As medical follow-up work is more and more valued, the task of obtaining information related to the follow-up guidance through medical image analysis has become increasingly important. However, most deep learning-based methods are not suitable for dealing with such task. In order to solve the problem, a Multi-term Knowledge Distillation (MKD) model was proposed. Firstly, with the advantage of knowledge distillation in model transfer, the classification task with long-term follow-up information was converted into a model transfer task based on domain knowledge. Then, the follow-up knowledge contained in the long-term medical images was fully utilized to realize the long-term classification of lung nodules. At the same time, facing the problem that the data collected during the follow-up process were relatively unbalanced every year, a meta-learning method based normalization method was proposed, and therefore improving the training accuracy of the model in the semi-supervised mode effectively. Experimental results on NLST dataset show that, the proposed MKD model has better classification accuracy in the task of long-term lung nodule classification than the deep learning classification models such as GoogleNet. When the amount of unbalanced long-term data reaches 800 cases, the MKD enhanced by meta-learning method can improve the accuracy by up to 7 percentage points compared with the existing state-of-the-art models.

Aiming at the problem of high misdiagnosis rate, high false positive rate and low detection accuracy in pulmonary nodule computer-aided detection, a method of nodular detection based on three-dimensional shape index and Hessian matrix eigenvalue was proposed. Firstly, the parenchyma region was extracted and the eigenvalues and eigenvectors of the Hessian matrix were calculated. Secondly, the three-dimensional shape index formula was deduced by the two-dimensional shape index, and the improved three-dimensional spherical like filter was constructed. Finally, in the parenchyma volume, the suspected nodule region was detected, and more false-positive regions were removed. The nodules were detected by the three-dimensional volume data, and the detected coordinates were input as the seeds of belief connect, and the three-dimensional data was splited to pick out three-dimensional nodules. The experimental results show that the proposed algorithm can effectively detect different types of pulmonary nodules, and has better detection effect on the ground glass nodules which are more difficult to detect, reduces the false positive rate of nodules, and finally reaches 92.36% accuracy rate and 96.52% sensitivity.

Kinship network is made up of marriage and parent-child relationship. Searching a special relationship on a huge kinship network is very difficult. This paper proposed two algorithms by extending breadth-first-search method: radius-search and directional-search. The data of the kinship network was extracted from Hebei province population database, which included about 4150000 vertexes, and about 10880000 edges. The network stored bilateral relationships, which declined some unnecessary back tracking. The experimental results show that the kinship retrospect algorithm can exactly locate some specific persons by the network. At the same time the algorithms can achieve high performance and guarantee high flexibility.

Concerning the classification problems of complex data distribution of scientific practice, economic life and many other fields, the correlation between variables could not be well described by using traditional Support Vector Machine (SVM), which would influence the classification performance. For this situation, Q-Gaussian function that was a parametric generalization of Gaussian function was put forward as the kernel function of SVM, and a financial early-warning model based on SVM with Q-Gaussian kernel was presented. Based on the financial data of A-share manufacturing listed companies of the Shanghai and Shenzhen stock markets, T-2 and T-3 financial early-warning model were constructed in experiments, the significance test was used to select some suitable indicators and the Cross Validation (CV) was used to determine model parameters. Compared to SVM model with Gaussian kernel, the forecasting accuracies of T-2 and T-3 model constructed by SVM with Q-Gaussian kernel were improved about 3%, and high-cost type I errors were reduced by at most 14.29%.

To resolve the problem with selecting the samples in the Differential Power Attack (DPA), this paper proposed a set of samples selection method. Based on the given experimental platform, the mode and amount of samples selection were proposed through theoretical analysis, and then were validated by experiments. For Advanced Encryption Standard (AES), this paper put forward the samples selection methods for simulation test and practical experimentation, and proved that the proposed method was right. The results show that the simulation sample plaintext attack should be taken in sequence, with the quantity of a full array. And the attack should be measured directly using a large number of random numbers. There is a big difference in the explicit requirements of the sample.