Installation package classification method, training method of installation package classification model and related equipment

By dynamic and static analysis of the installation package in the sandbox, extracting features and performing classification network analysis, the problem of poor interpretability of the model in the existing technology is solved, and accurate detection of the installation package of the black and gray industry is achieved.

CN119004162APending Publication Date: 2024-11-22BEIJING HONGTENG INTELLIGENT TECH CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202310583187.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-22
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

When detecting installation packages in the prior art, the model is poor interpretability and difficult to construct, resulting in the inability to effectively detect the installation packages of black and gray products.

Method used

By dynamically analyzing and static analysis of the installation package in the sandbox, dynamic features and static features are extracted, and dynamic classification network and static classification network are input respectively for analysis, and finally, the classification results of the installation package are comprehensively determined.

Benefits of technology

It reduces the difficulty of building the model, enhances the interpretability and application flexibility of the model, and realizes accurate detection of black and gray installation packages.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004162A_ABST
    Figure CN119004162A_ABST
Patent Text Reader

Abstract

The invention discloses an installation package classification method, a training method of an installation package classification model and related equipment. According to the installation package classification method, dynamic analysis and static analysis are carried out on an installation package in a sandbox to obtain corresponding dynamic features and static features, and then modeling is carried out on the dynamic features and the static features for analysis, so that the modeling difficulty is reduced, and corresponding classification results, namely a dynamic classification result and a static classification result, are obtained. And the interpretability of the model and the flexibility of application are enhanced. And finally, based on the dynamic classification result and the static classification result, a final classification result is comprehensively obtained, and the final classification result combines a dynamic analysis result and a static analysis result, so that the classification features are more accurate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of computers, and in particular to an installation package classification method, an installation package classification model training method, and related equipment. Background Art

[0002] As the Internet becomes more and more powerful, cybercrimes such as online fraud and online gambling are generally carried out through software. Therefore, it is only necessary to identify the corresponding black and gray product installation packages before the software is installed to effectively control cybercrime. However, the detection of installation packages is currently limited to the detection of their source code files. The model construction has poor interpretability and is difficult to build, resulting in the inability of the constructed model to detect black and gray product installation packages. Summary of the invention

[0003] The embodiments of the present application provide an installation package classification method, an installation package classification model training method and related equipment, which can quickly and accurately detect black and gray product installation packages.

[0004] In a first aspect, an embodiment of the present application provides an installation package classification method, comprising: parsing the installation package in a sandbox to obtain target dynamic features and target static features; inputting the dynamic features into a dynamic classification network to obtain dynamic classification results; inputting the static features into a static classification network to obtain static classification results; and determining a final classification result based on the dynamic classification results and the static classification results.

[0005] In a possible implementation, parsing the installation package in the sandbox to obtain target dynamic features and target static features specifically includes: statically parsing the installation package in the sandbox to obtain target static features; dynamically parsing the installation package in the sandbox to obtain target dynamic features.

[0006] In a possible implementation, statically parsing the installation package in a sandbox to obtain target static features specifically includes: statically parsing the installation package in a sandbox to obtain preliminary static features, wherein the preliminary static features include icons, texts, software names, file names, and file hashes; performing information gain on the preliminary static features to obtain target static features.

[0007] In a possible implementation, the target dynamic features include text information and screenshot information, and the dynamic parsing of the installation package in the sandbox to obtain the target dynamic features specifically includes: dynamically parsing the installation package in the sandbox to obtain preliminary dynamic features, and the preliminary dynamic features include network requests and running screenshots; parsing the network requests to obtain text information; identifying the running screenshots to obtain screenshot information.

[0008] In a possible implementation, the network request includes a protocol address and a text label, the text information includes an address file, a path file, label information and a text file, and the parsing of the network request to obtain the text information specifically includes: parsing the protocol address to obtain an address file and a path file; parsing the text label to obtain label information and a text file.

[0009] In a possible implementation, the identifying the running screenshot to obtain the screenshot information specifically includes: performing image recognition on the running screenshot to obtain the screenshot content; performing image hashing on the running screenshot to obtain the screenshot features.

[0010] In a second aspect, an embodiment of the present application provides a training method for an installation package classification model, characterized in that the installation package classification model includes a dynamic classification network and a static classification network, and the training method for the installation package classification model includes: obtaining an installation package sample set, the installation package sample set including multiple installation package samples, each installation package sample is marked with a corresponding installation package classification label; parsing the installation package samples in a sandbox to obtain sample dynamic features and sample static features; inputting the sample dynamic features into a dynamic classification network to obtain sample dynamic classification results; inputting the sample static features into a static classification network to obtain sample static classification results; determining the final classification results of the samples based on the sample dynamic classification results and the sample static classification results; adjusting parameters of the installation package classification model based on the sample final classification results and the installation package classification label until the training end conditions are met to obtain a trained installation package classification model.

[0011] In a third aspect, an embodiment of the present application provides an installation package classification device, which includes: a sandbox parsing module, used to parse the installation package in the sandbox to obtain target dynamic features and target static features; a dynamic classification module, used to input the dynamic features into a dynamic classification network to obtain dynamic classification results; a static classification module, used to input the static features into a static classification network to obtain static classification results; and a final classification module, used to determine the final classification result based on the dynamic classification result and the static classification result.

[0012] In a possible implementation, the sandbox parsing module specifically includes: a static parsing submodule, used to statically parse the installation package in the sandbox to obtain target static features; and a dynamic parsing submodule, used to dynamically parse the installation package in the sandbox to obtain target dynamic features.

[0013] In a possible implementation, the static parsing submodule specifically includes: a static feature unit, used to statically parse the installation package in a sandbox to obtain preliminary static features, wherein the preliminary static features include icons, texts, software names, file names, and file hashes; an information gain unit, used to perform information gain on the preliminary static features to obtain target static features.

