CN111027623A - Data-enhanced intelligent terminal security level classification method and system - Google Patents

Abstract

Translated fromChinese

本发明涉及通信安全技术领域，公开了一种数据增强的智能终端安全等级分类方法及系统。所述方法包括：获取智能终端的测试数据集；对所述测试数据集注入对应的标签得到输出样本集；构造新的输入信道信息样本，进行平均数据增强得到输入样本集，进行平均样本构造后的标签矩阵得到新的输出样本集，从而得到新的数据集；将新的数据集作为训练集，对安全等级分类器进行训练；利用安全等级分类器对智能终端进行安全等级分类。本发明利用终端安全测评样本的相关性构造新的伪测评样本，引入随机权重的概念来增加样本构造的随机性，以增强样本集的鲁棒性。该方法可适用于多种基于AI的终端安全等级分类器数据的增强。

The invention relates to the technical field of communication security, and discloses a data-enhanced intelligent terminal security level classification method and system. The method includes: acquiring a test data set of an intelligent terminal; injecting a corresponding label into the test data set to obtain an output sample set; constructing a new input channel information sample, performing average data enhancement to obtain an input sample set, and constructing an average sample The label matrix of , obtains a new output sample set, thereby obtaining a new data set; uses the new data set as a training set to train the security level classifier; uses the security level classifier to classify the security level of the smart terminal. The present invention utilizes the correlation of the terminal security evaluation samples to construct new pseudo-evaluation samples, and introduces the concept of random weight to increase the randomness of the sample structure to enhance the robustness of the sample set. This method can be applied to data enhancement of various AI-based terminal security level classifiers.

Description

Translated fromChinese

数据增强的智能终端安全等级分类方法及系统Data-enhanced intelligent terminal security level classification method and system

技术领域technical field

本发明涉及通信安全技术领域，具体而言，涉及一种数据增强的智能终端安全等级分类方法以及一种数据增强的智能终端安全等级分类系统。The invention relates to the technical field of communication security, in particular to a data-enhanced intelligent terminal security level classification method and a data-enhanced intelligent terminal security level classification system.

背景技术Background technique

随着网络的普及和4G/5G无线网络的发展，智能终端(以下简称终端)的应用深入我们日常的生活，广泛应用于工业、交通、医疗、城市等。相对于PC机，终端的体积受限，能量和计算能力受限，而终端又广泛分布于各种应用场景中，易于接近，更容易受到攻击。随着人们对智能终端的依赖程度加深，终端的安全问题日益凸显。尤其是终端功能越来越多，终端与互联网的结合越来越紧密，例如移动网购平台、手机银行、聊天软件等与我们财产隐私关系密切的第三方软件越来越多，使得使用者把财产信息、个人隐私、商业机密文件等存储在终端中。近来针对移动智能终端的各种攻击层出不穷，对终端的攻击成为攻击网络的一种切入点，终端的安全隐患成为网络安全的重要问题。因此，非常有必要对智能终端的安全进行测评。With the popularization of networks and the development of 4G/5G wireless networks, the application of intelligent terminals (hereinafter referred to as terminals) has penetrated into our daily life and is widely used in industry, transportation, medical care, and cities. Compared with PCs, terminals are limited in size, limited in energy and computing power, and are widely distributed in various application scenarios, making them easier to approach and more vulnerable to attacks. With the deepening of people's reliance on smart terminals, the security problems of terminals are becoming more and more prominent. In particular, there are more and more terminal functions, and the combination of terminals and the Internet is becoming more and more closely. For example, there are more and more third-party software that are closely related to our property privacy, such as mobile online shopping platforms, mobile banking, chat software, etc. Information, personal privacy, commercial confidential documents, etc. are stored in the terminal. Recently, various attacks on mobile intelligent terminals emerge in an endless stream. Attacks on terminals have become an entry point for attacking the network, and the security risks of terminals have become an important issue of network security. Therefore, it is very necessary to evaluate the security of smart terminals.

在移动智能终端安全测评中，根据各移动智能终端的各安全单项测试的结果进行科学量化，根据一定的评测依据，进行终端安全等级的划分，可以实现不同应用场景、不同用户对移动智能终端不同安全需求的重要依据，实现终端不同应用的不同安全级别需求的安全使用。移动智能终端安全测评成为保证终端安全使用的最有效手段之一，在移动智能终端安全测评中根据各项的测试结果科学进行终端安全等级的定级，是不同团体、不同个人对移动智能终端安全需求的重要判据，准确的评价可以实现不同安全级别需求的安全使用。In the mobile intelligent terminal security evaluation, scientific quantification is carried out according to the results of each security single test of each mobile intelligent terminal, and the terminal security level is divided according to a certain evaluation basis, which can realize different application scenarios and different users. It is an important basis for security requirements, and realizes the safe use of different security level requirements of different terminal applications. Mobile intelligent terminal security assessment has become one of the most effective means to ensure the safe use of terminals. In the mobile intelligent terminal security assessment, the terminal security level is scientifically graded according to the test results of various items. It is an important criterion for requirements, and accurate evaluation can realize the safe use of requirements of different security levels.

