Machine learning method and system based on privacy protection enhancement

Abstract

The invention discloses a machine learning method and system based on privacy protection enhancement, and relates to the technical field of data processing. The method comprises the steps of monitoring real-time data streams of a plurality of data sources based on real-time input data, analyzing relationships among various data, drawing a data association network diagram, evaluating information leakage risks and privacy sensitivity of a plurality of data points by analyzing interconnection density and path distribution of nodes in the network diagram, and generating sensitivity evaluation indexes. According to the method, the security and privacy protection level in the data processing process are improved by identifying the risk points in the data in real time and specifically adjusting the noise level of the data points, and the encryption processing efficiency and security are improved by comprehensively considering the data processing efficiency, security level requirements and computing resource consumption; the stability of encrypted data is tested by simulating network attack conditions, the capability of coping with network attacks is enhanced, input data is enabled to support various data analysis tasks, and the privacy and security of the data are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a machine learning method and system based on enhanced privacy protection. Background Art

[0002] The technical field of data processing involves processes from basic data input / output processing to complex information processing, including data cleaning, i.e., eliminating duplicate and correcting erroneous data points, data transformation, i.e., converting data from one format or structure to another to suit different analysis tools or technical requirements, data loading, i.e., storing the processed data into a data warehouse or other storage systems for subsequent use, combined with data indexing and query optimization to ensure the efficiency and accuracy of data retrieval, aiming to extract valuable information and support decision-making, including aspects such as database management, data encoding, data encryption, and data transformation.

[0003] Among them, the machine learning method refers to the process of using statistics, mathematics, and algorithms to learn from data and make predictions or decisions, realizing the development and optimization of machine learning algorithms, involving automatically adjusting model parameters through algorithms to adapt to data changes, including the application of supervised learning, unsupervised learning, and semi-supervised learning algorithms, such as decision trees, support vector machines, and neural networks, processing data through algorithm logic and mathematical models, identifying patterns and correlations, and ensuring data security and privacy protection during the processing to complete the target learning tasks.

[0004] Traditional methods have deficiencies in privacy protection and data security. They rely on static data cleaning and transformation processes, lack real-time dynamic data monitoring and immediate risk assessment, and are not effective when dealing with multi-data sources and high-speed changing information flows. In terms of data privacy, conventional data processing technologies fail to fully consider the complex correlations between data, resulting in information leakage or privacy infringement. The lack of effective real-time data flow monitoring will lead to the inability to detect and respond to data leakage risks in a timely manner. In the application of homomorphic encryption, the consumption of computing resources and processing efficiency are not optimized, which not only reduces the operating efficiency of the system, leading to a decline in response speed and affecting system performance. Summary of the Invention

[0005] To solve the technical problems of insufficient privacy protection and data security existing in the prior art, embodiments of the present invention provide a machine learning method and system based on enhanced privacy protection. The technical solutions are as follows:

[0006] On the one hand, a machine learning method based on enhanced privacy protection is provided, and the method includes:

[0007] S1. Based on real-time input data, monitor the real-time data streams of multiple data sources, analyze the relationships among various types of data, draw a data association network diagram, and evaluate the information leakage risk and privacy sensitivity of multiple data points by analyzing the interconnection density and path distribution of nodes in the network diagram, so as to generate sensitivity evaluation indicators;

[0008] S2. According to the sensitivity evaluation indicators and the privacy sensitivity of multiple data points, analyze the privacy requirement levels of multiple data points, and adjust the type and intensity of noise added to the target data point to generate a data noise addition record;

[0009] S3. Based on the data noise addition record, according to the data volume and security level requirements of the input data, match the homomorphic encryption configuration by analyzing the computing resource consumption, encryption intensity, and processing efficiency of various homomorphic encryption configurations, and generate data homomorphic encryption parameters;

[0010] S4. Based on the data homomorphic encryption parameters, by simulating various network attack conditions, analyze the response status and stability of the encrypted data under various loads and attacks, evaluate the actual performance of the target encryption parameters and make adjustments, and generate an encryption performance analysis result;

[0011] S5. Based on the encryption performance analysis result, perform homomorphic encryption processing on various input data, monitor the encrypted data stream in real time, analyze the compatibility and relevance among the data, and evaluate the consistency and integrity of the data in the encrypted state to generate an input data encryption record.

[0012] On the other hand, a machine learning system based on enhanced privacy protection is provided. This system is applied to the machine learning method based on enhanced privacy protection, and this system includes:

[0013] A data stream monitoring module, which is used to monitor the real-time data streams of multiple data sources based on real-time input data, analyze the relationships among various types of data, draw a data association network diagram, and evaluate the information leakage risk and privacy sensitivity of multiple data points by analyzing the interconnection density and path distribution of nodes in the network diagram, so as to generate sensitivity evaluation indicators;

[0014] A noise management module, which is used to analyze the privacy requirement levels of multiple data points according to the sensitivity evaluation indicators and the privacy sensitivity of multiple data points, and adjust the type and intensity of noise added to the target data point to generate a data noise addition record;

[0015] An encryption parameter configuration module, which is used to match the homomorphic encryption configuration based on the data noise addition record, according to the data volume and security level requirements of the input data, by analyzing the computing resource consumption, encryption intensity, and processing efficiency of various homomorphic encryption configurations, and generate data homomorphic encryption parameters;

[0016] A network attack simulation module, which is used to analyze the response status and stability of encrypted data under various loads and attack conditions based on the data homomorphic encryption parameters by simulating various network attack conditions, evaluate the actual performance of the target encryption parameters and make adjustments, and generate encryption performance analysis results;

[0017] An encrypted data monitoring module, which is used to perform homomorphic encryption processing on various input data based on the encryption performance analysis results, monitor the encrypted data stream in real time, analyze the compatibility and correlation between data, and evaluate the consistency and integrity of data in the encrypted state, and generate input data encryption records.

