Since the early Mild Cognitive Impairment (MCI) is very likely to be undiagnosed by the assessment of medical diagnostic cognitive scale, a multimodal network fusion method for the aided diagnosis and classification of MCI was proposed. The complex network analysis method based on graph theory has been widely used in the field of neuroimaging, but different effects of brain diseases on the network topology of the brain would be conducted by using imaging technologies based different modals. Firstly, the Diffusion Tensor Imaging (DTI) and resting-state functional Magnetic Resonance Imaging (rs-fMRI) data were used to construct the fusion network of brain function and structure connection. Then, the topological properties of the fusion network were analyzed by One-way ANalysis of VAriance (ANOVA), and the attributes with significant difference were selected as the classification features. Finally, the one way cross validation of Support Vector Machines (SVM) was used for the classification of healthy group and MCI group, and the accuracy was estimated. The experimental results show that, the classification result accuracy of the proposed method reaches 94.44%, which is significantly higher than that of single modal data method. Many brain regions, such as cingulate gyrus, superior temporal gyrus and parts of the frontal and parietal lobes, of the MCI patients diagnosed by the proposed method show significant differences, which is basically consistent with the existing research results.

Concerning the time-consuming and computational complexity in extracting face features of traditional Gabor filters, the face features were extracted by using three different local Gabor filters adaptively chosen by the Gabor images' energy from different directions, scales and overall situation. Firstly, the Gabor features of some images in the face database were extracted and analyzed, and the local Gabor filters were built by choosing the filters corresponding to the images with larger energy. And then, the Fisher features were extracted using Linear Discriminate Analysis (LDA) further. Finally, face recognition was realized using the nearest neighbor method. The experimental results based on ORL and YALE face database show that the proposed approach has better face recognition performance with less feature dimension and calculation time.

Label Propagation (LP) algorithm is one kind of semi-supervised learning methods. However, its performance in text classification is not good enough, because LP algorithm demands manifold assumption and it has high computational complexity in calculating the similarity of high dimension data. A new method was proposed to combine Latent Dirichlet Allocation (LDA) model with LP algorithm to solve the above problems after analyzing their principles and complexities. It represented documents with latent topics in LDA. On one hand, it reduces the dimension of matrixes; on the other hand, it can help LDA model lead to the classification results with manifold assumption. The experimental results show that the new method performs better than traditional supervised text classification methods in testing sets when labeled data is less than unlabeled data.