目前，关于终端安全等级的量化划分以及终端各个安全单项测试的方法都有一定的成果，然后通过综合得到终端的安全性能的一些量化数据，采用先进的分类方法进行终端的安全分级，尤其是基于人工智能(AI)技术，通过学习算法，对终端的安全性能进行客观分类。但是基于人工智能(AI)技术的分类方法，需要大量的数据对模型进行训练，测试数据的时间长，存在数据不足导致的分类精度不够准确的问题。At present, there are certain achievements in the quantitative division of terminal security levels and the methods of individual security tests of terminals. Then, some quantitative data of terminal security performance are obtained through synthesis, and advanced classification methods are used for terminal security classification, especially based on Artificial intelligence (AI) technology objectively classifies the security performance of terminals through learning algorithms. However, the classification method based on artificial intelligence (AI) technology requires a large amount of data to train the model, the test data takes a long time, and there is a problem that the classification accuracy is not accurate enough due to insufficient data.

发明内容SUMMARY OF THE INVENTION

本发明的目的是提供一种智能终端安全等级分类的数据增强方法及系统，以解决上述的基于人工智能技术的分类方法，测试数据的时间长，数据不足导致分类精度不够准确的问题。The purpose of the present invention is to provide a data enhancement method and system for intelligent terminal security level classification, so as to solve the problems of the above-mentioned classification method based on artificial intelligence technology, the time for testing data is long, and the classification accuracy is not accurate due to insufficient data.

为了实现上述目的，本发明第一方面提供一种数据增强的智能终端安全等级分类方法，所述方法包括：In order to achieve the above object, a first aspect of the present invention provides a data-enhanced intelligent terminal security level classification method, the method comprising:

S1)获取智能终端的测试数据集；S1) obtain the test data set of the intelligent terminal;

S2)对所述测试数据集注入对应的标签，得到输出样本集；S2) injecting corresponding labels into the test data set to obtain an output sample set;

S3)根据所述输出样本集构造新的输入信道信息样本；S3) construct a new input channel information sample according to the output sample set;

S4)将所述新的输入信道信息样本进行平均数据增强，得到输入样本集；S4) performing average data enhancement on the new input channel information samples to obtain an input sample set;

S5)将所述输入样本集进行平均样本构造后的标签矩阵，得到新的输出样本集；S5) carrying out the label matrix after the average sample construction on the input sample set to obtain a new output sample set;

S6)根据所述输入样本集和所述新的输出样本集得到新的数据集；S6) obtain a new data set according to the input sample set and the new output sample set;

S7)将所述新的数据集作为训练集，对安全等级分类器进行训练；S7) using the new data set as a training set, training the security level classifier;

S8)基于人工智能模型利用所述安全等级分类器对智能终端进行安全等级分类。S8) Using the security level classifier to classify the security level of the intelligent terminal based on the artificial intelligence model.

进一步地，步骤S1)获取智能终端的测试数据集，包括：Further, step S1) obtains the test data set of the intelligent terminal, including:

S11)对第k台智能终端测试S次，得到测试结果

每次测试结果均由n个测试单例的得分组成，即由向量

表示，其中m_j为第j个测试单例的得分；S11) Test the k-th intelligent terminal for S times, and obtain the test result

Each test result is composed of the scores of n test singles, which is composed of a vector

represents, where m_j is the score of the jth test single case;

S12)将每次测试结果乘以单例的权重函数H(n)得到智能终端的总分Y，其中权重函数H(n)为一个均匀概率密度函数，表示为H＝[h₁,h₂,…,h_S]^T，

即

同时，将智能终端安全等级y划分为W级，设定W-1个门限值为正数η₁,η₂,…,η_W-1，当满足0<Y≤η₁，则定义终端安全级别为1级，当满足η₁<Y≤η₂，则定义安全等级为2级，依此类推，当满足η_W-2<Y≤η_W-1，则定义终端安全等级为K-1级，当满足Y>η_W-1，则定义安全等级为W级；S12) Multiply each test result by the weight function H(n) of the single instance to obtain the total score Y of the smart terminal, wherein the weight function H(n) is a uniform probability density function, expressed as H=[h₁ , h₂ ,…,h_S ]^T ,

which is

At the same time, the security level y of the intelligent terminal is divided into the W level, and the W-1 thresholds are set as positive numbers η₁ , η₂ ,..., η_W-1 . When 0<Y≤η₁ is satisfied, the terminal is defined The security level is level 1. When η₁ <Y≤η₂ is satisfied, the security level is defined as level 2, and so on. When η_W-2 <Y≤η_W-1 is satisfied, the terminal security level is defined as K- Level 1, when Y>η_W-1 is satisfied, the security level is defined as level W;