[0018] The beneficial effects brought by the technical solutions provided by the embodiments of the present invention at least include:

[0019] By real-time identifying risk points in data, targeted adjustment of the noise level of data points is carried out to improve the security and privacy protection level in the data processing process. By comprehensively considering data processing efficiency, security level requirements, and computing resource consumption, the efficiency and security of encryption processing are improved. By simulating network attack conditions to test the stability of encrypted data, the ability to respond to network attacks is enhanced, enabling input data to support various data analysis tasks and ensuring the privacy and security of data. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 is a flowchart of a machine learning method based on enhanced privacy protection provided by an embodiment of the present invention;

[0022] Figure 2 is a block diagram of a machine learning system based on enhanced privacy protection provided by an embodiment of the present invention;

[0023] Figure 3 is a schematic structural diagram of a machine learning device based on enhanced privacy protection provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The following will describe the technical solutions in the present invention with reference to the drawings.

[0025] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to give examples, illustrations or explanations. Any embodiment or design described as an "example" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0026] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same.

[0027] In the embodiments of the present invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0028] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0029] The embodiments of the present invention provide a machine learning method based on enhanced privacy protection, such as Figure 1 shown in the flowchart of the machine learning method based on enhanced privacy protection. The processing flow of this method may include the following steps:

[0030] S1. Based on real-time input data, monitor the real-time data streams of multiple data sources, analyze the relationships between various data, draw a data association network diagram, and evaluate the information leakage risk and privacy sensitivity of multiple data points by analyzing the interconnection density and path distribution of nodes in the network diagram, and generate sensitivity evaluation indicators.

[0031] Optionally, the sensitivity evaluation indicators are specifically the data point connectivity, data path distribution, and data point risk level. The data noise addition record includes the data point sensitivity, privacy requirement level, and noise addition parameters. The data homomorphic encryption parameters specifically refer to the data processing efficiency of various configurations, the data security levels of various encryption configurations, and the encryption configuration matching records. The encryption performance analysis results include the data response status information, stability test results, and encryption parameter adjustment records. The input data encryption record is specifically the data stream monitoring record, data compatibility analysis results, and data integrity evaluation results.

[0032] Optionally, the specific operation steps of S2 include the following S101-S103:

[0033] S101. Based on real-time input data, real-time monitor the real-time data streams of multiple data sources, record data changes, and generate a data stream interaction analysis result by analyzing the mutual relationships between the data.

[0034] In a feasible implementation, signal and data changes sent by each data source are captured in real time, log records are made for the target data stream, and the record content involves the timestamp, source, type, and change details of the data. Analyze the data in the records, associate the interaction patterns and data transmission links between different data sources, use the graph theory analysis model, and apply the network flow analysis algorithm to process and analyze the dependencies and correlations between data streams. Each data change triggers an analysis process, which tracks the evolution of the data stream and reveals potential data dynamic relationships. Through this data monitoring and dynamic analysis, a data stream interaction analysis result is generated, which records the interaction status and change conditions of the data stream at different time points, providing a basis for data processing and analysis.

[0035] S102. Based on the data stream interaction analysis result, draw a data association network diagram according to the mutual relationships between the data, and generate data network structure information.

[0036] In a feasible implementation, each data point in the data stream analysis result is regarded as a node in the network, and the interaction relationship between data streams is transformed into a connection line between nodes. The strength and direction of the connection line reflect the degree of dependence between data streams and the master-slave relationship of data transmission. Use graphical software tools, including Gephi or Graphviz, to draw the network diagram. During the drawing process, calculate network topology characteristics such as the degree, centrality, and clustering coefficient of nodes. The target calculation helps to reveal the core data sources and potential information flow bottlenecks in the data network. After the drawing is completed, data network structure information is generated, and the target information is presented in the form of a visual chart, providing an intuitive structure reference and analysis basis for further privacy risk assessment.

[0037] S103. According to the data network structure information, identify the information leakage risks of multiple nodes in the network by calculating the interconnection density and path distribution of multiple nodes, evaluate the privacy sensitivity of multiple data points, and generate sensitivity evaluation indicators.

[0038] Optionally, the specific formula for evaluating the privacy sensitivity of multiple data points is as follows in formula (1):

[0039] (1)

[0040] Wherein, is the total number of nodes in the network, is the number of direct connections of the i-th node to other nodes, is the average value of the degrees of all nodes in the network, is the standard deviation of the node degrees in the network, is the clustering coefficient of the i-th node, represents the maximum clustering coefficient in the network, i is the index of the node, and S is the sensitivity score of each node.