[0014] In a possible implementation, the target dynamic features include text information and screenshot information, and the dynamic parsing submodule specifically includes: a dynamic feature unit, used to dynamically parse the installation package in a sandbox to obtain preliminary dynamic features, and the preliminary dynamic features include network requests and running screenshots; a request parsing unit, used to parse the network request to obtain text information; an image parsing unit, used to identify the running screenshot to obtain screenshot information.

[0015] In a possible implementation, the network request includes a protocol address and a text label, the text information includes an address file, a path file, label information and a text file, and the request parsing unit is specifically used to execute: parsing the protocol address to obtain an address file and a path file; parsing the text label to obtain label information and a text file.

[0016] In a possible implementation, the screenshot information includes screenshot content and screenshot features, and the image parsing unit is specifically used to execute: performing image recognition on the running screenshot to obtain the screenshot content; performing image hashing on the running screenshot to obtain the screenshot features.

[0017] In a fourth aspect, an embodiment of the present application provides a training device for an installation package classification model, characterized in that the installation package classification model includes a dynamic classification network and a static classification network, and the training device for the installation package classification model includes: a sample acquisition module, used to obtain an installation package sample set, the installation package sample set includes multiple installation package samples, and each installation package sample is marked with a corresponding installation package classification label; a sample parsing module, used to parse the installation package sample in a sandbox to obtain sample dynamic features and sample static features; a first classification module, used to input the sample dynamic features into a dynamic classification network to obtain a sample dynamic classification result; a second classification module, used to input the sample static features into a static classification network to obtain a sample static classification result; a sample result module, used to determine the final classification result of the sample based on the sample dynamic classification result and the sample static classification result; a parameter adjustment module, used to adjust the parameters of the installation package classification model according to the sample final classification result and the installation package classification label until the training end conditions are met to obtain a trained installation package classification model.

[0018] In a fifth aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded and executed by the processor: parsing the installation package in a sandbox to obtain target dynamic features and target static features; inputting the dynamic features into a dynamic classification network to obtain dynamic classification results; inputting the static features into a static classification network to obtain static classification results; and determining the final classification result based on the dynamic classification results and the static classification results.

[0019] In a possible implementation, when the processor parses the installation package in the sandbox to obtain the target dynamic features and the target static features, it specifically executes: statically parsing the installation package in the sandbox to obtain the target static features; dynamically parsing the installation package in the sandbox to obtain the target dynamic features.

[0020] In a possible implementation, when the processor statically parses the installation package in the sandbox to obtain target static features, the processor specifically executes: statically parses the installation package in the sandbox to obtain preliminary static features, where the preliminary static features include icons, texts, software names, file names, and file hashes; and performs information gain on the preliminary static features to obtain target static features.

[0021] In a possible implementation, the target dynamic features include text information and screenshot information. When the processor dynamically parses the installation package in the sandbox and obtains the target dynamic features, it specifically executes: dynamically parses the installation package in the sandbox to obtain preliminary dynamic features, and the preliminary dynamic features include network requests and running screenshots; parses the network request to obtain text information; and identifies the running screenshot to obtain screenshot information.

[0022] In a possible implementation, the network request includes a protocol address and a text label, and the text information includes an address file, a path file, label information, and a text file. When the processor parses the network request and obtains the text information, it specifically executes: parsing the protocol address to obtain the address file and the path file; parsing the text label to obtain the label information and the text file.

[0023] In a possible implementation, the screenshot information includes screenshot content and screenshot features. When the processor identifies the running screenshot and obtains the screenshot information, it specifically executes: performing image recognition on the running screenshot to obtain the screenshot content; performing image hashing on the running screenshot to obtain the screenshot features.

[0024] In a sixth aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory; wherein the memory stores a computer program, the installation package classification model comprises a dynamic classification network and a static classification network, and the computer program is suitable for being loaded and executed by the processor: obtaining an installation package sample set, the installation package sample set comprising multiple installation package samples, each installation package sample being marked with a corresponding installation package classification label; parsing the installation package samples in a sandbox to obtain sample dynamic features and sample static features; inputting the sample dynamic features into a dynamic classification network to obtain sample dynamic classification results; inputting the sample static features into a static classification network to obtain sample static classification results; determining a final classification result of the sample based on the sample dynamic classification results and the sample static classification results; adjusting parameters of the installation package classification model based on the sample final classification results and the installation package classification label until the training end conditions are met to obtain a trained installation package classification model.

[0025] In the seventh aspect, an embodiment of the present application provides a computer storage medium, which stores multiple instructions, and the instructions are suitable for being loaded by a processor and executing the method steps provided by the first aspect of the embodiment of the present application or any one of the implementation methods of the first aspect.

[0026] In an eighth aspect, an embodiment of the present application provides a computer program product comprising instructions, which, when executed on a computer or a processor, enables the above-mentioned computer or processor to execute the encoding method provided by the first aspect of the embodiment of the present application or any possible implementation of the first aspect.

[0027] In the embodiment of the present application, the installation package is dynamically parsed and statically parsed in a sandbox to obtain corresponding dynamic features and static features, and then the dynamic features and static features are modeled and analyzed respectively to reduce the difficulty of modeling, and the corresponding classification results, i.e., dynamic classification results and static classification results, are obtained, and the interpretability of the model and the flexibility of application are also enhanced. Finally, based on the dynamic classification results and the static classification results, the final classification results are obtained comprehensively, and the final classification results combine the results of dynamic analysis and the results of static analysis, so the classification features are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0029] Figure 1A schematic diagram of an exemplary system architecture to which the technical solution of the embodiments of the present application can be applied is shown.

[0030] Figure 2 A flow chart of an installation package classification method provided in an embodiment of the present application is shown.

