技术领域Technical field
本发明涉及数据安全技术领域，具体为一种数字平台的数据安全管理方法。The present invention relates to the field of data security technology, and is specifically a data security management method for a digital platform.
背景技术Background technique
数据安全管理方法是指通过采用一系列措施和策略来保护数据免受未经授权访问、损坏、泄漏或篡改的风险。数据安全管理的目标是确保数据的机密性、完整性和可用性，并符合适用的法律法规和行业标准。A data security management approach refers to protecting data from the risk of unauthorized access, damage, leakage or tampering by employing a series of measures and policies. The goal of data security management is to ensure the confidentiality, integrity and availability of data and to comply with applicable laws, regulations and industry standards.
而数字平台的数据安全管理方法是指针对数字平台上的数据，采取的一系列措施和策略来确保数据的保密性、完整性和可用性，以及防止数据被未经授权的访问、篡改、泄漏或损坏。数字平台可以是各种在线服务、应用程序、云服务，以及网络和移动应用，数字平台的数据安全管理方法的常见措施包括，身份认证和访问控制、数据加密、数据备份和灾难恢复、安全审计和监控、员工培训和意识、网络安全防护、更新和漏洞修补、数据分类和访问权限控制等，可以更好地保护其数据资源，确保数据安全，并为用户提供更可靠和安全的服务。随着技术的不断发展，数字平台的数据安全管理方法也会持续优化和更新。The data security management method of digital platforms refers to a series of measures and strategies adopted for data on digital platforms to ensure the confidentiality, integrity and availability of data, and to prevent data from unauthorized access, tampering, leakage or damage. Digital platforms can be various online services, applications, cloud services, as well as network and mobile applications. Common measures for data security management methods of digital platforms include identity authentication and access control, data encryption, data backup and disaster recovery, and security audits. and monitoring, employee training and awareness, network security protection, updates and vulnerability patching, data classification and access control, etc., to better protect its data resources, ensure data security, and provide users with more reliable and secure services. As technology continues to develop, data security management methods for digital platforms will continue to be optimized and updated.
然而传统的数据安全管理方法使用规则引擎、签名检测方式来识别已知的攻击模式，无法有效对外部供给进行识别，同时传统的数据共享方式可能会直接共享原始数据，可能造成关键数据泄漏，因此亟需一种数字平台的数据安全管理方法来解决此类问题。However, traditional data security management methods use rule engines and signature detection methods to identify known attack patterns, which cannot effectively identify external supplies. At the same time, traditional data sharing methods may directly share original data, which may cause key data leaks. Therefore There is an urgent need for a data security management method for digital platforms to solve such problems.
发明内容Contents of the invention
(一)解决的技术问题(1) Technical problems solved
针对现有技术的不足，本发明提供了一种数字平台的数据安全管理方法，解决现有技术中存在的使用规则引擎、签名检测方式来识别已知的攻击模式，无法有效对外部供给进行识别，同时传统的数据共享方式可能会直接共享原始数据，可能造成关键数据泄漏的问题。In view of the shortcomings of the existing technology, the present invention provides a data security management method for a digital platform, which solves the problem in the existing technology of using rule engines and signature detection methods to identify known attack patterns, which makes it impossible to effectively identify external supplies. , at the same time, the traditional data sharing method may directly share the original data, which may cause the problem of key data leakage.