S13)通过对智能终端的计算总分M_i和安全等级y的测试，得到第k台智能终端的S次测试数据集D_k：S13) Through the test of the total calculated score_Mi and the security level y of the intelligent terminal, the S test data set Dk of the_kth intelligent terminal is obtained:

D_k:D_k＝{X_k,Y_k}，D_k : D_k ={X_k ,Y_k },

其中

in

其中T＝{(M₁,y₁),(M₂,y₂),…,(M_N,y_N)}，y_i∈{1,2,3,4}，i＝1,2,…,N。where T={(M₁ ,y₁ ),(M₂ ,y₂ ),…,(M_N ,y_N )}, y_i ∈{1,2,3,4}, i=1,2, …, N.

进一步地，步骤S2)中所述输出样本集为：Further, the output sample set described in step S2) is:

其中y_k∈{1,2,…,W}。where y_k ∈ {1,2,…,W}.

进一步地，步骤S3)中所述新的输入信道信息样本为：Further, the new input channel information sample described in step S3) is:

其中α₀是一个正整数，表示每次参数评价样本构建的样本数。where α₀ is a positive integer, indicating the number of samples constructed for each parameter evaluation sample.

进一步地，步骤S4)中所述输入样本集为：Further, the input sample set described in step S4) is:

其中，N_k表示进行平均数据增强后的信道信息向量的个数；Among them, N_k represents the number of channel information vectors after average data enhancement;

步骤S5)中所述新的输出样本集为：The new output sample set described in step S5) is:

步骤S6)中所述新的数据集为：The new data set described in step S6) is:

进一步地，步骤S8)基于人工智能模型利用所述安全等级分类器对智能终端进行安全等级分类，包括：根据安全等级的级数W，采用W-1层支持向量机模型，计算智能终端安全等级。Further, step S8) utilizes the described security level classifier to classify the security level of the intelligent terminal based on the artificial intelligence model, including: according to the level W of the security level, adopting the W-1 layer support vector machine model to calculate the security level of the intelligent terminal .

进一步地，所述计算智能终端安全等级，包括以下步骤：Further, the computing intelligent terminal security level includes the following steps:

S81)初始化，令初始变量m＝1；S81) initialize, make initial variable m=1;

S82)将训练集分成两类，其中一类为y＝m，另一类为y＝m+1～W，即得到所述训练集

S82) Divide the training set into two categories, one of which is y=m, and the other is y=m+1～W, that is, the training set is obtained

其中

in

其中，

in,

S83)构造并求解约束最优化问题，公式如下：S83) Construct and solve the constrained optimization problem, the formula is as follows:

求出最优解

式中，

为拉格朗日乘子向量，x_i∈χ＝Rⁿ，y_i∈γ＝{+1,-1}，i＝1,2,3,…,S+N_k；find the optimal solution

In the formula,

is the Lagrange multiplier vector, x_i ∈ χ=Rⁿ , y_i ∈ γ={+1,-1}, i=1,2,3,...,S+N_k ;

S84)计算超平面的法向量值：S84) Calculate the normal vector value of the hyperplane:

式中，w表示高维空间中分类超平面的法向量值；In the formula, w represents the normal vector value of the classification hyperplane in the high-dimensional space;

同时，选择α^(m)的一个正分量

计算超平面的截距值:Also, choose a positive component of α^(m)

Compute the intercept value of the hyperplane:

式中，b表示高维空间中分类超平面的截距值；In the formula, b represents the intercept value of the classification hyperplane in high-dimensional space;

S85)计算得到超平面：S85) calculate the hyperplane:

通过分类决策函数：By classification decision function:

识别安全级别为m级的终端：Identify a terminal with a security level of m:

当f⁽¹⁾(M_i)＝1时，终端安全级别为m级；When f⁽¹⁾ (M_i )=1, the terminal security level is m level;

当f⁽¹⁾(M_i)＝-1时，终端安全级别为m+1～W级；When f⁽¹⁾ (M_i )=-1, the terminal security level is m+1～W level;

S86)判断m的值是否等于W-1：S86) Determine whether the value of m is equal to W-1:

若是，则完成所有安全等级分级；If so, complete all security level classifications;

若不是，则将m进行+1操作，并转至步骤S82)。If not, perform +1 operation on m, and go to step S82).