[0031] Figure 3 Shown according to Figure 2 A specific implementation flowchart of step S100 in the installation package classification method shown in the corresponding embodiment.

[0032] Figure 4 Shown according to Figure 3 A specific implementation flowchart of step S120 in the installation package classification method shown in the corresponding embodiment.

[0033] Figure 5 A flow chart of a method for training an installation package classification model provided in an embodiment of the present application is shown.

[0034] Figure 6 A structural schematic diagram of a packaging classification model provided in an embodiment of the present application is shown.

[0035] Figure 7 A schematic diagram of the structure of an installation package classification device provided in an embodiment of the present application.

[0036] Figure 8 A structural schematic diagram of a training device for an installation package classification model provided in an embodiment of the present application.

[0037] Fig. 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0038] Fig.10 A schematic diagram of the structure of another electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concept of the example embodiments to those skilled in the art.

[0040] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.

[0041] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0042] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0043] Figure 1 A schematic diagram of an exemplary system architecture to which the technical solution of the embodiments of the present application can be applied is shown.

[0044] like Figure 1 As shown, the system architecture may include terminal devices (such as Figure 1 The embodiment of the present invention is a schematic diagram of a mobile phone 101, a tablet computer 102, a portable computer 103, a desktop computer, etc., a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the terminal device and the server 105. The network 104 may include various connection types, such as a wired communication link, a wireless communication link, etc.

[0045] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. According to the implementation requirements, there may be any number of terminal devices, networks and servers. For example, the server 105 may be a server cluster composed of multiple servers.

[0046] The user can use the terminal device to interact with the server 105 through the network 104 to receive or send messages, etc. The server 105 can be a server that provides various services. For example, the user uses the terminal device 103 (or the terminal device 101 or 102) to upload the installation package to the server 105, and the server 105 can parse the installation package in the sandbox to obtain the target dynamic features and the target static features; the target dynamic features are input into the dynamic classification network to obtain the dynamic classification results; the target static features are input into the static classification network to obtain the static classification results; according to the dynamic classification results and the static classification results, the final classification results are determined. It should be noted that the installation package classification method provided in the embodiment of the present application is generally executed by the server 105, and accordingly, the installation package classification device is generally arranged in the server 105. However, in other embodiments of the present application, the terminal device may also have similar functions to the server, so as to execute the installation package classification scheme provided in the embodiment of the present application.

[0047] The implementation details of the technical solution of the embodiment of the present application are described in detail below:

[0048] Figure 2 A flowchart of an installation package classification method according to an embodiment of the present application is shown. The installation package classification method can be executed by a server, which can be Figure 1 Refer to the server shown in Figure 2 As shown, the installation package classification method at least includes:

[0049] Step S100, parsing the installation package in the sandbox to obtain target dynamic features and target static features.

[0050] Step S200, inputting the target dynamic features into a dynamic classification network to obtain a dynamic classification result.

[0051] Step S300, inputting the target static features into a static classification network to obtain a static classification result.

[0052] Step S400: determining a final classification result according to the dynamic classification result and the static classification result.

[0053] In an embodiment of the present application, the installation package is first parsed and extracted in a sandbox, and the parsing and extraction is divided into dynamic parsing and extraction and static parsing and extraction, and the target dynamic features and target static features are obtained respectively. Then, independent judgments are made based on the target dynamic features and target static features, and the corresponding classification results, i.e., dynamic classification results and static classification results, are determined respectively, so as to reduce the difficulty of modeling, and at the same time, the interpretability of the model and the flexibility of application can be enhanced. Finally, based on the dynamic classification results and the static classification results, the final classification results are obtained comprehensively, and the final classification results combine the results of dynamic analysis and the results of static analysis, so the classification features are more accurate.

[0054] In step S100, the installation package is sent to a sandbox for parsing, which can improve the security of the parsing. The sandbox is an integrated parsing tool for the installation package, which can be an apk sandbox for parsing apk installation packages, a windows sandbox for parsing exe installation packages, or an ipa sandbox for parsing ipa installation packages, which is not limited in this application.

[0055] The analysis of the installation package is divided into dynamic analysis and static analysis, which obtain static features and dynamic features respectively.

[0056] Specifically, in some embodiments, the specific implementation of step S100 can be found in Figure 3 . Figure 3 is based on Figure 2 The detailed description of step S100 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, step S100 may include the following steps:

[0057] Step S110, statically analyzing the installation package in the sandbox to obtain target static features.

[0058] Step S120, dynamically parse the installation package in the sandbox to obtain target dynamic features.

[0059] In the embodiment of the present application, static analysis and dynamic analysis of the installation package are performed separately, and the features obtained are the target static features and the target dynamic features. Step S110 and step S120 can be performed synchronously or asynchronously. When performed asynchronously, step S110 can be performed first and then step S120, or step S120 can be performed first and then step S110. The present application does not limit this.

[0060] In step S110, the installation package is statically parsed in the sandbox to obtain target static features. Static features refer to various information that can be directly extracted from the installation package, which may include installation package icon, installation package file, developer certificate, installation package text, software name, installation package name, etc.

[0061] Specifically, in some embodiments, the specific implementation of step S110 can refer to the following embodiments. Figure 3 The detailed description of step S110 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, step S110 may include the following steps:

[0062] The installation package is statically parsed in the sandbox to obtain preliminary static features, which include icons, texts, software names, file names, and file hashes.

[0063] Information gain is performed on the preliminary static features to obtain target static features.

[0064] In an embodiment of the present application, a preliminary static extraction is first performed on the installation package to obtain static features such as icons, texts, software names, file names, file hashes, etc. The aforementioned static features are all preliminary static features. Next, the preliminary static features are analyzed and screened to obtain the required target static features. The analysis and screening method can be specifically implemented through information gain.