[0041] The above formula (1) is explained by way of examples as follows:

[0042] The formula is used to calculate the privacy sensitivity scores of multiple nodes in the network, and the results are used to evaluate the information leakage risk of the nodes;

[0043] Meanings and set values of the parameters:

[0044] is the total number of nodes in the network, assumed to be 50, which reflects the scale of the network;

[0045] is the degree of the -th node, assumed to be 12, which reflects the

[0046] direct connection number of the node with other nodes;

[0047] is the standard deviation of the node degrees in the network, assumed to be 2, which is used to measure the dispersion degree of the node degrees;

[0048] is the clustering coefficient of the i-th node, assumed to be 0.5, which reflects the closeness among the neighbors of node i;

[0049] is the maximum clustering coefficient in the network, assumed to be 0.8, which is used for normalizing the comparison of the clustering coefficients.

[0050] Substitute the parameters into the formula for calculation:

[0051] ;

[0052] ;

[0053] ;

[0054] ;

[0055] The result Indicates that the privacy sensitivity score of the node is low, indicating that the risk of information leakage of the target node is relatively small compared to other nodes in the network. The calculation process is used to identify the privacy sensitivity of multiple data points and adjust the noise addition parameters.

[0056] S2. According to the sensitivity evaluation index and the privacy sensitivity of multiple data points, analyze the privacy requirement levels of multiple data points, and adjust the type and intensity of the noise added to the target data point to generate a data noise addition record.

[0057] Optionally, the specific operation steps of S2 include the following S201 - S203:

[0058] S201. According to the sensitivity evaluation index and the privacy sensitivity of multiple data points, evaluate the privacy protection requirements of multiple data points to generate a privacy requirement analysis result.

[0059] In a feasible implementation, use the sensitivity score to determine the privacy level of the data point, adopt a data classification algorithm, such as decision tree analysis, to classify the data points, convert the sensitivity score into privacy protection requirements, such as high, medium, and low levels. During the process, each data point is assigned a different protection level according to its role in the network and the content of sensitive data. Data points with high sensitivity receive more stringent protection measures. After the analysis is completed, the system generates a privacy requirement analysis result, listing the privacy protection requirements and recommended protection strategies for each data point, providing accurate guidance for data protection operations.

[0060] S202. Based on the privacy requirement analysis result, according to the type of data point, analyze the type of noise that needs to be added to multiple data to generate a noise type analysis result.

[0061] In a feasible implementation, adopt a probability model and noise pattern analysis technology, including Laplace noise and Gaussian noise addition models, to determine the most suitable noise type for various data sensitivity levels. The operation involves evaluating the balance requirement between the potential information leakage risk of each type of data and the original accuracy of the data, selecting the noise type that can effectively improve data privacy while having the least impact on data usability. During the analysis process, the usage frequency of the data and the privacy expectations of the users are also taken into account to ensure that the selected noise addition strategy is both practical and meets the user's needs, generating a noise type analysis result.

[0062] S203. Based on the noise type analysis result, calculate the noise intensity that needs to be added to multiple data points, and add noise to the target data point to generate a data noise addition record.

[0063] In a feasible implementation, differential privacy technology is used to determine the noise value. The calculation takes into account the privacy requirement level of the data points and the characteristics of the selected noise type. Mathematical optimization methods, such as linear programming, are adopted to ensure that the noise addition can achieve the expected privacy protection effect and retain the validity and accuracy of the data as much as possible. During the calculation process, the standard deviation, mean, and distribution of the noise in the dataset are calculated and adjusted in detail to adapt to different data application scenarios and privacy protection standards. After the target calculation is completed, the system adds noise to the specified data points, records the details of each addition operation, and generates a data noise addition record. The target record details the type, intensity, and addition time of the noise, providing a reliable record for subsequent data auditing and privacy verification.

[0064] S3. Based on the data noise addition record, according to the data volume and security level requirements of the input data, by analyzing the computing resource consumption, encryption intensity, and processing efficiency of multiple homomorphic encryption configurations, match the homomorphic encryption configuration and generate data homomorphic encryption parameters.

[0065] Optionally, the specific operation steps of S3 include the following S301 - S303:

[0066] S301. Based on the data noise addition record, identify the data volume and security level requirements of the input data and generate security requirement information.

[0067] Data analysis techniques are adopted, including statistical analysis and trend analysis methods, to identify the size and type of different datasets, including calculating the quantity and frequency of each data type, as well as the specific requirements of the target data type for security protection. Data classification and data clustering techniques are utilized during the analysis process to group the data according to sensitivity and usage. Each group of data is evaluated for its security level requirements based on its sensitivity and the situation of noise addition. This process ensures that the security requirements of all data points are accurately identified and recorded. The generated security requirement information provides a basis for the selection of subsequent encryption measures and ensures that the processing measures match the actual requirements of the data.

[0068] S302. Based on the security requirement information, analyze the encryption intensity, computing resource consumption, and data processing efficiency of multiple homomorphic encryption configurations and generate a configuration comparison and analysis result.

