Intrusion detection system (IDS) plays an important role in network security by discovering and preventing malicious activities. Due to the complex and time-varying network environment, the network intrusion samples are submerged into a large number of normal samples, which leads to insufficient samples for model training and detection results with a high false detection rate. According to the problem of data imbalance, we propose a network intrusion detection algorithm combined hybrid sampling with deep hierarchical network. Firstly, we use the one-side selection (OSS) to reduce the noise samples in majority category, and then increase the minority samples by Synthetic Minority Over-sampling Technique (SMOTE). In this way, a balanced dataset can be established to make the model fully learn the features of minority samples and greatly reduce the model training time. Secondly, we use convolution neural network (CNN) to extract spatial features and Bi-directional long short-term memory (BiLSTM) to extract temporal features, which forms a deep hierarchical network model. The proposed network intrusion detection algorithm was verified by experiments on the NSL-KDD and UNSW-NB15 dataset, and the classification accuracy can achieve 83.58% and 77.16%, respectively. INDEX TERMS Network intrusion detection, hybrid sampling, deep hierarchical network, convolution neural network, bi-directional long short-term memory.

Retinal vessel segmentation plays a vital role in computer-aided diagnosis and treatment of retinal diseases. Considering the low contrast between retinal vessels and the background image, complex structural information as well as blurred boundaries between tissue and blood vessels, the retinal vessel image segmentation algorithm based on the improved U-Net network is proposed in the paper. The algorithm introduces an attention mechanism and densely connected network into the original U-Net network and realizes the automatic segmentation of retinal vessels. According to the test results of the algorithm on commonly-used datasets of the DRIVE and STARE fundus images, respectively, the accuracy is 0.9663 and 0.9684; the sensitivity is 0.8075 and 0.8437; the specificity is 0.9814 and 0.9762; the AUC values are 0.9846 and 0.9765; and the F-measures are 0.8203 and 0.8419, respectively. In the paper, the Attention-Dense-UNet (AD-UNet) algorithm is applied to segment human bulbar conjunctival micro-vessels. The experimental results show that the algorithm can achieve ideal segmentation results.

The modulation recognition of digital signals under non-cooperative conditions is one of the important research contents here. With the rapid development of artificial intelligence technology, deep learning theory is also increasingly being applied to the field of modulation recognition. In this paper, a novel digital signal modulation recognition algorithm is proposed, which has combined the InceptionResNetV2 network with transfer adaptation, called InceptionResnetV2-TA. Firstly, the received signal is preprocessed and generated the constellation diagram. Then, the constellation diagram is used as the input of the InceptionResNetV2 network to identify different kinds of signals. Transfer adaptation is used for feature extraction and SVM classifier is used to identify the modulation mode of digital signal. The constellation diagram of three typical signals, including Binary Phase Shift Keying(BPSK), Quadrature Phase Shift Keying(QPSK) and 8 Phase Shift Keying(8PSK), was made for the experiments. When the signal-to-noise ratio(SNR) is 4dB, the recognition rates of BPSK, QPSK and 8PSK are respectively 1.0, 0.9966 and 0.9633 obtained by InceptionResnetV2-TA, and at the same time, the recognition rate can be 3% higher than other algorithms. Compared with the traditional modulation recognition algorithms, the experimental results show that the proposed algorithm in this paper has a higher accuracy rate for digital signal modulation recognition at low SNR.

No abstract

LiDAR data contain feature information such as the height and shape of the ground target and play an important role for land classification. The effect of convolutional neural network (CNN) for feature extraction on LiDAR data is very significant, however CNN cannot resolve the spatial relationship of features adequately. The capsule network (CapsNet) can identify the spatial variations of features and is widely used in supervised learning. In this article, the CapsNet is combined with the residual network (ResNet) to design a deep network-ResCapNet for improving the accuracy of LiDAR classification. The capsule network represents the features by vectors, which can account for the direction of the features and the relative position between the features. Therefore, more detailed feature information can be extracted. ResNet protects the integrity of information by passing input information to the output directly, which can solve the problem of network degradation caused by information loss in the traditional CNN propagation process to a certain extent. Two different LiDAR data sets and several classic machine learning algorithms are used for comparative experiments. The experimental results show that ResCapNet proposed in this article `improve the performance of LiDAR classification.