Performance, Scalability, and Computational Resource Consumption Analysis of QNN and QSVM Models in Ideal and Noisy Quantum Simulation Environments ?Ideal ve Gürültülü Kuantum Simülasyon Ortamlarinda QNN ve QSVM Modellerinin Performans, Ölçeklenebilirlik ve Hesaplamali Kaynak Tüketimi Analizi
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11636942
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Quantum Machine Learning, Quantum Neural Networks, Quantum Support Vector Machines
- Marmara Üniversitesi Adresli: Evet
Özet
The performance of quantum machine learning algorithms is subject to hardware noise and scalability constraints. In this study, Quantum Neural Networks (QNN) and Quantum Support Vector Machines (QSVM) were comparatively analysed under ideal and IBM noise models. Accuracy, F1-score, CPU/RAM consumption and iteration times were evaluated in 2, 4 and 6-qubit systems. In the ideal simulation, it was observed that an increasing number of qubits enhances the model's expressive power. Under noisy conditions, decoherence effects generally reduce performance, whilst it was found that noise exerts a regularising effect under specific hyperparameter settings. Furthermore, it was determined that the QNN optimisation process consumes more classical hardware resources compared to the core computations of QSVM. The research empirically establishes the noise tolerance limits of the models and defines the hardware requirements for Noisy Intermediate-Scale Quantum (NISQ) applications.