Abstract Neutral atoms devices represent a promising technology using optical tweezers to geometrically arrange atoms and modulated laser pulses to control their quantum states. They are exploited as noisy intermediate‐scale quantum (NISQ) processors. Indeed, like all real quantum devices, they are affected by noise introducing errors in the computation. Therefore, it is important to understand and characterize the noise sources and possibly to correct them. Here, two machine‐learning based approaches are proposed respectively to estimate the noise parameters and to mitigate their effects using only measurements of the final quantum state. Our analysis is then tested on a real neutral atom platform, comparing our predictions with a priori estimated parameters. It turns out that increasing the number of atoms is less effective than using more measurements on a smaller scale. The agreement is not always good but this may be due to the limited amount of real data that are obtained from a still under development device. Finally, reinforcement learning is employed to design a pulse that mitigates the noise effects. Our machine learning‐based approach is espected to be very useful for the noise benchmarking of NISQ processors and, more in general, of real quantum technologies.
Abstract Neutral atoms devices represent a promising technology using optical tweezers to geometrically arrange atoms and modulated laser pulses to control their quantum states. They are exploited as noisy intermediate‐scale quantum (NISQ) processors. Indeed, like all real quantum devices, they are affected by noise introducing errors in the computation. Therefore, it is important to understand and characterize the noise sources and possibly to correct them. Here, two machine‐learning based approaches are proposed respectively to estimate the noise parameters and to mitigate their effects using only measurements of the final quantum state. Our analysis is then tested on a real neutral atom platform, comparing our predictions with a priori estimated parameters. It turns out that increasing the number of atoms is less effective than using more measurements on a smaller scale. The agreement is not always good but this may be due to the limited amount of real data that are obtained from a still under development device. Finally, reinforcement learning is employed to design a pulse that mitigates the noise effects. Our machine learning‐based approach is espected to be very useful for the noise benchmarking of NISQ processors and, more in general, of real quantum technologies.
Abstract The effects of noise are one of the most important factors to consider when it comes to quantum computing in the noisy intermediate-scale quantum computing (NISQ) era that we are currently in. Therefore, it is important not only to gain more knowledge about the noise sources appearing in current quantum computing hardware in order to suppress and mitigate their contributions, but also to evaluate whether a given quantum algorithm can achieve reasonable results on a given hardware. To accomplish this, we need noise models that can describe the real hardware with sufficient accuracy. Here, we present a noise model that has been evaluated on superconducting hardware platforms and could be adapted to other common architectures such as trapped-ion or neutral atom devices. We then benchmark our model by simulating a 20-qubit superconducting quantum computer, and compare the accuracy of our model to similar approaches from the literature and demonstrate an improvement in the overall prediction accuracy.
Quantum computing has emerged as a transformative paradigm capable of addressing computational challenges beyond the reach of classical systems.In the current era of Noisy Intermediate-Scale Quantum (NISQ) devices, practical quantum hardware is characterized by limited qubit counts, short coherence times, gate errors, and the absence of full error correction.Despite these constraints, significant progress has been made in developing quantum machine learning (QML) techniques tailored to operate on near-term quantum processors.This paper presents a comprehensive review of Quantum Machine Learning in the context of NISQ computing.We first outline the fundamental principles of quantum computing, including qubits, superposition, entanglement, and quantum parallelism, followed by a detailed discussion of the NISQ computing model and its hardware limitations.The review then examines hybrid quantum-classical learning frameworks, variational quantum algorithms, quantum data encoding strategies, and noise mitigation techniques that enable practical implementation on existing platforms.Emerging applications across healthcare, finance, cybersecurity, climate modeling, and materials science are also surveyed to highlight the real-world potential of QML.Furthermore, key challenges such as scalability, trainability issues, and algorithmic robustness are analyzed, along with current research efforts aimed at overcoming these barriers.The paper concludes by identifying promising future directions for advancing QML toward quantum advantage.This review aims to provide researchers and practitioners with a structured understanding of the capabilities, limitations, and opportunities of quantum machine learning in the NISQ era .