本发明第二方面提供一种数据增强的智能终端安全等级分类系统，所述系统包括：A second aspect of the present invention provides a data-enhanced intelligent terminal security level classification system, the system comprising:

测试模块，用于对智能终端进行测试，得到测试数据集；The test module is used to test the intelligent terminal and obtain the test data set;

数据增强模块，用于对所述测试数据集注入对应的标签，得到输出样本集；根据所述输出样本集构造新的输入信道信息样本；将所述新的输入信道信息样本进行平均数据增强，得到输入样本集；将所述输入样本集进行平均样本构造后的标签矩阵，得到新的输出样本集；根据所述输入样本集和所述新的输出样本集得到新的数据集；a data enhancement module, configured to inject a corresponding label into the test data set to obtain an output sample set; construct a new input channel information sample according to the output sample set; perform average data enhancement on the new input channel information sample, Obtain an input sample set; perform a label matrix constructed by averaging samples on the input sample set to obtain a new output sample set; obtain a new data set according to the input sample set and the new output sample set;

模型训练模块，用于将所述新的数据集作为训练集，对安全等级分类器进行训练；a model training module, used for training the security level classifier by using the new data set as a training set;

分类模块，用于基于人工智能模型利用所述安全等级分类器对智能终端进行安全等级分类。The classification module is used to classify the security level of the intelligent terminal by using the security level classifier based on the artificial intelligence model.

进一步地，所述对智能终端进行测试，得到测试数据集，包括：Further, test the intelligent terminal to obtain a test data set, including:

对所述智能终端进行多次测试，得到测试结果，所述测试结果由至少一个测试单项的得分组成；Carry out multiple tests on the intelligent terminal to obtain a test result, and the test result consists of the score of at least one test item;

将所述测试结果乘以单例的权重函数得到所述智能终端的计算总分，并定义所述智能终端的安全等级；Multiply the test result by the weight function of the single instance to obtain the calculated total score of the intelligent terminal, and define the security level of the intelligent terminal;

根据所述智能终端的计算总分和安全等级得到所述智能终端的测试数据集。The test data set of the intelligent terminal is obtained according to the calculated total score and the security level of the intelligent terminal.

进一步地，所述基于人工智能模型利用所述安全等级分类器对智能终端进行安全等级分类，包括：根据安全等级的级数W，采用W-1层支持向量机模型，计算智能终端安全等级。Further, using the security level classifier to classify the security level of the intelligent terminal based on the artificial intelligence model includes: calculating the security level of the intelligent terminal by using the W-1 layer support vector machine model according to the level W of the security level.

本发明上述技术方案通过获取智能终端的测试数据集，利用终端安全测评样本的相关性构造新的伪测评样本，引入随机权重的概念来增加样本构造的随机性，以增强样本集的鲁棒性。本发明技术方案的数据集合增强方法可适用于多种基于AI的终端安全等级分类器数据的增强。本发明提供的安全等级分类的数据集合增强方法，可适用于多种智能终端设备，可移植性强。The above technical solution of the present invention increases the randomness of the sample structure by acquiring the test data set of the intelligent terminal, using the correlation of the terminal security evaluation samples to construct a new pseudo-evaluation sample, and introducing the concept of random weight to enhance the robustness of the sample set. . The data set enhancement method of the technical solution of the present invention is applicable to the enhancement of various AI-based terminal security level classifier data. The data set enhancement method for security level classification provided by the present invention can be applied to a variety of intelligent terminal devices and has strong portability.

附图说明Description of drawings

附图是用来提供对本发明实施方式的进一步理解，并且构成说明书的一部分，与下面的具体实施方式一起用于解释本发明实施方式，但并不构成对本发明实施方式的限制。在附图中：The accompanying drawings are used to provide a further understanding of the embodiments of the present invention, and constitute a part of the specification, and together with the following specific embodiments, are used to explain the embodiments of the present invention, but do not limit the embodiments of the present invention. In the attached image:

图1是本发明一种实施方式提供的数据增强的智能终端安全等级分类方法的流程图；1 is a flowchart of a data-enhanced smart terminal security level classification method provided by an embodiment of the present invention;

图2是本发明一种实施方式提供的数据增强的智能终端安全等级分类系统的框图。FIG. 2 is a block diagram of a data-enhanced intelligent terminal security level classification system provided by an embodiment of the present invention.