[0065] The specific steps of analyzing and screening preliminary static features by means of information gain may include:

[0066] Determine the information gain value of each preliminary static feature;

[0067] A preset number of preliminary static features with the highest information gain values ​​are selected as the target static features.

[0068] That is, the information gain value of each preliminary static feature is first calculated, and then sorted from large to small, and the preliminary static features at the front are selected as the target static features according to a preset percentage or quantity.

[0069] In some embodiments, after selecting a preset number of preliminary static features with the highest information gain values, these preliminary static features need to be encoded before they can be converted into target static features so as to be input into the static analysis network for processing.

[0070] In step S120, the installation package is dynamically parsed in the sandbox to obtain target dynamic features. Dynamic features refer to feature data obtained after dynamic operations are performed on the installation package in the sandbox, that is, the data of the installation package and the program corresponding to the installation package after running, which may specifically include running screenshots and network requests, etc.

[0071] Specifically, in some embodiments, the specific implementation of step S120 can be found in Figure 4 . Figure 4 is based on Figure 3The detailed description of step S120 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, the target dynamic feature includes text information and screenshot information, and step S120 may include the following steps:

[0072] Step S121, dynamically parse the installation package in the sandbox to obtain preliminary dynamic features, wherein the preliminary dynamic features include network requests and running screenshots.

[0073] Step S122, parsing the network request to obtain text information.

[0074] Step S123, identifying the running screenshot to obtain screenshot information.

[0075] In an embodiment of the present application, a preliminary dynamic extraction is first performed on the installation package to obtain dynamic features such as network requests and running screenshots. The aforementioned dynamic features are all preliminary dynamic features. Next, the preliminary dynamic features are analyzed and identified to obtain the required target dynamic features.

[0076] In step S121, the method of dynamic analysis is to first run the installation package and the corresponding application in the sandbox, perform dynamic analysis, and obtain various parameters of the installation package and the corresponding application in the running state, mainly including screenshot information obtained by screenshot and text information obtained by capturing channel information.

[0077] In step S122, various communication information between the installation package and the corresponding application and the outside world in the running state is intercepted and analyzed to obtain corresponding files.

[0078] Specifically, in some embodiments, the specific implementation of step S122 can refer to the following embodiments. Figure 4 The detailed description of step S122 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, the network request includes a protocol address and a text label, the text information includes an address file, a path file, label information and a text file, and step S120 may include the following steps:

[0079] Parsing the protocol address to obtain an address file and a path file;

[0080] The text label is parsed to obtain label information and a text file.

[0081] In an embodiment of the present application, a network request includes a protocol address and a text label, and the text information includes an address file, a path file, label information, and a text file. Parsing the protocol address obtains the address file and the path file; parsing the text label obtains the label information and the text file. The parsing of the protocol address and the parsing of the text label can be performed synchronously or asynchronously, which is not limited in the present application.

[0082] The protocol address can be a URL address, and correspondingly, the address file can be a host file, and the path file can be a path file. The host file stores the mapping relationship between the domain name and the IP address, and its custom mapping can be mastered. According to the custom mapped IP address, it can be determined whether the installation package and the corresponding application have interacted with the black and gray addresses. The path file is a path file, and it can be determined whether there is any black and gray data interaction based on the path data recorded in it.

[0083] The text tag may be an html tag or an xml tag, and correspondingly, the tag information may be the tag name, and the text file may be the Chinese text content thereof.

[0084] In step S123, screenshots are taken of various interfaces of the installation package and the corresponding application in the running state to obtain corresponding screenshot information.

[0085] Specifically, in some embodiments, the specific implementation of step S123 can refer to the following embodiments. Figure 4 The detailed description of step S123 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, the screenshot information includes the screenshot content and the screenshot features, and step S123 may include the following steps:

[0086] Perform image recognition on the running screenshot to obtain screenshot content.

[0087] Perform image hashing on the running screenshot to obtain screenshot features.

[0088] In the embodiment of the present application, the text and image content in the screenshot are identified by image recognition to obtain the screenshot content. The running screenshot is hashed by a hash algorithm to obtain the screenshot feature, which is a string of hash strings.

[0089] It should be noted that image recognition and image hashing can be performed synchronously or asynchronously, and this application does not limit this.

[0090] Specifically, the hash algorithm may adopt an image mean hash algorithm, an image perceptual hash algorithm, or an image difference hash algorithm, which is not limited in this application.

[0091] In step S200, the dynamic classification network may be an Extreme Gradient Boosting (XGBoost) network, which is an implementation of a boosting algorithm and has excellent effects, is easy to use, and is fast for classification or regression problems.

[0092] The specific training method of the dynamic classification network may include: obtaining a target dynamic feature sample set, wherein the target dynamic feature sample set includes multiple target dynamic feature samples, and each of the target dynamic feature samples includes a corresponding dynamic classification label; inputting the target dynamic feature samples in the target dynamic feature sample set into the dynamic classification network to obtain a dynamic classification result output by the dynamic classification network; if the corresponding dynamic classification labels of the dynamic classification results obtained after the target dynamic feature samples in the target dynamic feature sample set that do not exceed a predetermined dynamic proportion are input into the dynamic classification network are consistent, then adjusting the coefficient of the dynamic classification network; if the corresponding dynamic classification labels of the dynamic classification results obtained after the target dynamic feature samples in the target dynamic feature sample set that exceed a predetermined dynamic proportion are input into the dynamic classification network are consistent, the training ends.

