NN2QC: Systematic Transfer of Classical Neural Networks to Variational Quantum Circuits

We present NN2QC (Neural Network to Quantum Circuit), a systematic framework for transferring trained classical neural networks to variational quantum circuits. Through rigorous 3-fold cross-validation across four synthetic benchmarks and one real-world medical dataset (Breast Cancer Wisconsin, 569 samples), we evaluate the performance gap between classical and quantum implementations. While classical neural networks achieve 93.2% accuracy on linearly separable problems, quantum circuits demonstrate competitive 78.6% performance on real medical diagnosis tasks, significantly outperforming their 54-71% accuracy on synthetic data. We identify angle encoding as superior to amplitude encoding (70.8% vs 49.2%, p < 0.001) and find that circuit depth impacts performance more than qubit count (+39.5% vs +1.7%). Our empirical findings provide guidance for practitioners considering quantum neural network implementations and identify problem regimes where quantum approaches show promise.