具体实施方式Detailed ways

以下结合附图对本发明的具体实施方式进行详细说明。应当理解的是，此处所描述的具体实施方式仅用于说明和解释本发明，并不用于限制本发明。The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present invention, but not to limit the present invention.

图1是本发明一种实施方式提供的数据增强的智能终端安全等级分类方法的流程图。如图1所示，本发明实施方式提供的一种数据增强的智能终端安全等级分类方法，包括以下步骤：FIG. 1 is a flowchart of a data-enhanced smart terminal security level classification method provided by an embodiment of the present invention. As shown in FIG. 1 , a data-enhanced smart terminal security level classification method provided by an embodiment of the present invention includes the following steps:

S1)获取智能终端的测试数据集。S1) Obtain the test data set of the smart terminal.

该步骤包括以下子步骤：This step includes the following sub-steps:

S11)对第k台智能终端测试S次，得到测试结果

每次测试结果均由n个测试单例的得分组成，即由向量

表示，其中m_j为第j个测试单例的得分，得分越高表示安全性能越好。例如，所述测试单例包括短信功能、通话功能、第三方软件、内核漏洞、审计功能、存储和删除文件警告。S11) Test the k-th intelligent terminal for S times, and obtain the test result

Each test result is composed of the scores of n test singles, which is composed of a vector

represents, where m_j is the score of the jth test single case, and the higher the score, the better the security performance. For example, the test singleton includes SMS function, call function, third-party software, kernel vulnerability, audit function, storage and deletion file warning.

S12)将每次测试结果乘以单例的权重函数H(n)得到智能终端的总分Y，其中权重函数H(n)为一个均匀概率密度函数，表示为H＝[h₁,h₂,…,h_S]^T，

即

同时，将智能终端安全等级y划分为W级，设定W-1个门限值为正数η₁,η₂,…,η_W-1，当满足0<Y≤η₁，则定义终端安全级别为1级，当满足η₁<Y≤η₂，则定义安全等级为2级，依此类推，当满足η_W-2<Y≤η_W-1，则定义终端安全等级为K-1级，当满足Y>η_W-1，则定义安全等级为W级，安全等级越高表示终端越安全。S12) Multiply each test result by the weight function H(n) of the single instance to obtain the total score Y of the smart terminal, wherein the weight function H(n) is a uniform probability density function, expressed as H=[h₁ , h₂ ,…,h_S ]^T ,

which is

At the same time, the security level y of the intelligent terminal is divided into the W level, and the W-1 thresholds are set as positive numbers η₁ , η₂ ,..., η_W-1 . When 0<Y≤η₁ is satisfied, the terminal is defined The security level is level 1. When η₁ <Y≤η₂ is satisfied, the security level is defined as level 2, and so on. When η_W-2 <Y≤η_W-1 is satisfied, the terminal security level is defined as K- Level 1, when Y>η_W-1 is satisfied, the security level is defined as the W level, and the higher the security level, the safer the terminal.

S13)通过对智能终端的计算总分M_i和安全等级y的测试，得到第k台智能终端的S次测试数据集D_k：S13) Through the test of the total calculated score_Mi and the security level y of the intelligent terminal, the S test data set Dk of the_kth intelligent terminal is obtained:

D_k:D_k＝{X_k,Y_k}，D_k : D_k ={X_k ,Y_k },

其中

in

其中T＝{(M₁,y₁),(M₂,y₂),…,(M_N,y_N)}，y_i∈{1,2,3,4}，i＝1,2,…,N。where T={(M₁ ,y₁ ),(M₂ ,y₂ ),…,(M_N ,y_N )}, y_i ∈{1,2,3,4}, i=1,2, …, N.

S2)对所述测试数据集注入对应的标签，得到输出样本集。S2) Inject corresponding labels into the test data set to obtain an output sample set.

所述输出样本集为：The output sample set is:

其中y_k∈{1,2,…,W}。where y_k ∈ {1,2,…,W}.

S3)根据所述输出样本集构造新的输入信道信息样本。S3) Construct a new input channel information sample according to the output sample set.

所述新的输入信道信息样本为：The new input channel information sample is:

其中α₀是一个正整数，表示每次参数评价样本构建的样本数。where α₀ is a positive integer, indicating the number of samples constructed for each parameter evaluation sample.

S4)将所述新的输入信道信息样本进行平均数据增强，得到输入样本集。S4) Perform average data enhancement on the new input channel information samples to obtain an input sample set.

