Towards Artificial Neural Network Based Intrusion Detection with Enhanced Hyperparameter Tuning

Due to the development of complex communication paradigms and the rise in the number of inter-connected digital
devices, intrusion detection system (IDS) has become one basic and important security mechanism to identify cyber intrusions
and protect computer networks. Currently, various deep learning algorithms have been studied in intrusion detection to achieve
a high detection rate, whereas the detection performance may be still dependent on specific datasets. To maintain the detection
performance, parameter optimization is believed as an effective solution. Motivated by this observation, in this work, we propose a
concise but effective hyperparameter tuning process to enhance the artificial neural network (ANN) based IDS. In the evaluation, we
consider three ANN variants and four datasets. The experimental results indicate that our approach can outperform similar studies
and typical learning algorithms.