Vol. 2, Issue 1, Part A (2025)

The exponential growth of high-dimensional data in fields such as genomics, finance, and remote sensing has exposed the limitations of conventional machine learning models in terms of scalability, interpretability, and computational efficiency. This study presents a novel Quantum-Inspired Machine Learning (QIML) framework that integrates amplitude-based feature encoding, tensor network compression, and low-rank sketching to emulate quantum computational principles using classical resources. By drawing inspiration from quantum linear algebra and state-space representation, the proposed framework addresses the “curse of dimensionality” and reduces computational overhead while maintaining predictive accuracy.
The experimental design employed four diverse high-dimensional datasets—ImageHD, GenomicsHD, FinanceHD, and TextHD—to evaluate the model’s performance against established baselines including Support Vector Machines (SVM), PCA+SVM, and Deep Neural Networks (DNN). Statistical analyses using cross-validation and paired t-tests revealed that QIML consistently achieved comparable or superior accuracy to DNNs while reducing average runtime and memory usage by 30-40%. The results demonstrate that quantum-inspired kernel transformations effectively capture nonlinear dependencies and feature entanglement, enhancing model generalization without the need for quantum hardware. Moreover, the integration of tensor networks enabled compact data representation and interpretability, offering a transparent alternative to black-box deep learning architectures.
The study concludes that QIML provides a scalable, resource-efficient, and theoretically grounded approach for high-dimensional data analysis. Beyond advancing computational performance, it offers practical applicability for real-time analytics in resource-constrained environments. The findings pave the way for hybrid quantum-classical learning systems capable of harnessing quantum principles for practical machine intelligence. This research contributes a foundational step toward realizing quantum efficiency within classical computing frameworks and establishing QIML as a transformative paradigm in modern data science.