Hybrid Continuous Variational Quantum Neural Networks for Network Intrusion Detection

In order to solve the problem of excessive parameters and slow computing speed classical neural networks, this paper uses powerful parallel power quantum improve network models, proposes a hybrid continuous variational (HCVQNN) model that can be used for intrusion detection. The variable layer is realized through Gaussian gate non-Gaussian gate, in which performs operations such as state weight addition offset term setting, nonlinear operations, thereby improving overall expression ability model. addition, aiming at unbalanced UNSW-NB15 dataset, design algorithm from feature level level, using ReliefF selection oversampling undersampling processing by combining Borderline-SMOTE GMM algorithm. classification experimental results on dataset show compared with other two HCVQNN obtains higher accuracy lower loss function value both binary multi-classification tasks, minority categories also improved.