Research And Practice Of Mobile Application Software Security Detection Technology

With the rapid development of 4G,5G networks and the Internet of Things,mobile applications have penetrated into all aspects of people's lives and participated in people's smart lives.The Android operating system,as the mobile operating system that has the highest share and is most threatened by malicious applications,faces major challenges such as privacy leaks,malicious deductions,and system damage.Therefore,the research of mobile terminal application security detection technology for Android applications is a topic of great practical significance.This paper mainly uses machine learning classifiers to detect application security,and conducts research from the aspects of feature dimensionality reduction and imbalanced data set classification,as follows:1.Aiming at the problem of different contributions of applied features to classification,a feature selection algorithm based on feature contribution is proposed.The algorithm obtains the contribution weight of the feature to the original sample set classification detection by quantifying the degree of contribution of the feature within and between classes,and then sets a threshold to select a subset of low-dimensional features with high contributions for subsequent detection.The experimental results show that the algorithm has better detection effect and is applicable to the field of application security detection.Compared with the traditional feature selection algorithm,it can achieve an ideal detection effect with a small number of features.2.Aiming at the problems of fine feature granularity and high dimensionality in the application data set,a feature transformation algorithm based on deep belief network is proposed.The algorithm uses the non-linear transformation capability of the deep belief network to abstract features in depth to achieve the dimensionality reduction of the original high-dimensional features.Subsequent detection experiments show that the algorithm can better represent the structure of high-dimensional space,and has better detection results than other feature selection algorithms and linear feature transformation algorithms.3.Aiming at the problems of imbalanced classification of application data sets and multiclass classification tendency,K-Means clustering is proposed to improve Synthetic Minority Oversampling Technique(SMOTE)algorithm.K-Means clustering algorithm is used to perform clustering and to select clusters with a high proportion of fewer classes.Then,the SMOTE algorithm is used to synthesize a small number of samples in the cluster according to the sparseness,and finally change the category distribution of the original data set.The experimental results show that the algorithm effectively improves the shortcomings of the SMOTE algorithm and improves the classification performance of the classifier for small class samples and the overall sample set.In summary,the contribution point of this paper is to propose an application security detection method based on the mobile terminal application characteristics and the characteristics of the sample set category distribution.The results show that the application security detection method proposed in this paper can effectively improve the detection performance and has certain practical significance in the application security detection of mobile terminals.