230
Y. Zhou et al.
Decision tree is prone to overfitting problems and has poor performance in the
face of complex continuous features. These traditional machine learning methods
mostly have weak generalization ability and limited ability to express complex
functions, so they cannot deal with complex classification problems well.
In recent years, deep learning method has been widely used in face recognition, speech recognition, image recognition, and other fields, which make it
become a hot topic in machine learning. At the same time, deep learning also
has a good application in network security intrusion detection. By comparing
the intrusion detection technology based on deep learning with various traditional intrusion detection technologies, it is found that the intrusion detection
technology based on deep learning achieved better results in terms of accuracy
and false positive rate.
In this paper, an intrusion detection algorithm based on TensorFlow is proposed by using CNN, combining with mainstream deep learning techniques such
as dropout, Adam, and Softmax classifier. And KDDCUP’99 data set is used to
verify the feasibility of the proposed algorithm. This paper compares different
activation functions used in the proposed CNN model and then designs a fivecategory test experiment to test the performance of the proposed method for
intrusion detection.
The paper is organized as follows: Sect. 2 presents an intrusion detection
algorithm based on deep convolutional neural network. The experimental results
and analysis are discussed in Sect. 3. Finally, the paper is concluded with a
discussion of future work in Sect. 4.
2 Implementation of the Detection
Deep learning can automatically learn useful features from raw data to improve
classification accuracy. The basic architecture of machine learning is the combination of feature extraction and classification modules. The proposed architecture of deep learning intrusion detection method based on CNN is shown in
Fig. 1. It mainly consists of three modules: data preprocessing module, CNN
modeling module, and classifier module.
Raw Data
Data
Preprocessing
CNN Modeling
Classifier
0,2,2,181,5450…… 0,1,0
0,2,2,181,5450…… 0,1,0
Raw Data
Data
Preprocessing
CNN Modeling
Classifier
0,2,2,181,5450 0,1,0
0,2,2,181,5450 0,1,0
Fig. 1. Intrusion detection architecture
Y. Zhou et al.
Decision tree is prone to overfitting problems and has poor performance in the
face of complex continuous features. These traditional machine learning methods
mostly have weak generalization ability and limited ability to express complex
functions, so they cannot deal with complex classification problems well.
In recent years, deep learning method has been widely used in face recognition, speech recognition, image recognition, and other fields, which make it
become a hot topic in machine learning. At the same time, deep learning also
has a good application in network security intrusion detection. By comparing
the intrusion detection technology based on deep learning with various traditional intrusion detection technologies, it is found that the intrusion detection
technology based on deep learning achieved better results in terms of accuracy
and false positive rate.
In this paper, an intrusion detection algorithm based on TensorFlow is proposed by using CNN, combining with mainstream deep learning techniques such
as dropout, Adam, and Softmax classifier. And KDDCUP’99 data set is used to
verify the feasibility of the proposed algorithm. This paper compares different
activation functions used in the proposed CNN model and then designs a fivecategory test experiment to test the performance of the proposed method for
intrusion detection.
The paper is organized as follows: Sect. 2 presents an intrusion detection
algorithm based on deep convolutional neural network. The experimental results
and analysis are discussed in Sect. 3. Finally, the paper is concluded with a
discussion of future work in Sect. 4.
2 Implementation of the Detection
Deep learning can automatically learn useful features from raw data to improve
classification accuracy. The basic architecture of machine learning is the combination of feature extraction and classification modules. The proposed architecture of deep learning intrusion detection method based on CNN is shown in
Fig. 1. It mainly consists of three modules: data preprocessing module, CNN
modeling module, and classifier module.
Raw Data
Data
Preprocessing
CNN Modeling
Classifier
0,2,2,181,5450…… 0,1,0
0,2,2,181,5450…… 0,1,0
Raw Data
Data
Preprocessing
CNN Modeling
Classifier
0,2,2,181,5450 0,1,0
0,2,2,181,5450 0,1,0
Fig. 1. Intrusion detection architecture