(二)技术方案(2) Technical solutions
为实现以上目的，本发明通过以下技术方案予以实现，本发明提供了一种数字平台的数据安全管理方法，该方法包括：In order to achieve the above objectives, the present invention is implemented through the following technical solutions. The present invention provides a data security management method for a digital platform, which method includes:
搭建深度学习模型，使用历史数据进行训练，并持续监控网络流量和用户行为，识别异常活动，实时监控网络流量和用户行为，自动检测并警报异常和潜在威胁，提供高精度的入侵检测；Build a deep learning model, use historical data for training, and continuously monitor network traffic and user behavior, identify abnormal activities, monitor network traffic and user behavior in real time, automatically detect and alert anomalies and potential threats, and provide high-precision intrusion detection;
差分隐私数据共享，采用差分隐私技术对敏感数据进行加密和加噪处理；Differential privacy data sharing uses differential privacy technology to encrypt and add noise to sensitive data;
生成对抗网络防御，引入生成对抗网络，生成对抗性样本来测试和加强传统机器学习模型的安全性；Generative adversarial network defense introduces generative adversarial networks to generate adversarial samples to test and enhance the security of traditional machine learning models;
联邦学习，采用联邦学习方法，对多个数据源在本地训练模型，仅共享模型参数而非原始数据；Federated learning uses federated learning methods to train models locally on multiple data sources, sharing only model parameters rather than original data;
安全增强学习，采用安全增强学习技术，使系统与环境交互，自主学习和调整防御策略；Security-enhanced learning uses security-enhanced learning technology to enable the system to interact with the environment, learn and adjust defense strategies independently;
边缘智能，在终端设备上部署边缘智能技术，实现实时的安全监控和处理；Edge intelligence deploys edge intelligence technology on terminal devices to achieve real-time security monitoring and processing;
可解释性AI，使用可解释性的人工智能模型，对模型决策过程进行解释和可视化；Explainable AI uses explainable artificial intelligence models to explain and visualize the model decision-making process;
自动漏洞修复，利用机器学习技术自动检测系统中的漏洞，并实时生成修复策略。Automatic vulnerability repair uses machine learning technology to automatically detect vulnerabilities in the system and generate repair strategies in real time.
本发明进一步地设置为：所述搭建深度模型并进行训练的具体步骤为：The present invention is further configured as follows: the specific steps of building a deep model and conducting training are:
收集网络流量和用户行为的历史数据作为训练数据集，进行数据清洗、特征提取和标签标注；Collect historical data of network traffic and user behavior as training data sets for data cleaning, feature extraction and labeling;
在入侵检测任务中，选用深度学习算法包括卷积神经网络CNN、递归神经网络RNN、长短期记忆网络LSTM、变换器Transformer，构建模型架构，包括输入层、隐藏层和输出层，设置激活函数、损失函数和优化算法；In the intrusion detection task, deep learning algorithms are selected, including convolutional neural network CNN, recursive neural network RNN, long short-term memory network LSTM, and transformer Transformer. The model architecture is constructed, including input layer, hidden layer and output layer, and the activation function is set. Loss functions and optimization algorithms;
将数据集划分为训练集、验证集和测试集，训练集用于模型训练，验证集用于调整超参数和避免过拟合，测试集用于评估模型性能；Divide the data set into a training set, a validation set and a test set. The training set is used for model training, the validation set is used to adjust hyperparameters and avoid overfitting, and the test set is used to evaluate model performance;
使用训练集对深度学习模型进行训练，迭代优化模型参数，最小化损失函数，优化算法选用梯度下降法GradientDescent和其变种，包括Adam、RMSprop；Use the training set to train the deep learning model, iteratively optimize the model parameters, and minimize the loss function. The optimization algorithm uses the gradient descent method GradientDescent and its variants, including Adam and RMSprop;
根据验证集的性能，调整模型的超参数，包括学习率、正则化系数、隐藏层节点数，优化模型的性能和泛化能力；According to the performance of the verification set, adjust the hyperparameters of the model, including learning rate, regularization coefficient, and number of hidden layer nodes, to optimize the performance and generalization ability of the model;
使用测试集对训练好的深度学习模型进行评估，计算性能指标，包括准确率、召回率、F1值；Use the test set to evaluate the trained deep learning model and calculate performance indicators, including accuracy, recall, and F1 value;
