Quantum Kitchen Sinks: An algorithm for machine learning on near-term quantum computers
Christopher M. Wilson, Johannes Otterbach, Nikolas Tezak, Robert Smith, Anthony Polloreno, Peter J. Karalekas, Steven Heidel, Mahabubul Alam, Gavin E. Crooks, Marcus Palmer Da Silva
Noisy intermediate-scale quantum computing devices are an exciting platform for the exploration of the power of near-term quantum applications. Performing nontrivial tasks in such devices requires a fundamentally different approach than what would be used on an error-corrected quantum computer. One such approach is to use hybrid algorithms, where problems are reduced to a parameterized quantum circuit that is often optimized in a classical feedback loop. Here we describe one such hybrid algorithm for machine learning tasks by building upon the classical algorithm known as random kitchen sinks. Our technique, called quantum kitchen sinks, uses quantum circuits to nonlinearly transform classical inputs into features that can then be used in a number of machine learning algorithms. We demonstrate the power and flexibility of this proposal by using it to solve binary classification problems for synthetic datasets as well as handwritten digits from the MNIST database. Using the Rigetti quantum virtual machine, we show that small quantum circuits provide significant performance lift over standard linear classical algorithms, reducing classification error rates from 50% to $<0.1\%$, and from $4.1\%$ to $1.4\%$ in these two examples, respectively. Further, we are able to run the MNIST classification problem, using full-sized MNIST images, on a Rigetti quantum processing unit, finding a modest performance lift over the linear baseline.
Noisy intermediate-scale quantum computing devices are an exciting platform for the exploration of the power of near-term quantum applications. Performing nontrivial tasks in such devices requires a fundamentally different approach than what would be used on an error-corrected quantum computer. One such approach is to use hybrid algorithms, where problems are reduced to a parameterized quantum circuit that is often optimized in a classical feedback loop. Here we describe one such hybrid algorithm for machine learning tasks by building upon the classical algorithm known as random kitchen sinks. Our technique, called quantum kitchen sinks, uses quantum circuits to nonlinearly transform classical inputs into features that can then be used in a number of machine learning algorithms. We demonstrate the power and flexibility of this proposal by using it to solve binary classification problems for synthetic datasets as well as handwritten digits from the MNIST database. Using the Rigetti quantum virtual machine, we show that small quantum circuits provide significant performance lift over standard linear classical algorithms, reducing classification error rates from 50% to <0.1%, and from 4.1% to 1.4% in these two examples, respectively. Further, we are able to run the MNIST classification problem, using full-sized MNIST images, on a Rigetti quantum processing unit, finding a modest performance lift over the linear baseline.
Quantum information processing is likely to have a far-reaching impact in the field of artificial intelligence. Noisy, intermediate-scale quantum devices provide a platform for exploring the possibility of attaining a quantum advantage through hybrid quantum-classical machine-learning algorithms. One example of such a hybrid algorithm is ``quantum kitchen sinks,'' which builds upon a classical algorithm known as ``random kitchen sinks'' to leverage a gate model quantum computer for machine-learning applications. We propose an alternative algorithm called ``analog-quantum kitchen sinks'' (AQKSs), which employs an analog-quantum computer for mapping data features into new features in a nonlinear manner. The new features can then be used by a classical algorithm to perform machine-learning tasks. We show the effectiveness of our algorithm for performing binary classification on both a synthetic dataset and a real-world dataset by simulating the operations of a quantum annealer. We demonstrate that the AQKS algorithm reduces the classification error of a linear classifier from $50\mathrm{%}$ to $0.6\mathrm{%}$ for the synthetic dataset and from $4.4\mathrm{%}$ to $1.6\mathrm{%}$ for the other dataset. Our proposed AQKS algorithm presents the possibility to use current quantum annealers for solving practical machine-learning problems.
Several variational quantum circuits for machine learning have been proposed in recent years, with one promising class of variational algorithms involving tensor networks operating on states resulting from local feature maps. In contrast, a random feature approach known as quantum kitchen sinks (C. M. Wilson et al., Quantum kitchen sinks: An algorithm for machine learning on near-term quantum computers, arXiv:1806.08321) provides comparable performance, but leverages nonlocal feature maps. Here we combine these two approaches by proposing a circuit Ansatz where a tree tensor network coherently processes the nonlocal feature maps of quantum kitchen sinks and we run numerical experiments to evaluate the performance of the Ansatz on image classification. From the perspective of classification performance, we find that simply combining quantum kitchen sinks with tensor networks yields no qualitative improvements. However, the addition of feature optimization greatly boosts performance, leading to state-of-the-art quantum circuits for image classification, requiring only shallow circuits and a small number of qubits, both well within reach of near-term quantum devices.