[0093] In step S300, the static classification network may also be an XGBoost network, and the specific training method of the static classification network may include: obtaining a target static feature sample set, wherein the target static feature sample set includes multiple target static feature samples, and each of the target static feature samples includes a corresponding static classification label; inputting the target static feature samples in the target static feature sample set into the static classification network to obtain a static classification result output by the static classification network; if only target static feature samples in the target static feature sample set that do not exceed a predetermined static proportion are input into the static classification network, and the corresponding static classification labels of the static classification results obtained are consistent, then adjusting the coefficient of the static classification network; if target static feature samples in the target static feature sample set that exceed a predetermined static proportion are input into the static classification network, and the corresponding static classification labels of the static classification results obtained are consistent, the training ends.

[0094] It should be noted that step S200 and step S300 may be executed synchronously or asynchronously. Step S200 may be executed first and then step S300, or step S300 may be executed first and then step S200. This application does not limit this.

[0095] At the same time, in addition to independent training, static classification networks and dynamic classification networks can also be jointly trained. For joint training methods, please refer to Figure 5 A proposed training method for installation package classification model.

[0096] In step S400, the final classification result may be obtained by executing a preset rule or by inputting a joint decision network, which is not limited in the present application.

[0097] In an embodiment of the present application, the three classification results of the final classification result, the static classification result and the dynamic classification result may include the specific black and gray product type of the installation package, and may also include the probability that the installation package is of each type of black and gray product type.

[0098] Among them, the types of black and gray industries can include white industry, gray industry and black industry, etc. White industry means that the installation package and the corresponding application do not involve any cybercrime and are installation packages that can be used normally; gray industry means that the installation package and the corresponding application have the risk of involving cybercrime and are installation packages that need to be managed and controlled; black industry means that the installation package and the corresponding application clearly involve cybercrime and are installation packages that need to be strictly supervised.

[0099] Gray industries and black industries can be further divided into types involving gangs, violence, terrorism, pornography, gambling, drugs and fraud.

[0100] Therefore, the three classification results of the final classification result, the static classification result and the dynamic classification result can be in the form of: the installation package is a white product, the installation package is a gray product involving gambling, the installation package is a black product involving gangs, etc.; it can also be that the probability that the installation package is a white product is 5%, the probability that the installation package is a gray product involving gambling is 69%, and the probability that the installation package is a black product involving fraud is 26%.

[0101] Specifically, in some embodiments, the specific implementation of step S400 can refer to the following embodiments. Figure 2 The detailed description of step S400 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, step S400 may include the following steps:

[0102] If the dynamic classification result and the static classification result are both the same classification result, the same classification result is used as the final classification result.

[0103] If the dynamic classification result and the static classification result are not the same classification result, the final classification result is set as a pending classification, and the pending classification may include the dynamic classification result and the static classification result for manual reference, and the final classification result is determined manually.

[0104] After the final classification result is manually determined, the parameters of the static classification network can be updated according to the manual classification result and the static classification result, and the parameters of the dynamic classification network can be updated according to the manual classification result and the dynamic classification result.

[0105] If the classification result is the black-gray product type of the installation package, as long as the black-gray product type of the dynamic classification result and the black-gray product type of the static classification result are the same black-gray product type, the dynamic classification result and the static classification result are the same classification result, and the final classification result is the same black-gray product type.

[0106] For example, in one embodiment, the static classification result is that the installation package is related to gambling and illegal industries, and the dynamic classification result is that the installation package is related to gambling and illegal industries. If the static classification result and the dynamic classification result are consistent, the final classification result is related to gambling and illegal industries.

[0107] For example, in another embodiment, the static classification result is that the installation package is related to gambling and the dynamic classification result is that the installation package is related to fraud. If the static classification result and the dynamic classification result are inconsistent, the final classification result is set as pending classification.

[0108] If the classification result is the probability that the installation package is of various types of black and gray products, then not only is it necessary that the black and gray product type with the highest probability in the two classification results be the same black and gray product type, but the probability of the black and gray product type is also required to be greater than a predetermined probability threshold in order to determine that the dynamic classification result and the static classification result are the same classification result, and the final classification result can be determined as the same classification result.

[0109] Specifically, in some embodiments, the specific implementation of step S400 can refer to the following embodiments. Figure 2 The detailed description of step S400 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, step S400 may include the following steps:

[0110] The dynamic classification result and the static classification result are input into a joint decision network to obtain a final classification result.

[0111] The joint judgment network can determine the final classification result by calculating the residuals of the dynamic classification results and the static classification results. Specifically, if the classification result is the probability that the installation package is of various types of black and gray products, the residuals of each classification probability can be calculated one by one, and then the final probability of each classification can be determined based on the residuals. After normalization, it is the final classification result.

[0112] The joint decision network can be a convolutional neural network, a deep neural network, a residual attention network, etc. The training method of the joint decision network specifically includes: obtaining a classification result sample pair set, the classification result sample pair set includes multiple classification result sample pairs, each classification result sample pair includes a static classification result and a dynamic classification result, and each of the classification result sample pairs includes a corresponding final classification label; inputting the classification result sample pairs in the classification result sample pair set into the joint decision network to obtain the final classification result output by the joint decision network; if the classification result sample pairs in the classification result sample pair set that do not exceed a predetermined proportion are input into the joint decision network, and the corresponding final classification labels of the final classification results obtained are consistent, then adjusting the coefficient of the joint decision network; if the classification result sample pairs in the classification result sample pair set that exceed a predetermined proportion are input into the joint decision network, and the corresponding final classification labels of the final classification results obtained are consistent, the training ends.