将训练好的深度学习模型部署到数字平台上，持续监控网络流量和用户行为，实时输入数据样本到模型中进行预测，识别出异常活动并触发相应的响应措施；Deploy the trained deep learning model to the digital platform, continuously monitor network traffic and user behavior, input data samples into the model in real time for prediction, identify abnormal activities and trigger corresponding response measures;
本发明进一步地设置为：所述深度学习模型以卷积神经网络建立：The present invention is further configured as follows: the deep learning model is established with a convolutional neural network:
输入：XInput:
隐藏层h＝f(W*X+b)Hidden layer h=f(W*X+b)
输出层y＝g(V*h+c)Output layer y=g(V*h+c)
其中X为数据样本的特征向量，f为激活函数1，g为激活函数2；Among them, X is the feature vector of the data sample, f is the activation function 1, and g is the activation function 2;
损失函数定义：Loss function definition:
损失函数L(y，y’)Loss function L(y, y’)
其中y为实际标签，y’为预测标签；where y is the actual label and y’ is the predicted label;
算法优化，参数更新规则：Algorithm optimization, parameter update rules:
其中α为学习率，为梯度向量；where α is the learning rate and is the gradient vector;
本发明进一步地设置为：所述差份隐私数据共享步骤具体包括：The present invention is further configured such that the differential privacy data sharing step specifically includes:
对预处理后的数据添加噪声，选用拉普拉斯噪声来进行加噪处理，具体噪声添加公式为：To add noise to the preprocessed data, Laplace noise is used for noise processing. The specific noise addition formula is:
其中，表示拉普拉斯分布的噪声，epsilon是隐私预算，sensitivity是敏感度；in, Represents the noise of Laplace distribution, epsilon is the privacy budget, and sensitivity is the sensitivity;
设置差分隐私的隐私预算epsilon，将经过加密和加噪处理后的数据共享给授权的数据使用方；Set the privacy budget epsilon for differential privacy, and share the encrypted and denoised data with authorized data users;
在接收到数据使用方的查询时，将加密和加噪后的数据进行解密和处理，然后返回响应结果；When receiving a query from the data user, the encrypted and noisy data will be decrypted and processed, and then the response result will be returned;
进行隐私保护分析，评估差分隐私技术的效果，确保共享数据满足隐私保护要求；Conduct privacy protection analysis, evaluate the effect of differential privacy technology, and ensure that shared data meets privacy protection requirements;
本发明进一步地设置为：所述隐私保护分析步骤还包括：The present invention is further configured such that: the privacy protection analysis step further includes:
对于共享的敏感数据，计算其敏感度；For shared sensitive data, calculate its sensitivity;
根据隐私预算epsilon的设置，确定噪声的大小；Determine the size of the noise according to the setting of the privacy budget epsilon;
使用差分隐私的数学定义来评估共享数据的隐私保护效果，具体为：Use the mathematical definition of differential privacy to evaluate the privacy protection effect of shared data, specifically:
对于任意相邻的数据集D和D’，以及任意查询Q，对于所有可能的查询结果S，满足以下条件：For any adjacent data sets D and D', and any query Q, for all possible query results S, the following conditions are satisfied:
Pr[Q(D)∈S]<＝exp(epsilon)*Pr[Q(D’)∈S]Pr[Q(D)∈S]<=exp(epsilon)*Pr[Q(D’)∈S]
其中，epsilon表示隐私预算，Q(D)表示在数据集D上查询D的结果，exp(epsilon)表示隐私预算的指数化；Among them, epsilon represents the privacy budget, Q(D) represents the result of querying D on the data set D, and exp(epsilon) represents the indexation of the privacy budget;
使用差分隐私失真来评估共享数据的隐私保护效果；Use differential privacy distortion to evaluate the privacy protection effect of shared data;
在进行隐私保护处理后，对共享数据进行性能评估，包括模型准确性、数据可用性和查询响应时间；After privacy-preserving processing, perform performance evaluation on shared data, including model accuracy, data availability, and query response time;
根据隐私保护分析的结果，调整差分隐私技术中的参数；Adjust parameters in differential privacy technology based on the results of privacy protection analysis;
本发明进一步地设置为：所述引入生成对抗网络来测试和加强传统机器学习模型的安全性步骤具体包括：The present invention is further configured as follows: the steps of introducing a generative adversarial network to test and enhance the security of traditional machine learning models specifically include:
采用生成器网络和判别器网络，并准备用于训练GAN的数据集，包括真实数据和噪声数据；使用真实数据和噪声数据训练GAN，生成器网络试图生成接近真实数据的样本，而判别器网络则试图区分真实数据和生成器生成的数据；Adopt a generator network and a discriminator network, and prepare a data set for training GAN, including real data and noise data; use real data and noise data to train GAN, the generator network attempts to generate samples close to real data, and the discriminator network then attempts to distinguish between real data and data generated by the generator;
使用训练好的生成器网络，生成对抗性样本；Use the trained generator network to generate adversarial samples;
使用生成的对抗性样本对传统机器学习模型进行测试，将对抗性样本作为输入输入到传统模型中，观察模型的输出结果；Use the generated adversarial samples to test the traditional machine learning model, input the adversarial samples as input into the traditional model, and observe the output results of the model;
根据对传统模型的测试结果，选择性进行对抗性U型捏脸和防御方法改进：Based on the test results of traditional models, adversarial U-shaped face pinching and defense method improvements are selectively carried out:
对抗性训练：将生成的对抗性样本与原始训练数据混合在一起，重新训练传统模型；Adversarial training: Mix the generated adversarial samples with the original training data to retrain the traditional model;
本发明进一步地设置为：所述采用联邦学习方法本地训练步骤具体包括：The present invention is further configured as follows: the local training step using the federated learning method specifically includes:
将需要参与联邦学习的多个数据源分别收集起来；Collect multiple data sources that need to participate in federated learning separately;
在开始联邦学习之前，随机初始化联邦学习模型的参数；Before starting federated learning, randomly initialize the parameters of the federated learning model;
在每次联邦学习迭代中，数据源进行顺序具体为：In each federated learning iteration, the order of data sources is specifically:
每个数据源使用本地数据在本地训练模型；Each data source trains the model locally using local data;
在本地训练完成后，每个数据源将本地训练得到的模型参数上传到中央服务器；After local training is completed, each data source uploads the model parameters obtained by local training to the central server;
中央服务器将收集到的模型参数进行聚合，并将中央服务器将聚合后的模型参数发送回各个数据源，更新各自的本地模型参数；The central server aggregates the collected model parameters, and the central server sends the aggregated model parameters back to each data source to update their respective local model parameters;
重复进行上述联邦学习迭代，直到模型收敛；Repeat the above federated learning iterations until the model converges;
联邦学习过程中的参数聚合公式为：The parameter aggregation formula in the federated learning process is:
其中，ω_avg是参数的平均值，N是数据源的数量，ω_i是第i个数据源的本地模型参数；Among them, ω_avg is the average value of the parameters, N is the number of data sources, and ω_i is the local model parameter of the i-th data source;
本发明进一步地设置为：所述安全增强学习步骤具体包括：The present invention is further configured such that the security enhanced learning step specifically includes:
安全增强学习中，使系统与环境交互，自主学习和调整防御策略的具体步骤为：In security enhancement learning, the specific steps to enable the system to interact with the environment, learn independently and adjust defense strategies are:
对系统环境进行建模，包括将系统运行环境、网络结构、攻击者行为抽象为数学模型；Model the system environment, including abstracting the system operating environment, network structure, and attacker behavior into mathematical models;
定义奖励函数，用于评估系统在不同状态下的表现；Define a reward function to evaluate the performance of the system in different states;
采用Q-learning算法，将建立的增强学习模型与系统的防御部分进行连接，使系统能够与环境交互；The Q-learning algorithm is used to connect the established reinforcement learning model with the defense part of the system so that the system can interact with the environment;
根据增强学习算法进行学习和优化，持续与环境交互和学习；Learn and optimize according to reinforcement learning algorithms, and continuously interact and learn with the environment;
增强学习中的强化学习算法的更新规则公式为：The update rule formula of the reinforcement learning algorithm in reinforcement learning is:
Q(s,a)＝Q(s,a)+a*(r+γ*max(Q(s’,a’))-Q(s,a))Q(s,a)＝Q(s,a)+a*(r+γ*max(Q(s’,a’))-Q(s,a))
其中，Q(s,a)表示在状态s下执行动作a的预期回报，a是学习率，r是在状态s下执行动作a后获得的奖励，γ是折扣因子，s’是执行动作a后的新状态，a’是在新状态s’下选择的最优动作。Among them, Q(s,a) represents the expected return of executing action a in state s, a is the learning rate, r is the reward obtained after executing action a in state s, γ is the discount factor, and s' is the execution of action a. After the new state, a' is the optimal action selected under the new state s'.