[0069] Using an encryption strength evaluation algorithm and a resource consumption model, the target algorithms and models help evaluate the performance of each encryption configuration in actual operations. For example, the computational resource consumption of a certain configuration can be quantified by measuring the time required for encryption operations and the CPU usage rate. A comprehensive scoring system is utilized, taking into account encryption efficiency, cost, and maintenance complexity, to ensure that the recommended configuration not only meets security standards but also optimizes the economy and practicality of operations. The generated configuration comparison analysis results detail the scores, advantages, and disadvantages of each configuration, providing a scientific basis for selecting the most suitable encryption configuration for current data processing requirements.

[0070] S303. Based on the configuration comparison analysis results, considering security requirements and processing efficiency, match the homomorphic encryption configuration and generate data homomorphic encryption parameters.

[0071] Optionally, the specific formula for matching the homomorphic encryption configuration is as follows in Equation (2):

[0072] (2)

[0073] where F is the score of the target encryption parameter configuration, j is the index of the configuration factor, represents the weight of the j-th configuration factor, represents the performance value of the j-th configuration factor in the current environment, and m is the number of evaluation factors.

[0074] The above Equation (2) is explained by way of an example, and the process is as follows:

[0075] The formula is used to calculate the comprehensive scores of different homomorphic encryption configurations, and the results are used to select the optimal encryption configuration;

[0076] Parameter meanings and setting values:

[0077] is the weight of the j-th configuration factor, assumed to be encryption strength, computational resource consumption, and processing efficiency respectively, with corresponding values of 0.6, 0.3, and 0.1; is the performance value of the j-th configuration factor in the current environment, assumed to be 0.9, 0.7, and 0.5, representing the actual performance of encryption strength, resource consumption, and efficiency; m is the total number of evaluation factors, assumed to be 3;

[0078] Substitute the parameters into the formula for calculation:

[0079] ;

[0080] ;

[0081] ;

[0082] ;

[0083] Result It shows that the target configuration can effectively control resource consumption and maintain a moderate processing efficiency while ensuring high encryption strength. The calculation process helps to balance multiple important factors and optimize the security and data processing efficiency of the encryption configuration.

[0084] S4. Based on the data homomorphic encryption parameters, by simulating various network attack conditions, analyze the response status and stability of the encrypted data under various loads and attacks, evaluate the actual performance of the target encryption parameters and make adjustments to generate the encryption performance analysis results.

[0085] Optionally, the specific operation steps of S4 include the following S401 - S403:

[0086] S401. Based on the data homomorphic encryption parameters, simulate the network attack scenario, test the response status of the encrypted data, record the response time and data loss rate under various conditions, and generate the performance test results.

[0087] In the network attack simulation sub - step, based on the determined data homomorphic encryption parameters, test the response ability of the encrypted data by setting different network attack scenarios. Adopt the multi - condition stress test method to simulate common network threats such as DDoS attacks and SQL injections, monitor the response time T and data loss rate D of the encrypted data under various attacks, and calculate the response efficiency under various scenarios according to the formula , and generate the performance test results;

[0088] In the formula, T represents the average response time, D represents the data loss rate, e is the base of the natural logarithm, and R represents the response efficiency;

[0089] The detailed explanation of the formula and the formula calculation derivation process are as follows:

[0090] Suppose in the target DDoS attack test, the average response time of the encrypted data is 2 seconds, , and the data loss rate is 0.05, , calculate R:

[0091] ;

[0092] The result 0.475 indicates that under the preset attack scenario, the response efficiency score of the target encryption configuration is 0.475. The calculation process is used to evaluate the data homomorphic encryption performance under different network attack conditions, obtain the response efficiency value, and help judge the adaptability of the current encryption settings and the effectiveness of security protection.

[0093] S402. Analyze the performance test results, detect the stability and security of encrypted data under various loads and attack conditions, and generate data stability information.

[0094] Use data analysis software such as MATLAB or the Pandas library of Python to process the performance test data, analyze the statistical characteristics of the data, including the average response time of encrypted data, the distribution of the longest and shortest response times, and the retention rate of data integrity. The analysis helps to determine which encryption configurations perform most stably in a specific network environment and which configurations can still maintain high data security under high attack risks, generating data stability information. The target information indicates the security and stability performance of encrypted data under various environmental conditions, providing key data for further adjustment of the encryption strategy.

[0095] S403. Based on the data stability information, consider the security requirements and processing efficiency of the data, and adjust the parameter settings of the homomorphic encryption algorithm to generate encryption performance analysis results.

[0096] Evaluate the existing parameter settings of the homomorphic encryption algorithm according to the data stability information, and adjust the configuration parameters to match the security requirements of the data and optimize the processing efficiency. Consider the priority of data protection and the resource consumption of encryption operations during the process. Use decision analysis methods, including cost-benefit analysis, to compare the security and efficiency performance under different parameter settings. Adjust the key length, encryption rounds, and operation complexity of the encryption algorithm during the process to ensure that the encryption process meets the security standards without excessive consumption of computing resources, generating encryption performance analysis results. The target results describe in detail the adjusted parameter settings and their expected impact on encryption performance, ensuring that the encryption strategy is both secure and efficient.

