Neural Structure Adaptive Quantum Neural Networks

Quantum Neural Networks (QNNs), as a pivotal tool in quantum machine learning, have demonstrated significant potential in tackling complex problems. However, QNNs still face numerous challenges in practical applications, particularly in terms of quantum circuit complexity and parameter optimization. To address these issues, this paper proposes a Neural Structure-Adaptive Quantum Neural Network (NSA-QNN) model based on sparse regularization optimization. The model optimizes parameters via L1 and group sparsity regularization, enhancing training efficiency and generalization. Meanwhile, this paper also incorporates a system adaptive strategy so that the quantum circuit can dynamically adjust the number of network layers according to the input data, which further improves the flexibility and performance of the model. The experimental results show that the number of parameters of the NSA-QNN model is reduced by up to 21.43%, while the accuracy reaches 98.65% compared with the existing state-of-the-art models. A new method is provided for the optimization of quantum neural networks.