It is an important task to carry out reasoning and calculation on graph structure data. The main challenge of this task is how to represent graph-structured knowledge so that machines can easily understand and use graph structure data. After comparing the existing representation learning models， it is found that the models based on random walk methods are likely to ignore the special effect of attributes on the association between nodes. Therefore， a hybrid random walk method based on node adjacency and attribute association was proposed. Firstly the attribute weights were calculated through the common attribute distribution among adjacent nodes， and the sampling probability from the node to each attribute was obtained. Then， the network information was extracted from adjacent nodes and non-adjacent nodes with common attributes respectively. Finally， the network representation learning model based on node attribute bipartite graph was constructed， and the node vector representations were obtained through the above sampling sequence learning. Experimental results on Flickr， BlogCatalog and Cora public datasets show that the Micro-F1 average accuracy of node classification by the node vector representations obtained by the proposed model is 89.38%， which is 2.02 percentage points higher than that of GraphRNA （Graph Recurrent Networks with Attributed random walk） and 21.12 percentage points higher than that of classical work DeepWalk. At the same time， by comparing different random walk methods， it is found that increasing the sampling probabilities of attributes that promote node association can improve the information contained in the sampling sequence.

Aiming at the problem of influence of the channel on the characteristics of the vowel formant， a systematic experiment was carried out. Firstly， the standard recordings of 8 volunteers were collected. Then， the standard recordings were played with the mouth simulator， and 104 channel recordings were recorded using 13 different channels. Finally， the characteristic voice segments were extracted， and chi-square test analysis was used in the qualitative analysis of the spectral characteristics， and one-sample t-test was used in the quantitative analysis of acoustic parameters. The statistical results show that about 69% of the channels have a significant influence on the overall form of the high-order formants， and about 85% of the channels have significant differences in the relative intensity of the formants. The one-sample t-test results show that there is no significant difference between the standard recordings and the channel recordings in center frequency of the formant. Experimental results show that the frequency characteristics of formants should be paid more attention to when processing the identification of voices in different channels.

The evaluation of correspondences was introduced to the 3D reconstruction form binoview, which was defined as match measure function and gave a guideline to use the correspondences for estimating the Fundamental Matrix (FM). The works of features matching, FM computing and 3D reconstruction became one union process. The detecting and matching of features were analyzed firstly; then two match measure functions were defined according to the correlating and feature-distance-comparing matching methods. The new linear algorithm computed FM from the correspondences weighted by its match measure which was based on the normalized 8-points algorithm, meanwhile the RANdom SAmple Consensus （RANSAC） algorithm was employed to overcome the outliers. With the internal camera parameters, the camera relative motion was computed from FM. An optimal algorithm was employed to get the accurate result At last, reconstruction was realized through triangulation. Experimental results with simulative and real images show that the algorithm works robustly with high accuracy.