[0097] S5. Based on the encryption performance analysis results, perform homomorphic encryption processing on various input data, monitor the encrypted data stream in real time, analyze the compatibility and correlation between data, and evaluate the consistency and integrity of the data in the encrypted state to generate input data encryption records.

[0098] Optionally, the specific operation steps of S5 include the following S501 - S503:

[0099] S501. Use the encryption performance analysis results to perform homomorphic encryption on various input data to generate an encrypted processing data set.

[0100] Use RSA or ElGamal, adjust the configuration of the encryption algorithm according to the type and sensitivity of the data, including selecting an appropriate key length and complexity, setting security protocols and standards during the encryption process, apply the encryption algorithm to each type of data, and record and verify each encryption operation generated to ensure the encryption quality. The process ensures that all input data is adequately protected before storage or transmission, generating a processed dataset containing various types of encrypted data.

[0101] S502. Based on the encrypted processed dataset, monitor the encrypted data stream in real time, record the interactions between multiple data sources and the data transmission status, and generate an encrypted data stream monitoring log.

[0102] Use network monitoring tools such as Wireshark and data monitoring scripts to track and record the transmission status of encrypted data in the network and the interactions between multiple data sources, record the data transmission speed, access frequency, and packet loss situation. Any abnormal data stream detected will trigger a security alert. The target monitoring log helps identify potential data leakage points or security vulnerabilities, provides real-time data support for the network security team, and generates an encrypted data stream monitoring log, providing basic data for subsequent security analysis and optimization.

[0103] S503. Analyze the encrypted data stream monitoring log, analyze the compatibility and relevance between the data after encryption processing, and evaluate the consistency and integrity of the data in the encrypted state, generating an input data encryption record.

[0104] Use data analysis tools such as the data processing libraries NumPy and Pandas in Python to analyze the encrypted data stream recorded in the monitoring log, evaluate the consistency and integrity of the data during the encryption process. During the analysis, calculate the change rate and error range before and after data encryption to ensure that the data still maintains its original data structure and relevance without being damaged after encryption. The operation results reveal whether the encryption processing has changed the data content or the relationship between the data, ensuring that all outputs after data processing are highly consistent and complete. Based on the target analysis, generate an input data encryption record, recording the encryption effect and the consistency status of the data, providing an accurate record for the further use and protection of the data.

[0105] In the embodiments of the present invention, by real-time identifying the risk points in the data, specifically adjusting the noise level of the data points, improving the security and privacy protection level during the data processing process, by comprehensively considering the data processing efficiency, security level requirements, and computing resource consumption, improving the efficiency and security of the encryption processing, by simulating network attack conditions to test the stability of the encrypted data, enhancing the ability to respond to network attacks, enabling the input data to support multiple data analysis tasks, and ensuring the privacy and security of the data.

[0106] Figure 2 This is a block diagram of a machine learning system based on enhanced privacy protection provided by an embodiment of the present invention. This system is used for a machine learning method based on enhanced privacy protection. Refer to Figure 2 , this device includes a data flow monitoring module 210, a noise management module 220, an encryption parameter configuration module 230, a network attack simulation module 240, and an encrypted data monitoring module 250. Among them:

[0107] The data flow monitoring module 210 is used to monitor the real-time data flows of multiple data sources based on real-time input data, analyze the relationships between various types of data, draw a data association network diagram, and evaluate the information leakage risk and privacy sensitivity of multiple data points by analyzing the interconnection density and path distribution of nodes in the network diagram, and generate a sensitivity evaluation index;

[0108] The noise management module 220 is used to analyze the privacy requirement levels of multiple data points according to the sensitivity evaluation index and the privacy sensitivities of multiple data points, and adjust the types and intensities of noises added to the target data points, and generate a data noise addition record;

[0109] The encryption parameter configuration module 230 is used to match the homomorphic encryption configuration based on the data noise addition record, according to the data volume and security level requirements of the input data, and generate data homomorphic encryption parameters by analyzing the computing resource consumption, encryption intensity, and processing efficiency of various homomorphic encryption configurations;

[0110] The network attack simulation module 240 is used to analyze the response status and stability of encrypted data under various loads and attack conditions based on the data homomorphic encryption parameters by simulating various network attack conditions, evaluate the actual performance of the target encryption parameters and make adjustments, and generate an encryption performance analysis result;

[0111] The encrypted data monitoring module 250 is used to perform homomorphic encryption processing on various input data based on the encryption performance analysis result, monitor the encrypted data flow in real time, analyze the compatibility and relevance between data, and evaluate the consistency and integrity of the data in the encrypted state, and generate an input data encryption record.

[0112] Optionally, the sensitivity evaluation index is specifically the data point connectivity, data path distribution, and data point risk level. The data noise addition record includes data point sensitivity, privacy requirement level, and noise addition parameters. The data homomorphic encryption parameters specifically refer to the data processing efficiency of various configurations, the data security levels of various encryption configurations, and the encryption configuration matching record. The encryption performance analysis result includes data response status information, stability test results, and encryption parameter adjustment records. The input data encryption record is specifically the data flow monitoring record, data compatibility analysis result, and data integrity evaluation result.

