MLNN: A Novel Network Intrusion Detection Based on Multilayer Neural Network
Chia-Fen Hsieh, Che-Min Su · 2021
The rapid development of network technology and related services has led to an increase in data traffic. Although some researches use machine learning (ML)-based intrusion detection schemes to detect intrusion. For network attacks, the changes in network traffic may lead to lower accuracy of machine learning-based models. It focuses on feature values ineffective learning materials, machine learning, and human learning similarly, and classifying data to analyze understanding, and take actions. The neural networks in deep learning used artificial neural network (ANNs) that imitate the functions of the human brain. Deep learning is a type of machine learning. The difference lies in inexperienced. In this paper, we proposed an intrusion detection architecture that based on a multi-layer neural network (MLNN). It processes data traffic and build a reliable intrusion detection model based on deep learning (DL). Compared with other machine learning or algorithms, Deep learning has the function of automatically extracting features and uses TensorFlow to execute Keras to analyze data. Through Keras, an open-source neural network library, intrusion detection targets can be achieved in a faster and more effective way. The main contribution of this paper includes considering various factors to evaluate and select, and let the integrated method perform intrusion detection.