Automated language detection system using Raussendorf lattice pattern features with EEG signals

Abstract

This study introduces a quantum-inspired and self-organized model for EEG-based language detection. The Raussendorf Lattice Pattern (RLP) is proposed as a graph-based feature extractor inspired by quantum topology. It defines 15 energy-based patterns that adapt to signal dynamics and generate explainable texture features. A five-level Twin Wavelet Transform produces 18 wavelet bands for multilevel frequency-domain features. Statistical and RLP features are fused into a single vector. Iterative NCA selects the most discriminative features, while kNN and SVM perform channel-wise classification. Iterative Majority Voting fuses outputs for optimal accuracy. The model achieves 99.54% (tenfold CV) and 92.84% (LOSO) accuracies. Semantic cortical maps show dominant frontal activation near Broca's area. The results confirm that quantum-inspired self-organized feature extraction offers efficient and explainable solutions for EEG-based inner-speech and language detection.