[0113] Optionally, the data stream monitoring module 210 is used for:

[0114] S101. Based on the real-time input data, monitor the real-time data streams of multiple data sources in real time, record data changes, and generate a data stream interaction analysis result by analyzing the mutual relationships between the data;

[0115] S102. Based on the data stream interaction analysis result, draw a data association network diagram according to the mutual relationships between the data, and generate data network structure information;

[0116] S103. According to the data network structure information, identify the information leakage risks of multiple nodes by calculating the interconnect density and path distribution of multiple nodes in the network, evaluate the privacy sensitivity of multiple data points, and generate a sensitivity evaluation index.

[0117] Optionally, the specific formula for evaluating the privacy sensitivity of multiple data points is as follows in formula (1):

[0118] (1)

[0119] Wherein, is the total number of nodes in the network, is the number of direct connections of the i-th node to other nodes, is the average value of the degrees of all nodes in the network, is the standard deviation of the node degrees in the network, is the clustering coefficient of the i-th node, represents the maximum clustering coefficient in the network, i is the index of the node, and S is the sensitivity score of each node.

[0120] Optionally, the noise management module 220 is used for:

[0121] S201. Evaluate the privacy protection requirements of multiple data points according to the sensitivity evaluation index and the privacy sensitivity of multiple data points, and generate a privacy requirement analysis result;

[0122] S202. Based on the privacy requirement analysis result, analyze the types of noise that need to be added to multiple data according to the types of data points, and generate a noise type analysis result;

[0123] S203. Based on the noise type analysis result, calculate the noise intensity that needs to be added to multiple data points, and add noise to the target data points to generate a data noise addition record.

[0124] Optionally, the encryption parameter configuration module 230 is used for:

[0125] S301. Identify the data volume and security level requirements of the input data based on the data noise addition record, and generate security requirement information;

[0126] S302. Analyze the encryption strength, computing resource consumption, and data processing efficiency of multiple homomorphic encryption configurations based on the security requirement information, and generate a configuration comparison and analysis result;

[0127] S303. Match a homomorphic encryption configuration considering the security requirements and processing efficiency based on the configuration comparison and analysis result, and generate data homomorphic encryption parameters.

[0128] Optionally, the specific formula for matching the homomorphic encryption configuration is as follows in Equation (2):

[0129] (2)

[0130] Where F is the score of the target encryption parameter configuration, j is the index of the configuration factor, represents the weight of the j-th configuration factor, represents the performance value of the j-th configuration factor in the current environment, and m is the number of evaluation factors.

[0131] Optionally, the network attack simulation module 240 is used for:

[0132] S401. Simulate a network attack scenario based on the data homomorphic encryption parameters, test the response status of the encrypted data, record the response time and data loss rate under multiple conditions, and generate a performance test result;

[0133] S402. Analyze the performance test result, detect the stability and security of the encrypted data under multiple load and attack conditions, and generate data stability information;

[0134] S403. Adjust the parameter settings of the homomorphic encryption algorithm considering the security requirements and processing efficiency of the data based on the data stability information, and generate an encryption performance analysis result.

[0135] Optionally, the encrypted data monitoring module 250 is used for:

[0136] S501. Perform homomorphic encryption on multiple input data using the encryption performance analysis result, and generate an encrypted processing data set;

[0137] S502. Based on the encrypted processing data set, monitor the encrypted data stream in real time, record the interactions and data transmission status between multiple data sources, and generate an encrypted data stream monitoring log;

[0138] S503: Analyze the encrypted data flow monitoring log, analyze the compatibility and relevance of the encrypted data, evaluate the consistency and integrity of the encrypted data, and generate an input data encryption record.

[0139] In the embodiments of the present invention, risk points in the data are identified in real time and the noise level of the data points is adjusted in a targeted manner to improve the security and privacy protection level in the data processing process. The efficiency and security of encryption processing are improved by comprehensively considering data processing efficiency, security level requirements, and computing resource consumption. The stability of encrypted data is tested by simulating network attack conditions to enhance the ability to respond to network attacks, so that the input data supports a variety of data analysis tasks, thereby ensuring the privacy and security of the data.

[0140] Figure 3 is a structural diagram of a machine learning device based on privacy protection enhancement provided by an embodiment of the present invention, such as Figure 3 As shown, the machine learning device based on privacy protection enhancement may include the above Figure 2 Optionally, the machine learning device 310 based on privacy protection enhancement may include a first processor 2001.

[0141] Optionally, the privacy protection enhanced machine learning device 310 may further include a memory 2002 and a transceiver 2003 .

[0142] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0143] Combine the following Figure 3 The components of the privacy protection-enhanced machine learning device 310 are specifically introduced as follows:

[0144] The first processor 2001 is the control center of the privacy-enhanced machine learning device 310, which can be a processor or a general term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement an embodiment of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (field programmable gate arrays, FPGAs).

[0145] Optionally, the first processor 2001 may execute various functions of the machine learning device 310 with enhanced privacy protection by running or executing software programs stored in the memory 2002 and invoking data stored in the memory 2002.

