“HYBRID QUANTUM-CLASSICAL ARCHITECTURES FOR SPARSE TENSOR DECOMPOSITION IN HIGH-DIMENSIONAL DATA STREAMS WITH APPLICATIONS IN SUSTAINABLE SMART SYSTEMS”

Abstract

The past years have witnessed the generation of a high amount of high-dimensional data from various domains including smart cities, healthcare, environment, transportation systems, Industrial Internet of Things (IIoT) platforms, etc. The process of analyzing these multidimensional data streams poses a major challenge for effective analysis. A variety of classical machine learning and tensor decomposition methods are used to analyze high-dimensional sparse data. However, the practical applicability and performance of these techniques might be limited when it comes to large-scale sparse data streams because of their high computation cost… memory overhead and slow convergence rate. The current study presents a hybrid quantum-classical architecture for sparse tensor decomposition in high-dimensional streaming settings to overcome classical limitations. The framework that we proposed utilizes classical tensor factorization along with quantum optimization for efficient computation, dimensionality reduction and sparse pattern extraction. The study addresses the development of the new sustainable smart systems, where continuous application of data for real time analysis is crucial for energy optimization failure prediction resource management and intelligent decision making. The architecture utilizes quantum-inspired optimization, tensor train decomposition, and adaptive sparse learning to facilitate the processing of multifaceted datasets. The results of the investigation reveal that the hybrid model performs better than the other existing tensor decomposition techniques in terms of decomposition accuracy, reconstruction error, execution time and scalability. The research shows that quantum-classical computing can improve the performance of data analytics in future sustainable smart systems and large-scale smart infrastructures.