriminator. The readout crosstalk is decreased by more than $80\mathrm{%}$. The training and optimization process of the neural network consumes only about 10 s.
Quantum processors require rapid and high-fidelity simultaneous measurements of many qubits. While superconducting qubits are among the leading modalities toward a useful quantum processor, their readout remains a bottleneck. Traditional approaches to processing measurement data often struggle to account for crosstalk present in frequency-multiplexed readout, the preferred method to reduce the resource overhead. Recent approaches to address this challenge use neural networks to improve the state-discrimination fidelity. However, they are computationally expensive to train and evaluate, resulting in increased latency and poor scalability as the number of qubits increases. We present an alternative machine learning approach based on next-generation reservoir computing that constructs polynomial features from the measurement signals and maps them to the corresponding qubit states. This method is highly parallelizable, avoids the costly nonlinear activation functions common in neural networks, and supports real-time training, enabling fast evaluation, adaptability, and scalability. Despite its lower computational complexity, our reservoir approach is able to maintain high qubit-state-discrimination fidelity. Relative to traditional methods, our approach achieves error reductions of up to 50% and 11% on single- and five-qubit datasets, respectively, and delivers up to 2.5x crosstalk reduction on the five-qubit dataset. Compared with recent machine-learning methods, evaluating our model requires 100x fewer multiplications for single-qubit and 2.5x fewer for five-qubit models. This work demonstrates that reservoir computing can enhance qubit-state discrimination while maintaining scalability for future quantum processors.
We report the realization of a versatile classifier based on the quantum mechanics of a single atom. The problem of classification has been extensively studied by the classical machine learning community, with plenty of proposed algorithms that have been refined over time. Quantum computation must necessarily develop quantum classifiers and benchmark them against their classical counterparts. It is not obvious how to make use of our increasing ability to precisely control and evolve a quantum state to solve this kind of problems, while there is only a limited number of strong theorems backing the quantum algorithms for classification. Here we show that both of these limitations can be successfully addressed by the implementation of a recently proposed data re-uploading algorithm in an ion trap based quantum processing unit. The quantum classifier is trained in two steps: first, the quantum circuit is fed with an optimal set of variational parameters found by classical simulation; then, the variational circuit is optimized by inspecting the parameter landscape with only the quantum processing unit. This second step provides a partial cancellation of the systematic errors inherent to the quantum device. The accuracy of our quantum supervised classifier is benchmarked on a variety of datasets, that imply, finding the separation of classes associated to regions in a plane in both binary and multi-class problems, as well as in higher-dimensional feature spaces. Our experiments show that a single-ion quantum classifier circuit made out of k gates is as powerful as a neural network with one intermediate hidden layer of k neurons.