[0113] Figure 5 A flowchart of a method for training an installation package classification model according to an embodiment of the present application is shown. The method for training an installation package classification model can be executed by a server, which can be Figure 1 Refer to the server shown in Figure 5 As shown, the training method of the installation package classification model at least includes:

[0114] Step S000, obtaining an installation package sample set, wherein the installation package sample set includes a plurality of installation package samples, each installation package sample is marked with a corresponding installation package classification label;

[0115] Step S010, parsing the installation package sample in the sandbox to obtain dynamic features and static features of the sample;

[0116] Step S020, inputting the sample dynamic features into a dynamic classification network to obtain a sample dynamic classification result;

[0117] Step S030, inputting the sample static features into a static classification network to obtain a sample static classification result;

[0118] Step S040, determining a final classification result of the sample according to the dynamic classification result of the sample and the static classification result of the sample;

[0119] Step S050, adjusting parameters of the installation package classification model according to the final classification result of the sample and the installation package classification label until the training end condition is met, thereby obtaining a trained installation package classification model.

[0120] In the embodiments of the present application, Figure 2The specific steps of the installation package classification method shown are packaged into an installation package classification model, such as Figure 6 As shown in FIG. 1 , after inputting the installation package into the installation package classification model, the corresponding installation package classification, i.e., the final classification result, is obtained. Then the installation package classification model can be used Figure 5 The method shown is used to train the model. In this embodiment, a joint training method is used to train all model networks together. Each time a final classification result of a sample is obtained, the parameters of each model network in the installation package classification model are updated, such as the dynamic classification network and the static classification network. If it is determined that the sample classification result uses a joint decision network, the parameters of the joint decision network are also updated.

[0121] In step S050, the parameters of the installation package classification model can be adjusted according to the loss function. Specifically, in some embodiments, the specific implementation of step S050 can refer to the following embodiments. Figure 5 The detailed description of step S050 in the installation package classification method shown in the corresponding embodiment, in the installation package classification method, step S050 may include the following steps:

[0122] Obtaining a loss function according to the final classification result of the sample and the classification label of the installation package;

[0123] The parameters of the installation package classification model are adjusted according to the loss function until the training end condition is met, thereby obtaining a trained installation package classification model.

[0124] The training end condition may be that the loss function converges, or the loss function is less than a predetermined loss function threshold.

[0125] There may be multiple loss functions mentioned above. In some embodiments, the loss function includes a first loss function, a second loss function and a third loss function.

[0126] Among them, the first loss function is determined according to the dynamic classification results of the samples and the classification labels of the installation packages, and can be used to update the parameters of the dynamic classification network; the second loss function is determined according to the static classification results of the samples and the classification labels of the installation packages, and can be used to update the parameters of the static classification network; the third loss function is determined according to the final classification results of the samples and the classification labels of the installation packages, and can be used to update the parameters of the dynamic classification network, the static classification network and the joint judgment network.

[0127] Figure 7 The following is a schematic diagram showing the structure of an installation package classification device provided in an embodiment of the present application. Figure 7As shown, the installation package classification device 700 may at least include: a sandbox analysis module 710, a dynamic classification module 720, a static classification module 730, and a final classification module 740. Among them:

[0128] The sandbox parsing module 710 is used to parse the installation package in the sandbox to obtain target dynamic features and target static features;

[0129] A dynamic classification module 720, used for inputting the dynamic features into a dynamic classification network to obtain a dynamic classification result;

[0130] A static classification module 730, used for inputting the static features into a static classification network to obtain a static classification result;

[0131] The final classification module 740 is used to determine the final classification result according to the dynamic classification result and the static classification result.

[0132] In a possible implementation, the sandbox parsing module 710 specifically includes: a static parsing submodule, which is used to statically parse the installation package in the sandbox to obtain target static features; and a dynamic parsing submodule, which is used to dynamically parse the installation package in the sandbox to obtain target dynamic features.

[0133] In a possible implementation, the static parsing submodule specifically includes: a static feature unit, used to statically parse the installation package in a sandbox to obtain preliminary static features, wherein the preliminary static features include icons, texts, software names, file names, and file hashes; an information gain unit, used to perform information gain on the preliminary static features to obtain target static features.

[0134] In a possible implementation, the target dynamic features include text information and screenshot information, and the dynamic parsing submodule specifically includes: a dynamic feature unit, used to dynamically parse the installation package in a sandbox to obtain preliminary dynamic features, and the preliminary dynamic features include network requests and running screenshots; a request parsing unit, used to parse the network request to obtain text information; an image parsing unit, used to identify the running screenshot to obtain screenshot information.

[0135] In a possible implementation, the network request includes a protocol address and a text label, the text information includes an address file, a path file, label information and a text file, and the request parsing unit is specifically used to execute: parsing the protocol address to obtain an address file and a path file; parsing the text label to obtain label information and a text file.

[0136] In a possible implementation, the screenshot information includes screenshot content and screenshot features, and the image parsing unit is specifically used to execute: performing image recognition on the running screenshot to obtain the screenshot content; performing image hashing on the running screenshot to obtain the screenshot features.

[0137] Figure 8 The structural diagram of a training device for an installation package classification model provided in an embodiment of the present application is exemplarily shown. Figure 7 As shown, the training device 800 for the installation package classification model may at least include: a sample acquisition module 810, a sample analysis module 820, a first classification module 830, a second classification module 840, a sample result module 850, and a parameter adjustment module 860. Among them:

[0138] The sample acquisition module 810 is used to acquire an installation package sample set, wherein the installation package sample set includes a plurality of installation package samples, each of which is marked with a corresponding installation package classification label;

[0139] The sample analysis module 820 is used to analyze the installation package sample in the sandbox to obtain the sample dynamic features and sample static features;

[0140] The first classification module 830 is used to input the sample dynamic features into a dynamic classification network to obtain a sample dynamic classification result;

[0141] The second classification module 840 is used to input the sample static features into a static classification network to obtain a sample static classification result;

[0142] A sample result module 850 is used to determine a final classification result of the sample according to the dynamic classification result of the sample and the static classification result of the sample;

[0143] The parameter adjustment module 860 is used to adjust the parameters of the installation package classification model according to the final classification result of the sample and the installation package classification label until the training end condition is met to obtain the trained installation package classification model.

