Hybrid Quantum–Classical Framework for Urban Traffic State Classification

This paper presents a hybrid quantum–classical approach for multiclass traffic state classification using a benchmark urban mobility dataset collected in Daejeon, South Korea. The proposed architecture, called the Multilayer Perceptron–Variational Quantum Classifier (MVQC), combines classical feature extraction via a multilayer perceptron (MLP) with quantum-based classification using a variational quantum classifier (VQC). On the medium-scale subset of the dataset, the MVQC model achieved a test accuracy of 91.4%, outperforming the standalone VQC model, which achieved 61.8%. Although MVQC did not reach the accuracy of the classical MLP model (98.6%), it consistently outperformed purely quantum models and demonstrated better scalability with increasing data complexity. All quantum operations were simulated under noise-free conditions using IBM Qiskit’s statevector simulator. These results demonstrate that classical feature extraction can effectively guide quantum circuits, enhance learning performance, and highlight the hybrid MVQC architecture as a viable solution for time-series classification in intelligent transportation systems.