[0146] In a specific implementation, as an example, the first processor 2001 may include one or more CPUs, such as Figure 3 CPU0 and CPU1 shown in

[0147] In a specific implementation, as an example, the machine learning device 310 with enhanced privacy protection may also include multiple processors, such as Figure 3 the first processor 2001 and the second processor 2004 shown in

[0148] Among them, the memory 2002 is used to store software programs for implementing the solution of the present invention and is controlled by the first processor 2001 for execution. The specific implementation manner may refer to the above method embodiments and will not be elaborated here.

[0149] Optionally, the memory 2002 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or may also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently and be coupled to the first processor 2001 through an interface circuit ( Figure 3 not shown in

[0150] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0151] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 3 not shown separately in the figure). Among them, the receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0152] Optionally, the transceiver 2003 may be integrated with the first processor 2001, or may exist independently, and is coupled to the first processor 2001 through an interface circuit ( Figure 3 not shown in the figure) of the machine learning device 310 with enhanced privacy protection. The embodiments of the present invention do not make specific limitations in this regard.

[0153] It should be noted that Figure 3 the structure of the machine learning device 310 with enhanced privacy protection shown in the figure does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0154] In addition, the technical effects of the machine learning device 310 with enhanced privacy protection may refer to the technical effects of the machine learning device method with enhanced privacy protection described in the above method embodiments, and will not be elaborated here.

[0155] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), and this processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0156] It should also be understood that the memory in the embodiments of the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0157] The above-described embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any combination thereof. When implemented using software, the above-described embodiments 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 or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are 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 from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0158] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context before and after.

[0159] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0160] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the order of execution is prior or subsequent. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0161] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0162] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described devices, apparatuses, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0163] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be in an electrical, mechanical, or other form.

[0164] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0165] In addition, the functional units in each embodiment of the present invention can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0166] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0167] As described above, the above are only specific implementation manners of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A machine learning method based on enhanced privacy protection, characterized in that The method includes: S1. Based on real-time input data, monitor the real-time data streams of multiple data sources, analyze the relationships between various types of data, draw a data association network diagram, and evaluate the information leakage risk and privacy sensitivity of multiple data points by analyzing the interconnection density and path distribution of nodes in the network diagram, so as to generate sensitivity evaluation indicators; S2. According to the sensitivity evaluation indicators and the privacy sensitivity of multiple data points, analyze the privacy requirement levels of multiple data points, and adjust the types and intensities of noise added to the target data points to generate data noise addition records; S3. Based on the data noise addition records, according to the data volume and security level requirements of the input data, match the homomorphic encryption configuration by analyzing the computing resource consumption, encryption intensity, and processing efficiency of various homomorphic encryption configurations, so as to generate data homomorphic encryption parameters; S4. Based on the data homomorphic encryption parameters, simulate various network attack conditions, analyze the response status and stability of the encrypted data under various loads and attack conditions, evaluate the actual performance of the target encryption parameters and make adjustments, so as to generate encryption performance analysis results; S5. Based on the encryption performance analysis results, perform homomorphic encryption processing on various input data, monitor the encrypted data streams in real time and analyze the compatibility and relevance between the data, and evaluate the consistency and integrity of the data in the encrypted state to generate input data encryption records.

2. The machine learning method based on enhanced privacy protection according to claim 1, wherein, The sensitivity evaluation indicators of S1 are specifically the data point connectivity, data path distribution, and data point risk level. The data noise addition records include data point sensitivity, privacy requirement level, and noise addition parameters. The data homomorphic encryption parameters specifically refer to the data processing efficiency of various configurations, the data security levels of various encryption configurations, and the encryption configuration matching records. The encryption performance analysis results include data response status information, stability test results, and encryption parameter adjustment records. The input data encryption records are specifically data stream monitoring records, data compatibility analysis results, and data integrity evaluation results.

3. The machine learning method based on enhanced privacy protection according to claim 1, wherein The steps of S1 for monitoring the real-time data streams of multiple data sources based on real-time input data, analyzing the relationships between various types of data, drawing a data association network diagram, and evaluating the information leakage risk and privacy sensitivity of multiple data points by analyzing the interconnection density and path distribution of nodes in the network diagram to generate sensitivity evaluation indicators are specifically as follows: S101. Based on real-time input data, monitor the real-time data streams of multiple data sources in real time, record data changes, and generate data stream interaction analysis results by analyzing the mutual relationships between the data; S102. Based on the data stream interaction analysis results, draw a data association network diagram according to the mutual relationships between the data to generate data network structure information; S103. According to the data network structure information, identify the information leakage risks of multiple nodes by calculating the interconnection density and path distribution of multiple nodes in the network, evaluate the privacy sensitivity of multiple data points, and generate sensitivity evaluation indicators.

4. The machine learning method based on enhanced privacy protection according to claim 3, wherein The specific formula for evaluating the privacy sensitivity of multiple data points is as follows, Equation (1): （1） Among them, is the total number of nodes in the network, is the number of direct connections of the i-th node to other nodes, is the average value of the degrees of all nodes in the network, is the standard deviation of the node degrees in the network, is the clustering coefficient of the i-th node, represents the maximum clustering coefficient in the network, i is the index of the node, and S is the sensitivity score of each node.

