SOFTWARE METHODS AND ARCHITECTURAL SOLUTIONS FOR BIG DATA PROCESSING VIA QUANTUM NEURAL NETWORKS AND METAHEURISTIC OPTIMIZERS

Abstract

The article presents proposed software methods and a hybrid classical–quantum architecture for big data processing using a quantum kernel method QSVC and a variational quantum classifier VQC/QNN to address the problem of computing mental energy from EEG signals. The proposed framework operates on data acquired from a wearable EEG device that records δ, θ, α, and β bands across four channels (the 4-channel Muse-S Athena EEG device). From these signals, feature vectors are constructed and mapped into quantum states via a quantum feature map, which are then used either in a quantum kernel SVM or in a variational quantum neural network for classification or regression of mental energy–related indicators. To optimize the resulting quantum circuits (“quantum neural networks”), the work employs metaheuristic algorithms such as Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Grey Wolf Optimizer (GWO), and multi-objective variants like NSGA. These optimizers jointly tune circuit parameters, feature-encoding scales and elements of the circuit structure under realistic NISQ constraints. The main task is to reduce an inherently multi- objective problem to a tractable single-objective formulation by defining a cost function that minimizes a weighted sum of key parameters: the number of measurement shots on the quantum computer, the depth of the quantum circuit, and the total execution time, while preserving an acceptable classification quality of mental states. The paper also proposes a software architecture based on quantum microservices (for quantum kernel evaluation and VQC execution) combined with a metaheuristic optimization orchestrator and resource-tracking components. This architecture is designed to integrate with EEG-driven mental energy models, enabling scalable experimentation with quantum models on large EEG datasets. The results obtained provide a methodological basis for using quantum neural networks and metaheuristic optimizers in cognitive and neuroengineering applications, with a particular focus on quantifying mental energy in real- world conditions.