This paper presents a comparative analysis of two quantum machine learning models based on amplitude encoding Quantum Support Vector Regression (QSVR) and Quantum Neural Network (QNN) applied to the problem of delay (demand/replenishment lag) prediction in inventory management systems. The study uses a dataset comprising 52 months of sales history for 51 product categories. Classical feature vectors are mapped into quantum states via amplitude encoding, which embeds a d-dimensional vector into$\mathbf{n}=\left\lceil\log _{2} \mathbf{d}\right\rceil$qubits with exponential space efficiency. The QSVR model employs a quantum kernel while the QNN model uses a data re-uploading variational circuit (4 qubits, 2 layers, R_Y rotation gates with CNOT entangling gates) trained with the COBYLA optimizer (200 iterations). Experimental results show that the QNN model achieved a slightly higher coefficient of determination$\left(\mathrm{R}^{2} \text{QNN}=0.8371\right.$vs.$\left.\mathrm{R}^{2} \text{QSVR}=0.8185\right)$, whereas the QSVR model achieved a lower mean absolute error (MAE=0.4145 vs. 0.4630 for QNN), while in terms of RMSE the QNN model outperformed QSVR (0.4916 vs. 0.7270), indicating fewer large prediction errors. All experiments were simulated in Python using the IBM Qiskit quantum computing framework. The findings suggest that quantum machine learning models are a promising direction for practical inventory replenishment and delay-forecasting tasks.
This paper presents a comparative analysis of two quantum machine learning models based on amplitude encoding Quantum Support Vector Regression (QSVR) and Quantum Neural Network (QNN) applied to the problem of delay (demand/replenishment lag) prediction in inventory management systems. The study uses a dataset comprising 52 months of sales history for 51 product categories. Classical feature vectors are mapped into quantum states via amplitude encoding, which embeds a d-dimensional vector inton=⌈log2d⌉qubits with exponential space efficiency. The QSVR model employs a quantum kernel while the QNN model uses a data re-uploading variational circuit (4 qubits, 2 layers, R_Y rotation gates with CNOT entangling gates) trained with the COBYLA optimizer (200 iterations). Experimental results show that the QNN model achieved a slightly higher coefficient of determination(R2QNN=0.8371vs.R2QSVR=0.8185), whereas the QSVR model achieved a lower mean absolute error (MAE=0.4145 vs. 0.4630 for QNN), while in terms of RMSE the QNN model outperformed QSVR (0.4916 vs. 0.7270), indicating fewer large prediction errors. All experiments were simulated in Python using the IBM Qiskit quantum computing framework. The findings suggest that quantum machine learning models are a promising direction for practical inventory replenishment and delay-forecasting tasks.
This article analyzes the issue of accurately predicting the price and sales volume of retail products based on classical and quantum approaches. The classical Support Vector Regression (SVR) model was evaluated in comparison with the Quantum Support Vector Regression (QSVR) model based on quantum computing. Latent representations of the data were obtained through dimensionality reduction using an autoencoder and utilized for forecasting. The experimental results demonstrated that the QSVR model outperforms the classical SVR model in terms of MAE, MSE, and R2 indicators. The study effectively employed quantum kernel functions, data re-uploading feature maps, and the COBYLA optimization algorithm. The findings confirm the practical potential of quantum machine learning methods.
Abstract Quantum machine learning is a rapidly growing field at the intersection of quantum computing and machine learning. In this work, we examine our quantum machine learning models, which are based on quantum support vector classification (QSVC) and quantum support vector regression (QSVR). We investigate these models using a quantum circuit simulator, both with and without noise, as well as the IonQ Harmony quantum processor. For the QSVC tasks, we use a dataset containing fraudulent credit card transactions and image datasets (the MNIST and the Fashion-MNIST datasets); for the QSVR tasks, we use a financial dataset and a materials dataset. For the classification tasks, the performance of our QSVC models using 4 qubits of the trapped-ion quantum computer was comparable to that obtained from noiseless quantum circuit simulations. The result is consistent with the analysis of our device noise simulations with varying qubit gate error rates. For the regression tasks, applying a low-rank approximation to the noisy quantum kernel, in combination with hyperparameter tuning in ε-SVR, improved the performance of the QSVR models on the near-term quantum device. The alignment, as measured by the Frobenius inner product between the noiseless and noisy quantum kernels, can serve as an indicator of the relative prediction performance on noisy quantum devices in comparison with their ideal counterparts. Our results suggest that the quantum kernel, as described by our shallow quantum circuit, can be effectively used for both QSVC and QSVR tasks, indicating its resistance to noise and its adaptability to various datasets.
