Deep learning has superior performance in defect detection， however， due to the low defect probability， the detection process of defect-free images occupies most of the calculation time， which seriously limits the overall effective detection speed. In order to solve the above problem， a section steel surface defect detection algorithm based on cascade network named SDNet （Select and Detect Network） was proposed. The proposed algorithm was divided into two stages： the pre-inspection stage and the precise detection stage. In the pre-inspection stage， the lightweight ResNet pre-inspection network based on Depthwise Separable Convolution （DSC） and multi-scale parallel convolution was used to determine whether there were defects in the surface image of the section steel. In the precise detection stage， the YOLOv3 was used as the baseline network to accurately classify and locate the defects in the image. In addition， the improved Atrous Spatial Pyramid Pooling （ASPP） module and dual attention module were introduced in the backbone feature extraction network and prediction branches to improve the network detection performance. Experimental results show that the detection speed and the accuracy of SDNet on 1 024 pixel×1 024 pixel images reach 120.63 frames per second and 92.1% respectively. Compared to the original YOLOv3 algorithm， the proposed algorithm has the detection speed of about 3.7 times and the detection precision improved by 10.4 percentage points. The proposed algorithm can be applied to the rapid detection of section steel surface defects.

In view of the problem of subjectivity and easy misdiagnosis in the artificial identification of Alzheimer's Disease (AD) through brain imaging, a method of automatic identification of AD by constructing brain network based on Magnetic Resonance Imaging (MRI) image was proposed. Firstly, MRI images were superimposed and were divided into structural blocks, and the Structural SIMilarity (SSIM) between any two structural blocks was calculated to construct the network. Then, the complex network theory was used to extract structural parameters, which were used as the input of machine learning algorithm to realize the AD automatic identification. The analysis found that the classification effect was optimal with two parameters, especially the node betweenness and edge betweenness were taken as the input. Further study found that the classification effect was optimal when MRI image was divided into 27 structural blocks, and the accuracy of weighted network and unweighted network was up to 91.04% and 94.51% respectively. The experimental results show that the complex network of structural similarity based on MRI block division can identify AD with higher accuracy.

To resolve the problems of feature homogeneity in unsupervised training of Restricted Boltzmann Machine (RBM) and non-adaptiveness of Sparse Restricted Boltzmann Machine (SRBM), a new sparse mechanism method of RBM based on competitive learning was designed. Firstly, a distance measurement was designed based on the cosine value between the neuron weight vector and the input vector to evaluate the similarity. Secondly, the optimal matching implicit unit based on distance measurement was selected for different samples during training. Thirdly, the sparse penalty for other hidden units was calculated according to the activation state of the optimal matching hidden unit. Finally, the parameters were updated and the competitive sparseness was applied to the construction of Deep Boltzmann Machine (DBM) based on the deep model training process. The handwritten number recognition results show that, compared with the mechanism using the sum of squared errors as the regularization factor, the classification accuracy of DBM based on new sparse mechanism is improved by 0.74%, and the average sparsity measurement is increased by 5.6%, without the need to set sparse parameters. Therefore, the proposed sparse mechanism can improve the training efficiency of unsupervised training model, such as RBM, and can be applied into the construction of deep models.

Focused on poor robustness to noise of the Visibility Graph (VG) algorithm, an improved Limited Penetrable Visibility Graph (LPVG) algorithm was proposed. LPVG algorithm could map time series into networks by connecting the points of time series which satisfy the certain conditions based on the visibility criterion and the limited penetrable distance. Firstly, the performance of LPVG algorithm was analyzed. Secondly, LPVG algorithm was combined with Power Spectrum Density (PSD) to apply to the automatic identification of epileptic ElectroEncephaloGram (EEG) before, during and after the seizure. Finally, the characteristic parameters of the LPVG network in the three states were extracted to study the influence of epilepsy seizures on the network topology. The simulation results show that compared with VG and Horizontal Visibility Graph (HVG), although LPVG had a high time complexity, it had strong robustness to noise in the signal:when mapping the typical periodic, random, fractal and chaos time series into networks by LPVG, it was found that as the noise intensity increased, the fluctuation rates of clustering coefficient by LPVG network were always the lowest, respectively 6.73%, 0.05%, 0.99% and 3.20%. By the PSD and LPVG analysis, it was found that epilepsy seizure had great influence on the brain energy. PSD was obviously enhanced in the delta frequency band, and significantly reduced in the theta frequency band; the topological structure of the LPVG network changed during the seizure, characterized by the independent enhanced network module, increased average path length and decreased graph index complexity. The PSD and LPVG applied in this paper could be taken as an effective measure to characterize the abnormality of the energy distribution and topological structure of single EEG signal channel, which would provide help for the pathological study and clinical diagnosis of epilepsy.