5. The machine learning method based on enhanced privacy protection according to claim 1, wherein The steps of S2 for analyzing the privacy requirement levels of multiple data points according to the sensitivity evaluation index and the privacy sensitivities of multiple data points, and adjusting the noise type and intensity added to the target data point to generate a data noise addition record are specifically as follows: S201. Evaluate the privacy protection requirements of multiple data points according to the sensitivity evaluation index and the privacy sensitivities of multiple data points, and generate a privacy requirement analysis result; S202. Based on the privacy requirement analysis result, analyze the noise types to be added to multiple data according to the types of data points, and generate a noise type analysis result; S203. Based on the noise type analysis result, calculate the noise intensity to be added to multiple data points, and add noise to the target data point to generate a data noise addition record.

6. The machine learning method based on enhanced privacy protection according to claim 1, characterized in that, The steps of S3 for matching a homomorphic encryption configuration based on the data noise addition record, according to the data volume and security level requirements of the input data, by analyzing the computing resource consumption, encryption intensity, and processing efficiency of multiple homomorphic encryption configurations, and generating data homomorphic encryption parameters are specifically as follows: S301. Based on the data noise addition record, identify the data volume and security level requirements of the input data, and generate security requirement information; S302. Based on the security requirement information, analyze the encryption intensity, computing resource consumption, and data processing efficiency of multiple homomorphic encryption configurations, and generate a configuration comparison analysis result; S303. Based on the configuration comparison analysis result, considering the security requirements and processing efficiency, match a homomorphic encryption configuration and generate data homomorphic encryption parameters.

7. The machine learning method based on enhanced privacy protection according to claim 6, wherein The specific formula for matching the homomorphic encryption configuration is as follows, formula (2): （2） Among them, F is the score configured for the target encryption parameter, j is the index of the configuration factor, represents the weight of the j-th configuration factor, represents the performance value of the j-th configuration factor in the current environment, and m is the number of evaluation factors.

8. The machine learning method based on enhanced privacy protection according to claim 1, characterized in that The steps of S4 for analyzing the response status and stability of encrypted data under multiple loads and attack conditions based on the data homomorphic encryption parameters, simulating multiple network attack conditions, evaluating the actual performance of the target encryption parameters and making adjustments, and generating an encryption performance analysis result are specifically as follows: S401. Based on the data homomorphic encryption parameters, simulate a network attack scenario, test the response status of the encrypted data, record the response time and data loss rate under multiple conditions, and generate a performance test result; S402. Analyze the performance test result, detect the stability and security of the encrypted data under multiple loads and attack conditions, and generate data stability information; S403. Based on the data stability information, considering the security requirements and processing efficiency of the data, adjust the parameter settings of the homomorphic encryption algorithm to generate an encryption performance analysis result.

9. The machine learning method based on enhanced privacy protection according to claim 1, characterized in that, The steps of S5 for performing homomorphic encryption processing on multiple input data based on the encryption performance analysis result, real-time monitoring the encrypted data stream and analyzing the compatibility and correlation between data, and evaluating the consistency and integrity of the data in the encrypted state, and generating an input data encryption record are specifically as follows: S501. Use the encryption performance analysis result to perform homomorphic encryption on multiple input data to generate an encrypted processing data set; S502. Based on the encrypted processing data set, real-time monitor the encrypted data stream, record the interactions and data transmission status between multiple data sources, and generate an encrypted data stream monitoring log; S503. Analyze the encrypted data stream monitoring log, analyze the compatibility and correlation between the data after encryption processing, and evaluate the consistency and integrity of the data in the encrypted state to generate an input data encryption record.

10. A machine learning system with enhanced privacy protection, characterized in that, The machine learning system with enhanced privacy protection is used to implement the machine learning method with enhanced privacy protection according to any one of claims 1-9. It is characterized in that the system includes: A data stream monitoring module, which is used to monitor the real-time data streams of multiple data sources based on real-time input data, draw a relationship graph between the data, analyze the connection density and path distribution of the nodes in the graph, evaluate the leakage risk and privacy sensitivity of multiple data points, and generate a sensitivity evaluation index; A noise management module, which is used to add noise to multiple data points according to the sensitivity evaluation index, and adjust the noise type and intensity according to the privacy requirement levels of multiple data points to generate a data noise addition record; An encryption parameter configuration module, which is used to utilize the data noise addition record, analyze the data volume and security level requirements of the input data, match the data homomorphic encryption configuration by calculating the data processing efficiency, computing resource consumption, and encryption intensity of multiple encryption configurations, and generate data homomorphic encryption parameters; A network attack simulation module, which is used to use the data homomorphic encryption parameters to simulate various network attack conditions, analyze the response and stability of the encrypted data under various load and attack states, and adjust the encryption parameters to generate an encryption performance analysis result; An encrypted data monitoring module, which is used to perform homomorphic encryption processing on various input data according to the encryption performance analysis result, monitor the encrypted data stream in real time, analyze the compatibility and correlation between the data, evaluate the consistency and integrity of the data, and generate an input data encryption record.

Images