所述输入样本集为：The input sample set is:

其中，N_k表示进行平均数据增强后的信道信息向量的个数。Among them, N_k represents the number of channel information vectors after average data enhancement.

S5)将所述输入样本集进行平均样本构造后的标签矩阵，得到新的输出样本集。S5) The input sample set is averaged to construct a label matrix to obtain a new output sample set.

所述新的输出样本集为：The new output sample set is:

S6)根据所述输入样本集和所述新的输出样本集得到新的数据集。S6) Obtain a new data set according to the input sample set and the new output sample set.

所述新的数据集为：

The new dataset is:

S7)将所述新的数据集作为训练集，对安全等级分类器进行训练。S7) Using the new data set as a training set, train the security level classifier.

S8)基于人工智能模型利用所述安全等级分类器对智能终端进行安全等级分类。S8) Using the security level classifier to classify the security level of the intelligent terminal based on the artificial intelligence model.

根据安全等级的级数W，采用W-1层支持向量机模型，计算智能终端安全等级，包括以下子步骤:According to the level W of the security level, the W-1 layer support vector machine model is used to calculate the security level of the intelligent terminal, including the following sub-steps:

S81)初始化，令初始变量m＝1。S81) Initialize, let the initial variable m=1.

S82)将训练集分成两类，其中一类为y＝m，另一类为y＝m+1～W，即得到所述训练集

S82) Divide the training set into two categories, one of which is y=m, and the other is y=m+1～W, that is, the training set is obtained

其中

in

其中，

in,

S83)构造并求解约束最优化问题，公式如下：S83) Construct and solve the constrained optimization problem, the formula is as follows:

求出最优解

式中，

为拉格朗日乘子向量，x_i∈χ＝Rⁿ，y_i∈γ＝{+1,-1}，i＝1,2,3,…,S+N_k。find the optimal solution

In the formula,

is the Lagrange multiplier vector, x_i ∈ χ=Rⁿ , y_i ∈ γ={+1,-1}, i=1,2,3,...,S+N_k .

S84)计算超平面的法向量值：S84) Calculate the normal vector value of the hyperplane:

式中，w表示高维空间中分类超平面的法向量值；In the formula, w represents the normal vector value of the classification hyperplane in the high-dimensional space;

同时，选择α^(m)的一个正分量

计算超平面的截距值:Also, choose a positive component of α^(m)

Compute the intercept value of the hyperplane:

式中，b表示高维空间中分类超平面的截距值。In the formula, b represents the intercept value of the classification hyperplane in the high-dimensional space.

S85)计算得到超平面：S85) calculate the hyperplane:

通过分类决策函数：By classification decision function:

识别安全级别为m级的终端：Identify a terminal with a security level of m:

当f⁽¹⁾(M_i)＝1时，终端安全级别为m级；When f⁽¹⁾ (M_i )=1, the terminal security level is m level;

当f⁽¹⁾(M_i)＝-1时，终端安全级别为m+1～W级。When f⁽¹⁾ (M_i )=-1, the terminal security level is m+1～W level.

S86)判断m的值是否等于W-1：S86) Determine whether the value of m is equal to W-1:

若是，则完成所有安全等级分级；If so, complete all security level classifications;

若不是，则将m进行+1操作，并转至步骤S82)。If not, perform +1 operation on m, and go to step S82).

例如，安全等级的级数为4级，在步骤S12)中需设定3个门限值η₁,η₂,η₃，在步骤S8)中采用3层支持向量机模型。For example, if the level of security level is 4, three threshold values η₁ , η₂ , and η₃ need to be set in step S12), and a three-layer support vector machine model is used in step S8).

图2是本发明一种实施方式提供的数据增强的智能终端安全等级分类系统的框图。如图2所示，本发明实施方式提供的一种数据增强的智能终端安全等级分类系统，包括测试模块、数据增强模块、模型训练模块和分类模块。FIG. 2 is a block diagram of a data-enhanced intelligent terminal security level classification system provided by an embodiment of the present invention. As shown in FIG. 2 , a data-enhanced intelligent terminal security level classification system provided by an embodiment of the present invention includes a test module, a data enhancement module, a model training module and a classification module.