Quantum computing represents one of the most transformative technological developments of the twenty-first century. However, the practical realization of its potential on current Noisy Intermediate-Scale Quantum (NISQ) hardware remains severely constrained by noise and computational errors that arise from the inherent fragility of quantum states. This paper presents a systematic comparative analysis of three prominent quantum error mitigation techniques, partial quantum error correction (QEC), zero-noise extrapolation (ZNE), and deep learning-based noise prediction, and develops a predictive assessment of a proposed hybrid framework that integrates all three into a unified system. Through a thorough review of twelve peer-reviewed publications spanning 2017 to 2026, we establish what each technique has accomplished individually and to what degree of accuracy. We identify a critical and previously unaddressed gap: no existing work has systematically combined these three complementary approaches. Drawing on documented performance trends and theoretical complementarity, we predict that a well-designed hybrid framework could achieve classification accuracies of 85 to 90 percent on standard quantum machine learning benchmarks, compared to only 60 to 65 percent without error mitigation. These predictions are grounded in peer-reviewed experimental data. Our analysis provides a clear evidence-based roadmap for future experimental validation and establishes a strong foundation for the development of practical, resource-efficient quantum machine learning systems on near-term hardware. CCS Concepts: Hardware → Quantum computation; • Computing methodologies → Machine learning; • Theory of computation → Quantum complexity theory
ABSTRACT: Recent research has tackled the problem of mitigating noise present in quantum computers in the Noisy Intermediate-Scale Quantum (NISQ) era to enable precise computations and to benefit from the intrinsic properties of quantum mechanics. For this matter, an important task is the characterization of qubits available in quantum devices so as to provide insights on how to reduce noise on the final output of a quantum circuit. Characterization comprises analysis of noise sources and this information can be used to reduce noise with methods such as Cycle Benchmarking, Quantum Error Mitigation, Quantum Error Correction and others. Here, we study optimization of pulses through Quantum Optimal Control (QOC) to obtain higher gate fidelity. We will explore an algorithm that performs calibration on specific quantum gates by implementing optimized pulse schedules to subsequently use the algorithm for analysis of noise sources. Using calibrated gates as input, several benchmarking protocols, including pulse noise extrapolation, leakage analysis from quantum optimal control, and machine learning based classification of qubit readout, will be tested to extract precise information on how noise influences the analyzed qubits. We will explain and discuss different techniques for obtaining properties of qubits and quantum computers. We will implement state discrimination with a Machine Learning (ML) focus to analyze readout errors caused by factors such as cross-talk and leakage into higher quantum states. We will perform noise fitting of optimized pulses and evaluate the effectiveness of important quantum algorithms at the pulse level.
A quantum error mitigation technique based on machine learning is proposed, which learns how to adjust the probabilities estimated by measurement in the computational basis. Neural networks in two different designs are trained with random quantum circuits consisting of a set of one- and two-qubit unitary gates whose measurement statistics in the ideal (noiseless) and real (noisy) cases are known. Once the neural networks are trained, they infer the amount of probability adjustment to be made on the measurement obtained from executing an unseen quantum circuit to reduce the error. The proposed schemes are tested with two-, three-, five-, and seven-qubit quantum circuits of depth up to 20 by computer simulations with realistic error models and experiments using the IBM quantum cloud platform. In all test cases, the proposed mitigation technique reduces the error effectively. Our method can be used to improve the accuracy of noisy intermediate-scale quantum (NISQ) algorithms without relying on extensive error characterization or quantum error correction.
Noisy intermediate-scale quantum (NISQ) devices are nowadays starting to become available to the final user, hence potentially allowing the quantum speedups predicted by quantum-information theory to be shown. However, before implementing any quantum algorithm, it is crucial to have at least a partial or possibly full knowledge on the type and amount of noise affecting the quantum machine. Here, by generalizing quantum generative adversarial learning from quantum states (QGANs) to quantum operations, superoperators, and channels (here named super QGANs), we propose a very promising framework to characterize noise in a realistic quantum device, even in the case of spatially and temporally correlated noise (memory channels) affecting quantum circuits. The key idea is to learn about the noise by mimicking it in a way that one cannot distinguish between the real (to be sensed) and the fake (generated) one. We find that, when applied to the benchmarking case of Pauli channels, the super-QGAN protocol is able to learn the associated error rates even in the case of spatially and temporally correlated noise. Moreover, we also show how to employ it for quantum metrology applications. We believe our super QGANs pave the way for the development of hybrid quantum-classical machine-learning protocols for a better characterization and control of the current and future unavoidably noisy quantum devices.