Enhanced Quantum-Inspired Supervised Learning for Telecommunication Networks
Subhadeep Mondal, M Saravanan
Quantum-Inspired Supervised Learning (QISL) leverages principles from quantum mechanics to enhance classical machine learning models, offering a promising pathway to address computational bottlenecks without the need for quantum hardware. By exploiting the high-dimensional expressibility of tensor network structures and efficient optimization techniques, QISL offers reduced parameter complexity and improved scalability compared to conventional deep learning approaches. In this paper, we investigate the application of QISL to key telecommunication network challenges, specifically focusing on throughput prediction and handover optimization. We propose a hybrid architecture that integrates QISL-based tensor representations with neural networks, employing techniques such as Adaptive Cross Approximation (ACA)-based tensor decomposition for memory efficiency and optimized loss functions for stable convergence. Through extensive numerical simulations, we demonstrate that QISL significantly reduces parameter overhead while maintaining competitive or superior prediction performance, highlighting its potential as a scalable and resource-efficient alternative for 5G and Future 6G communication systems.
Computing, Communications and IoT Applications (ComComAp)•Dec 14, 2025•0 citations