所述测试模块用于对智能终端进行测试，得到测试数据集。具体地，包括：对所述智能终端进行多次测试，得到测试结果，所述测试结果由至少一个测试单项的得分组成；将所述测试结果乘以单例的权重函数得到所述智能终端的计算总分，并定义所述智能终端的安全等级；根据所述智能终端的计算总分和安全等级得到所述智能终端的测试数据集。The test module is used to test the intelligent terminal to obtain a test data set. Specifically, it includes: performing multiple tests on the smart terminal to obtain a test result, where the test result consists of the score of at least one test item; multiplying the test result by the weight function of the single instance to obtain the test result of the smart terminal Calculate the total score and define the security level of the intelligent terminal; obtain the test data set of the intelligent terminal according to the calculated total score and the security level of the intelligent terminal.

所述数据增强模块用于对所述测试数据集注入对应的标签，得到输出样本集；根据所述输出样本集构造新的输入信道信息样本；将所述新的输入信道信息样本进行平均数据增强，得到输入样本集；将所述输入样本集进行平均样本构造后的标签矩阵，得到新的输出样本集；根据所述输入样本集和所述新的输出样本集得到新的数据集。The data enhancement module is used for injecting corresponding labels into the test data set to obtain an output sample set; constructing a new input channel information sample according to the output sample set; performing average data enhancement on the new input channel information sample , to obtain an input sample set; the label matrix constructed by averaging the samples is performed on the input sample set to obtain a new output sample set; a new data set is obtained according to the input sample set and the new output sample set.

所述模型训练模块用于将所述新的数据集作为训练集，对安全等级分类器进行训练。The model training module is used for training the security level classifier by using the new data set as a training set.

所述分类模块用于基于人工智能模型利用所述安全等级分类器对智能终端进行安全等级分类。例如：根据安全等级的级数W，采用W-1层支持向量机模型，计算智能终端安全等级。The classification module is configured to use the security level classifier to classify the security level of the intelligent terminal based on the artificial intelligence model. For example, according to the level W of the security level, the W-1 layer support vector machine model is used to calculate the security level of the intelligent terminal.

本发明实施方式还提供一种机器可读存储介质，该机器可读存储介质上存储有计算机程序指令，该计算机程序指令被处理器执行时实现上述的数据增强的智能终端安全等级分类方法。Embodiments of the present invention further provide a machine-readable storage medium, where computer program instructions are stored thereon, and when the computer program instructions are executed by a processor, the above-mentioned data-enhanced smart terminal security level classification method is implemented.

本领域技术人员可以理解实现上述实施方式的方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成，该程序存储在一个存储介质中，包括若干指令用以使得单片机、芯片或处理器(processor)执行本发明各个实施方式所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。Those skilled in the art can understand that all or part of the steps in the method for implementing the above embodiments can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium and includes several instructions to make the single chip, chip or processing The processor (processor) executes all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .

以上结合附图详细描述了本发明的可选实施方式，但是，本发明实施方式并不限于上述实施方式中的具体细节，在本发明实施方式的技术构思范围内，可以对本发明实施方式的技术方案进行多种简单变型，这些简单变型均属于本发明实施方式的保护范围。此外，本发明的各种不同的实施方式之间也可以进行任意组合，只要其不违背本发明实施方式的思想，其同样应当视为本发明实施方式所公开的内容。The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details of the above-mentioned embodiments. Within the scope of the technical concept of the embodiments of the present invention, the technical The scheme undergoes various simple modifications, and these simple modifications all belong to the protection scope of the embodiments of the present invention. In addition, various different embodiments of the present invention can also be combined arbitrarily, as long as they do not violate the idea of the embodiments of the present invention, they should also be regarded as the contents disclosed by the embodiments of the present invention.

Claims (10)

1. A data-enhanced intelligent terminal security level classification method is characterized by comprising the following steps:

s1) acquiring a test data set of the intelligent terminal;

s2) injecting corresponding labels into the test data set to obtain an output sample set;

s3) constructing new input channel information samples from the output sample set;

s4) carrying out average data enhancement on the new input channel information sample to obtain an input sample set;

s5) carrying out label matrix after average sample construction on the input sample set to obtain a new output sample set;

s6) obtaining a new data set according to the input sample set and the new output sample set;

s7) taking the new data set as a training set, and training a safety class classifier;

s8) carrying out safety level classification on the intelligent terminal by using the safety level classifier based on the artificial intelligence model.