Source code comments is an important part of the software, so researchers need to use manual or automated methods to generate comments. In the past, the quality assessment of this kind of comments is done manually, which is inefficient and not objective. In order to solve this problem, an assessment criterion was built in which four aspects of the comments including comment format, language form, content and code-related degree were considered. Then a code comments quality assessment method based on an aggregation of classification algorithms was proposed, in which machine learning and natural language processing technology were introduced into comments quality assessment, by using classification algorithms the comments were classified into four levels, including unqualified, qualified, good and excellent ones. The evaluation results were improved by the aggregation of the basic classification algorithms. The precision and F1 measure of the aggregated classification algorithm were improved about 20 percentage points compared with using a single classification algorithm, and all the indexes have reached more than 70% except the macro average F1 measure. The experimental results show that this method can be applied to assess the quality of comments effectively.

As the instantaneous Channel State Information (CSI) of source to relay and relay to destination affects the overall Bit Error Rate (BER) of the cooperative communication system, a relay selection scheme which evaluated the two-stage channel coefficients was proposed. Firstly, the channel coefficients of source-relay channel and the channel coefficients of relay-destination channel were compared according to the CSI of each candidate relay, and the worse one was found out. Moreover, a node set containing the approximate optimal relays was obtained by sorting the candidate relays based on their worse channel coefficients. Finally, the relay with the highest summation of the two-stage channel coefficients in the set was selected as the one which participated in the cooperative transmission. The simulation results reveal that the Signal-to-Noise Ratio (SNR) of the proposed relay selection scheme respectively decreases by 0.4 dB and 0.2 dB compared with the best worse channel selection scheme and the delay selection scheme based on the nearest neighbor relation, when the number of candidate relay nodes is 100 and 5, and the BER decreases to 10^-4 and 10^-5. In general, the proposed scheme can expend the information transmission range and improve the reliability in the wireless relay network.

In order to meet the high requirements of real-time collision detection in increasingly complex virtual environment, a fast collision detection algorithm based on image space was proposed. It made efficiently use of the Graphics Processing Unit (GPU). Based on the hierarchical binary tree and the collision detection between Oriented Bounding Boxes (OBB), the algorithm could quickly eliminate disjoint bumps of the virtual scene. With the potential collision set, the efficiency of the algorithm has a significantly improvement on the basis of RECODE algorithm. The experimental results show that the algorithm achieves good results, and has a higher efficiency, especially in a highly complex virtual environment.

This paper proposed a solution to do research on real-time data stream analyzing and entire process quality tracing based on PI (Plant information) in order to solve these problems that the production information was blocked and product quality was unable to be traced in the steel production. The proposed solution focused on real-time data stream partition and process monitoring, and presented statistical monitoring methods based on Statistical Quality Control (SQC) charts and process capability indices. Furthermore, a product technique and quality monitoring system was developed. The applied results indicate the implementation of real-time data stream analysis and product quality monitoring based on PI can efficiently monitor production process quality, the identification and improvement of key production technology as well.

The interpersonal contacts of the Social Network Service (SNS) customers are often researched based on the information of the graph theory. The preference of the customers themselves is often ignored, when discussing the nodes and the edges of the relationship graph of SNS. Thus, a second-level interpersonal contacts method based on the subjects of users' preference was proposed in this paper. Utilizing text mining technology and the Least Mean Square (LMS) algorithm, the authors transformed the subjects of users' preference into feature vectors reasonably. In order to ensure the set of the customers similar to the subjects of users' preference and complete the second-level recommendation of the customers, the similarity of the customers was computed with the similarity measurement. The experimental results show that the recommendation accuracy for good friend of this algorithm is very high. The acceptance rate of the recommended good friends is 70%.

During the restoration of actual motion blurred images based on Wiener filtering, restoration results get affected by serious ringing effect and unsatisfactory local restoration. Its main reasons were found through theoretical analysis, experimental comparisons and the study of the characteristics of the actual motion blurring process. It was proposed that the artificial boundary compensation and block-restoration were used to restrain ringing effect and local unsatisfactory restoration. The relations of blur parameters, space positions and speeds, even the standard of blocking partition were given. The experimental results verify that the proposed method of the boundary compensation and the block-restoration can effectively reduce ringing effect and maintain the consistency of the overall image restoration effect.

The theory of IPID(IP Identification) covert network scanning was introduced,then the design and implementation of IPID covert network scanning under the operation Windows was proposed.In order to improve the efficiency of scanning, the chunk binary algorithm was introduced, then its performance was analyzed and was compared with other algorithm. The results show that the chunk binary algorithm is a good algorithm of IPID covert scanning,and the correctness and speed of IPID covert scanning is decided by setting delay time between getting two IPID.