[0144] The embodiment of the present application obtains the corresponding dynamic features and static features by dynamically parsing and statically parsing the installation package in a sandbox, and then models and analyzes the dynamic features and static features respectively to reduce the difficulty of modeling, obtain the corresponding classification results, i.e., the dynamic classification results and the static classification results, and also enhance the interpretability of the model and the flexibility of application. Finally, based on the dynamic classification results and the static classification results, the final classification results are obtained comprehensively, and the final classification results combine the results of dynamic analysis and the results of static analysis, so the classification features are more accurate.

[0145] It should be noted that when the installation package classification device provided in the above embodiment executes the installation package classification method, it only uses the division of the above functional modules as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the wireless screen projection connection device and the wireless screen projection connection method embodiment provided in the above embodiment belong to the same concept. The implementation process is detailed in the method embodiment, which will not be repeated here.

[0146] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0147] See also Fig. 9 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig. 9 As shown, the electronic device 900 may include: at least one processor 901 , at least one network interface 904 , a user interface 903 , a memory 905 , and at least one communication bus 902 .

[0148] The communication bus 902 is used to realize the connection and communication between these components.

[0149] The user interface 903 may include a display screen (Display) and a camera (Camera), and the optional user interface 903 may also include a standard wired interface and a wireless interface.

[0150] The network interface 904 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0151] Among them, the processor 901 may include one or more processing cores. The processor 901 uses various interfaces and lines to connect various parts of the entire electronic device 900, and executes various functions and processes data of the electronic device 60 by running or executing instructions, programs, code sets or instruction sets stored in the memory 905, and calling data stored in the memory 905. Optionally, the processor 901 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 901 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 901, and it can be implemented separately through a chip.

[0152] Among them, the memory 905 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 905 includes a non-transitory computer-readable storage medium. The memory 905 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 905 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 905 may also be optionally at least one storage device located away from the aforementioned processor 901. As Fig. 9 As shown, the memory 905 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program.

[0153] exist Fig. 9In the electronic device 900 shown, the user interface 903 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 901 can be used to call the application stored in the memory 905, and specifically perform the following operations: parse the installation package in the sandbox to obtain target dynamic features and target static features; input the dynamic features into the dynamic classification network to obtain dynamic classification results; input the static features into the static classification network to obtain static classification results; determine the final classification result based on the dynamic classification results and the static classification results, and the final classification result, the dynamic classification result, and the static classification result include the black and gray product category of the installation package.

[0154] In a possible implementation, when the processor parses the installation package in the sandbox to obtain the target dynamic features and the target static features, it specifically executes: statically parsing the installation package in the sandbox to obtain the target static features; dynamically parsing the installation package in the sandbox to obtain the target dynamic features.

[0155] In a possible implementation, when the processor statically parses the installation package in the sandbox to obtain target static features, the processor specifically executes: statically parses the installation package in the sandbox to obtain preliminary static features, where the preliminary static features include icons, texts, software names, file names, and file hashes; and performs information gain on the preliminary static features to obtain target static features.

[0156] In a possible implementation, the target dynamic features include text information and screenshot information. When the processor dynamically parses the installation package in the sandbox and obtains the target dynamic features, it specifically executes: dynamically parses the installation package in the sandbox to obtain preliminary dynamic features, and the preliminary dynamic features include network requests and running screenshots; parses the network request to obtain text information; and identifies the running screenshot to obtain screenshot information.

[0157] In a possible implementation, the network request includes a protocol address and a text label, and the text information includes an address file, a path file, label information, and a text file. When the processor parses the network request and obtains the text information, it specifically executes: parsing the protocol address to obtain the address file and the path file; parsing the text label to obtain the label information and the text file.

[0158] In a possible implementation, the screenshot information includes screenshot content and screenshot features. When the processor identifies the running screenshot and obtains the screenshot information, it specifically executes: performing image recognition on the running screenshot to obtain the screenshot content; performing image hashing on the running screenshot to obtain the screenshot features.

[0159] See also Fig.10 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig.10 As shown, the electronic device 1000 may include: at least one processor 1001 , at least one network interface 1004 , a user interface 1003 , a memory 1005 , and at least one communication bus 1002 .

[0160] The communication bus 1002 is used to realize the connection and communication between these components.

[0161] The user interface 1003 may include a display screen (Display) and a camera (Camera), and the optional user interface 1003 may also include a standard wired interface and a wireless interface.

[0162] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0163] Among them, the processor 1001 may include one or more processing cores. The processor 1001 uses various interfaces and lines to connect various parts of the entire electronic device 1000, and executes various functions and processes data of the electronic device 1000 by running or executing instructions, programs, code sets or instruction sets stored in the memory 1005, and calling data stored in the memory 1005. Optionally, the processor 1001 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 1001 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 1001, and it can be implemented separately through a chip.

[0164] Among them, the memory 1005 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 1005 includes a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 1005 may optionally be at least one storage device located away from the aforementioned processor 1001. As Fig.10 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module and an application program.

[0165] exist Fig.10 In the electronic device 1000 shown, the user interface 1003 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 1001 can be used to call the application stored in the memory 1005 and specifically perform the following operations:

[0166] Acquire an installation package sample set, wherein the installation package sample set includes multiple installation package samples, and each installation package sample is marked with a corresponding installation package classification label; parse the installation package samples in a sandbox to obtain sample dynamic features and sample static features; input the sample dynamic features into a dynamic classification network to obtain sample dynamic classification results; input the sample static features into a static classification network to obtain sample static classification results; determine the sample final classification results based on the sample dynamic classification results and the sample static classification results; adjust the parameters of the installation package classification model based on the sample final classification results and the installation package classification label until the training end conditions are met, and obtain the trained installation package classification model.