2. The method for classifying security levels of an intelligent terminal with enhanced data according to claim 1, wherein the step S1) of obtaining a test data set of the intelligent terminal comprises:

s11) testing the kth intelligent terminal S times to obtain a test result

Each test result is composed of scores of n test singletons, namely vectors

Is represented by the formula (I) in which m_jScore for jth test case;

s12), multiplying each test result by a weight function H (n) of a single case to obtain a total score Y of the intelligent terminal, wherein the weight function H (n) is a uniform probability density function and is expressed as H ═ H₁,h₂,…,h_S]^T，

Namely, it is

Meanwhile, the security level y of the intelligent terminal is divided into W levels, and W-1 threshold values are set to be positive numbers η₁,η₂,…,η_W-1When it satisfies 0<Y≤η₁Then define the terminal security level as level 1, when η is satisfied₁<Y≤η₂Then the security level is defined as level 2, and so on, when η is satisfied_W-2<Y≤η_W-1Defining the terminal security level as K-1 level, and satisfying Y>η_W-1Defining the security level as W level;

s13) total score M calculated through intelligent terminal_iAnd testing the security level y to obtain an S-time test data set D of the kth intelligent terminal_k：

D_k:D_k＝{X_k,Y_k}，

Wherein

Wherein T { (M)₁,y₁),(M₂,y₂),…,(M_N,y_N)}，y_i∈{1,2,3,4}，i＝1,2,…,N。

3. The method for classifying security levels of a data-enhanced intelligent terminal according to claim 2, wherein the output sample set in step S2) is:

wherein y is_k∈{1,2,…,W}。

4. The method for classifying security levels of a data-enhanced intelligent terminal according to claim 3, wherein the new input channel information samples in step S3) are:

α therein₀Is a positive integer representing the number of samples constructed for each parametric evaluation sample.

5. The data-enhanced intelligent terminal security level classification method according to claim 4,

the input sample set in step S4) is:

wherein N is_kRepresenting the number of channel information vectors after average data enhancement;

step S5) the new output sample set is:

the new data set in step S6) is:

6. the data-enhanced intelligent terminal security level classification method according to claim 5, wherein the step S8) of performing security level classification on the intelligent terminal by using the security level classifier based on the artificial intelligence model comprises:

and calculating the safety level of the intelligent terminal by adopting a W-1 layer support vector machine model according to the level W of the safety level.

7. The data-enhanced intelligent terminal security level classification method according to claim 6, wherein the calculating of the intelligent terminal security level comprises the following steps:

s81), initializing the initial variable m to 1;

s82) dividing the training set into two types, wherein one type is y ═ m, and the other type is y ═ m + 1-W, and obtaining the training set

Wherein

Wherein,

s83) constructing and solving a constrained optimization problem, the formula is as follows:

finding the optimal solution

In the formula,

is a Lagrange multiplier vector, x_i∈χ＝Rⁿ，y_i∈γ＝{+1,-1}，i＝1,2,3,…,S+N_k；

S84) calculating the normal magnitude of the hyperplane:

in the formula, w represents the normal magnitude of the classification hyperplane in the high-dimensional space;

at the same time, select α^(m)A positive component of

Calculating the intercept value of the hyperplane:

in the formula, b represents an intercept value of a classification hyperplane in a high-dimensional space;

s85) calculating a hyperplane:

by means of a classification decision function:

identifying a terminal with a security level of m:

when f is⁽¹⁾(M_i) When the terminal is 1, the security level of the terminal is m level;

when f is⁽¹⁾(M_i) When the terminal is equal to-1, the security level of the terminal is m + 1-W;

s86) determining whether the value of m is equal to W-1:

if yes, finishing all safety grade grading;

if not, perform operation +1 on m, and go to step S82).

8. A data-enhanced intelligent terminal security level classification system is characterized by comprising:

the testing module is used for testing the intelligent terminal to obtain a testing data set;

the data enhancement module is used for injecting corresponding labels into the test data set to obtain an output sample set; constructing a new input channel information sample according to the output sample set; carrying out average data enhancement on the new input channel information sample to obtain an input sample set; carrying out label matrix after average sample construction on the input sample set to obtain a new output sample set; obtaining a new data set according to the input sample set and the new output sample set;

the model training module is used for taking the new data set as a training set and training a safety class classifier;

and the classification module is used for classifying the safety level of the intelligent terminal by using the safety level classifier based on the artificial intelligence model.

9. The system according to claim 8, wherein the testing the smart terminal to obtain a test data set comprises:

testing the intelligent terminal for multiple times to obtain a test result, wherein the test result consists of the score of at least one test single item;

multiplying the test result by a weight function of a single case to obtain a total calculation score of the intelligent terminal, and defining the safety level of the intelligent terminal;

and obtaining a test data set of the intelligent terminal according to the calculated total score and the safety level of the intelligent terminal.

10. The system according to claim 8, wherein the artificial intelligence model-based security level classification of the smart terminal using the security level classifier comprises:

and calculating the safety level of the intelligent terminal by adopting a W-1 layer support vector machine model according to the level W of the safety level.

Images