This study explores the performance of Quantum Support Vector Classifiers (QSVCs) and Quantum Neural Networks (QNNs) in comparison to classical models for machine learning tasks. By evaluating these models on the Iris and MNIST-PCA datasets, we find that quantum models tend to outperform classical approaches as the problem complexity increases. While QSVCs generally provide more consistent results, QNNs exhibit superior performance in higher-complexity tasks due to their increased quantum load. Additionally, we analyze the impact of hyperparameter tuning, showing that feature maps and ansatz configurations significantly influence model accuracy. We also compare the PennyLane and Qiskit frameworks, concluding that Qiskit provides better optimization and efficiency for our implementation. These findings highlight the potential of Quantum Machine Learning (QML) for complex classification problems and provide insights into model selection and optimization strategies
Support vector machines are powerful and frequently used machine learning methods for classification and regression tasks, which rely on the construction of kernel matrices. While crucial for the performance of this machine learning approach, choosing the most suitable kernel is highly problem-dependent. The emergence of quantum computers and quantum machine learning techniques provides new possibilities for generating powerful quantum kernels. Within this work, we solve a real-world price forecasting problem using fidelity and projected quantum kernels, which are promising candidates for the utility of near-term quantum computing. In our analysis, we examine and validate the most auspicious quantum kernels from literature and compare their performance with an optimized classical kernel. Unlike previous work on quantum support vector machines, our dataset includes categorical features that need to be encoded as numerical features, which we realize by using the one-hot-encoding scheme. One-hot-encoding, however, increases the dimensionality of the dataset significantly, which collides with the current limitations of noisy intermediate scale quantum computers. To overcome these limitations, we use autoencoders to learn a low-dimensional representation of the feature space that still maintains the most important information of the original data. To examine the impact of autoencoding, we compare the results of the encoded date with the results of the original, unencoded dataset. We could demonstrate that quantum kernels are comparable to or even better than the classical support vector machine kernels regarding the mean absolute percentage error scores for both encoded and unencoded datasets.
Financial markets, particularly equity indices like the SP 500, exhibit complex, nonlinear, and frequently chaotic behaviors driven by a combination of macroeconomic, geopolitical, and psychological factors. Traditional machine learning techniques, despite their notable success in modeling market trends, often encounter performance limitations when dealing with high-dimensional feature spaces and intricate historical dependencies. Quantum computing offers a promising approach for addressing these challenges through its ability to represent and process information in superposition states. This study proposes and implements a quantum neural network (QNN) architecture that utilizes angle encoding to embed carefully engineered features, including volatility measures, momentum indicators, and lagged price signals, into quantum states. These encoded features are processed through a customized quantum circuit comprising three variational layers that combine linear entanglement patterns with single-qubit rotation operations. The training process is conducted via gradient-based optimization through the PennyLane framework, operating within a noiseless simulation environment to establish theoretical performance bounds. The experimental results demonstrate that the QNN achieves an accuracy of 81.3% on the held-out test set in simulation, outperforming traditional baseline models including Long Short-Term Memory networks, Random Forest classifiers, and Support Vector Machines under identical conditions. The model exhibits balanced precision (0.83) and recall (0.79). While the experiments were conducted in simulation due to current quantum hardware limitations, the findings illustrate the potential of quantum-enhanced algorithms for financial forecasting applications. Keywords Quantum Neural Networks, Financial Forecasting, SP 500, Variational Quantum Circuits, Time Series Prediction.
Recovery rate prediction plays a pivotal role in bond investment strategies, enhancing risk assessment, optimizing portfolio allocation, improving pricing accuracy, and supporting effective credit risk management. However, forecasting faces challenges like high-dimensional features, small sample sizes, and overfitting. We propose a hybrid Quantum Machine Learning model incorporating Parameterized Quantum Circuits (PQC) within a neural network framework. PQCs inherently preserve unitarity, avoiding computationally costly orthogonality constraints, while amplitude encoding enables exponential data compression, reducing qubit requirements logarithmically. Applied to a global dataset of 1,725 observations (1996–2023), our method achieved superior accuracy (RMSE 0.228) compared to classical neural networks (0.246) and quantum models with angle encoding (0.242), with efficient computation times. This work highlights the potential of hybrid quantum-classical architectures in advancing recovery rate forecasting.