[0167] The embodiment of the present application obtains the corresponding dynamic features and static features by dynamically parsing and statically parsing the installation package in a sandbox, and then models and analyzes the dynamic features and static features respectively to reduce the difficulty of modeling, obtain the corresponding classification results, i.e., the dynamic classification results and the static classification results, and also enhance the interpretability of the model and the flexibility of application. Finally, based on the dynamic classification results and the static classification results, the final classification results are obtained comprehensively, and the final classification results combine the results of dynamic analysis and the results of static analysis, so the classification features are more accurate.

[0168] The embodiment of the present application also provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a computer or a processor, the computer or the processor executes the above-mentioned Figure 2-Figure 6 One or more steps in the illustrated embodiment. If the components of the electronic device described above are implemented in the form of software functional units and sold or used as independent products, they can be stored in the computer-readable storage medium.

[0169] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted by the computer-readable storage medium. The computer instructions can be transmitted from a website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, a data center, etc. that contains one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0170] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The aforementioned storage medium includes: various media that can store program codes, such as system memory (Read Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk. In the absence of conflict, the technical features in this embodiment and the implementation scheme can be combined arbitrarily.

[0171] The embodiments described above are merely preferred embodiments of the present application and are not intended to limit the scope of the present application. Without departing from the design spirit of the present application, various modifications and improvements made to the technical solutions of the present application by ordinary technicians in this field should fall within the protection scope determined by the claims of the present application.

Claims

1. A method for classifying installation packages, characterized in that: The installation package classification method comprises: Parse the installation package in the sandbox to obtain the target dynamic features and target static features; Inputting the target dynamic features into a dynamic classification network to obtain a dynamic classification result; Inputting the target static features into a static classification network to obtain a static classification result; A final classification result is determined according to the dynamic classification result and the static classification result.

2. The method according to claim 1, characterized in that The installation package is parsed in the sandbox to obtain the target dynamic features and target static features, specifically including: Performing static analysis on the installation package in the sandbox to obtain target static features; The installation package is dynamically parsed in the sandbox to obtain target dynamic features.

3. The method according to claim 2, characterized in that The static analysis of the installation package in the sandbox to obtain the target static features specifically includes: Perform static analysis on the installation package in the sandbox to obtain preliminary static features, where the preliminary static features include icons, texts, software names, file names, and file hashes; Information gain is performed on the preliminary static features to obtain target static features.

4. The method according to claim 2, characterized in that The target dynamic features include text information and screenshot information. The dynamic analysis of the installation package in the sandbox to obtain the target dynamic features specifically includes: Dynamically analyzing the installation package in the sandbox to obtain preliminary dynamic features, wherein the preliminary dynamic features include network requests and running screenshots; Parsing the network request to obtain text information; The running screenshot is identified to obtain screenshot information.

5. The method according to claim 4, characterized in that The network request includes a protocol address and a text label, the text information includes an address file, a path file, label information and a text file, and the network request is parsed to obtain text information, specifically including: Parsing the protocol address to obtain an address file and a path file; The text label is parsed to obtain label information and a text file.

6. A training method for an installation package classification model, characterized in that: The installation package classification model includes a dynamic classification network and a static classification network, and the training method of the installation package classification model includes: Obtaining an installation package sample set, wherein the installation package sample set includes a plurality of installation package samples, each installation package sample being marked with a corresponding installation package classification label; Parsing the installation package sample in the sandbox to obtain dynamic features and static features of the sample; Inputting the sample dynamic features into a dynamic classification network to obtain a sample dynamic classification result; Inputting the sample static features into a static classification network to obtain a sample static classification result; Determine a final classification result of the sample according to the dynamic classification result of the sample and the static classification result of the sample; According to the final classification result of the sample and the installation package classification label, the parameters of the installation package classification model are adjusted until the training end condition is met, thereby obtaining a trained installation package classification model.

7. An installation package classification device, characterized in that: The installation package classification device comprises: The sandbox parsing module is used to parse the installation package in the sandbox to obtain the target dynamic features and target static features; A dynamic classification module, used for inputting the dynamic features into a dynamic classification network to obtain a dynamic classification result; A static classification module, used for inputting the static features into a static classification network to obtain a static classification result; The final classification module is used to determine the final classification result according to the dynamic classification result and the static classification result.

8. A training device for an installation package classification model, characterized in that: The installation package classification model includes a dynamic classification network and a static classification network, and the training device of the installation package classification model includes: A sample acquisition module is used to acquire an installation package sample set, wherein the installation package sample set includes multiple installation package samples, each installation package sample is marked with a corresponding installation package classification label; A sample analysis module, used to analyze the installation package sample in the sandbox to obtain dynamic features and static features of the sample; A first classification module, used for inputting the sample dynamic features into a dynamic classification network to obtain a sample dynamic classification result; The second classification module is used to input the sample static features into a static classification network to obtain a sample static classification result; A sample result module, used to determine the final classification result of the sample according to the dynamic classification result of the sample and the static classification result of the sample; The parameter adjustment module is used to adjust the parameters of the installation package classification model according to the final classification result of the sample and the installation package classification label until the training end condition is met to obtain the trained installation package classification model.

9. An electronic device, characterized in that: include: A processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the method steps as claimed in any one of claims 1 to 6.

10. A computer storage medium, characterized in that: The computer storage medium stores a plurality of instructions, which are suitable for being loaded by a processor and executing the method steps according to any one of claims 1 to 6.

Citation Information

Patent Citations

  • Method for detecting malicious phishing application for Android platform

    CN105095751A

  • APP detection method, apparatus and device, and storage medium

    CN112685737A

  • Malicious code detection method for multi-dimensional data decision fusion

    CN115828239A

  • Adaptive filtering of malware using machine-learning based classification and sandboxing

    US20220067146A1