Quantum-inspired feature engineering is a proposed route to practical gains in machine learning without requiring quantum hardware. We conduct a controlled and matched evaluation of a quantum-inspired implementation of Quantum Kitchen Sinks (QKS)—originally proposed as a quantum algorithm—for network traffic classification across tasks of differing complexity using two real datasets—ISCXVPN2016 (14 classes) and MIT VNAT (5 classes)—and a binary VPN detection task. Holding LightGBM hyperparameters, training regimen, and cross-validation fixed, we compare classical baselines to models augmented with two QKS-inspired features selected in a supervised manner from 3,990 simulated circuits. On the complex 14-class ISCXVPN2016 task, this quantum-inspired implementation yields a statistically significant but modest gain; on simpler problems (MIT VNAT and binary VPN detection) it provides no measurable benefit. These findings suggest task-dependent boundary conditions: quantum-inspired features are helpful when classical performance retains headroom and are redundant near the performance ceiling. Our evaluation specifies datasets, feature engineering, model settings, fixed CV splits, and significance testing to support reproducibility, and offers actionable guidance for when QKS-inspired augmentation is likely to add value in networking applications.
Abstract Image classification, a pivotal task in multiple industries, faces computational challenges due to the burgeoning volume of visual data. This research addresses these challenges by introducing two quantum machine learning models that leverage the principles of quantum mechanics for effective computations. Our first model, a hybrid quantum neural network with parallel quantum circuits, enables the execution of computations even in the noisy intermediate-scale quantum era, where circuits with a large number of qubits are currently infeasible. This model demonstrated a record-breaking classification accuracy of 99.21% on the full MNIST dataset, surpassing the performance of known quantum–classical models, while having eight times fewer parameters than its classical counterpart. Also, the results of testing this hybrid model on a Medical MNIST (classification accuracy over 99%), and on CIFAR-10 (classification accuracy over 82%), can serve as evidence of the generalizability of the model and highlights the efficiency of quantum layers in distinguishing common features of input data. Our second model introduces a hybrid quantum neural network with a Quanvolutional layer, reducing image resolution via a convolution process. The model matches the performance of its classical counterpart, having four times fewer trainable parameters, and outperforms a classical model with equal weight parameters. These models represent advancements in quantum machine learning research and illuminate the path towards more accurate image classification systems.
Quantum computing has emerged as one of the most influential paradigms capable of transforming computational processes that struggle under classical limitations. In recent years, hybrid quantum–classical algorithms have gained significant attention for their ability to leverage the strengths of both architectures, especially within the Noisy Intermediate-Scale Quantum (NISQ) era. This research paper explores a quantum-enhanced optimization framework designed to improve machine learning model performance through parameterized quantum circuits (PQCs). The goal of this study is to provide a practical, scalable approach that can be integrated into existing machine learning workflows without requiring fully developed fault-tolerant quantum hardware. The proposed framework uses classical preprocessing to encode data into quantum states, followed by quantum transformations that explore richer feature spaces due to quantum superposition and entanglement. A classical optimizer then updates the quantum circuit parameters based on measurement results, forming an end-to-end feedback loop. This hybrid model aims to demonstrate improved optimization efficiency, reduction in loss-function plateaus, and better generalization for specific ML tasks involving small to medium-sized datasets. This paper provides a comprehensive investigation of methodology, literature background, and experimental reasoning with statistical analysis, along with conceptual diagrams illustrating system architecture and circuit layouts. The outcomes further strengthen the viability of hybrid quantum–classical computation as a promising direction for advanced machine learning systems.
Hybrid quantum-classical computing offers a promising pathway to harness quantum advantages while mitigating the hardware limitations of noisy intermediate-scale quantum (NISQ) devices. This work proposes a modular hybrid architecture that combines classical preprocessing, quantum-enhanced feature mapping through parameterized quantum circuits (PQCs), and adaptive classical optimization. The framework is designed to improve robustness and generalization of machine learning models in noisy environments, addressing critical issues such as overfitting, noise sensitivity, and hardware-induced errors. Experimental evaluations were conducted on benchmark datasets, including Iris, reduced-dimension MNIST, synthetic noisy XOR, Wine Quality, and Breast Cancer Wisconsin. The proposed model was tested under multiple noise models, such as Gaussian, salt-and-pepper, Poisson, and quantum-specific errors (bit-flip, phase-flip, and depolarizing), using 5-fold cross-validation and repeated trials for statistical rigor. Key performance metrics included Classification Accuracy (CA), Robustness Index (RI), and Training Stability (TS), with results reported as mean ± standard deviation and supported by paired t-tests (p < 0.05). Results demonstrate that the hybrid model consistently outperforms classical support vector machines (SVM) and standalone variational quantum classifiers (VQC) in noisy conditions, achieving smaller accuracy degradation and higher stability. The combination of quantum- enhanced feature mapping, optimizer synergy, and modular design is shown to be central to its resilience.