本发明还提供一种终端设备，该设备包括：存储器、处理器及存储在存储器上并可在处理器上运行的数字平台的数据安全管理方法的控制程序，所述数字平台的数据安全管理方法的控制程序被所述处理器执行时实现上述的数字平台的数据安全管理方法；The invention also provides a terminal device, which includes: a memory, a processor, and a control program for a data security management method of a digital platform that is stored in the memory and can be run on the processor. The data security management method of the digital platform When the control program is executed by the processor, the above-mentioned data security management method of the digital platform is implemented;
本发明还提供一种存储介质，该介质应用于计算机，所述存储介质上存储有数字平台的数据安全管理方法的控制程序，所述数字平台的数据安全管理方法的控制程序被所述处理器执行时实现上述的数字平台的数据安全管理方法。The present invention also provides a storage medium, which is applied to a computer. The storage medium stores a control program of a data security management method of a digital platform. The control program of a data security management method of a digital platform is controlled by the processor. The above-mentioned data security management method of the digital platform is implemented during execution.
(三)有益效果(3) Beneficial effects
本发明提供了一种数字平台的数据安全管理方法。具备以下有益效果：The invention provides a data security management method for a digital platform. It has the following beneficial effects:
本发明所提供的数字平台的数据安全管理方法使用深度学习模型进行入侵检测，搜集网络流量和用户行为的历史数据作为训练数据集，在入侵检测任务中构建模型架构，包括输入层、隐藏层和输出层，并设置激活函数、损失函数和优化算法，将数据集划分为训练集、验证集和测试集，用于模型的训练、超参数调整和评估，使用训练集对深度学习模型进行训练，通过迭代优化模型参数，最小化损失函数，采用梯度下降法进行优化，并根据验证集的性能，调整模型的超参数，包括学习率、正则化系数和隐藏层节点数，以提高模型的性能和泛化能力。The data security management method of the digital platform provided by the present invention uses a deep learning model for intrusion detection, collects historical data of network traffic and user behavior as a training data set, and builds a model architecture in the intrusion detection task, including an input layer, a hidden layer and Output layer, and set activation function, loss function and optimization algorithm, divide the data set into training set, verification set and test set for model training, hyperparameter adjustment and evaluation, use the training set to train the deep learning model, By iteratively optimizing the model parameters, minimizing the loss function, using the gradient descent method for optimization, and adjusting the model's hyperparameters, including learning rate, regularization coefficient and number of hidden layer nodes, according to the performance of the validation set, to improve the performance and performance of the model. Generalization.
针对实时入侵，将训练好的深度学习模型部署到数字平台上，持续监控网络流量和用户行为，实时输入数据样本到模型中进行预测，识别出异常活动并触发相应的响应措施。For real-time intrusions, deploy the trained deep learning model to the digital platform, continuously monitor network traffic and user behavior, input data samples into the model in real time for prediction, identify abnormal activities and trigger corresponding response measures.
针对隐私数据，利用差分隐私技术对敏感数据进行加密和加噪处理，保护数据隐私，同时允许授权的数据使用方获得有限的、不可逆的洞察，根据隐私预算将加密和加噪后的数据共享给授权的数据使用方。For private data, differential privacy technology is used to encrypt and denoise sensitive data to protect data privacy, while allowing authorized data users to obtain limited and irreversible insights, and share the encrypted and denoised data with the privacy budget. Authorized Data Users.
引入生成对抗网络，生成对抗性样本来测试和加强传统机器学习模型的安全性，通过对抗性训练、改进防御策略来增强模型的安全性，使用联邦学习方法在本地设备上进行模型训练，仅共享模型参数而非原始数据，以减少数据泄露风险，提高数据安全性，同时采用安全增强学习技术，使系统与环境交互，自主学习和调整防御策略，以适应不断变化的安全威胁。Introduce generative adversarial networks, generate adversarial samples to test and enhance the security of traditional machine learning models, enhance the security of models through adversarial training and improved defense strategies, use federated learning methods to train models on local devices, and only share Model parameters instead of raw data to reduce the risk of data leakage and improve data security. At the same time, security enhancement learning technology is used to enable the system to interact with the environment, learn and adjust defense strategies autonomously to adapt to changing security threats.
解决了现有技术中存在的使用规则引擎、签名检测方式来识别已知的攻击模式，无法有效对外部供给进行识别，同时传统的数据共享方式可能会直接共享原始数据，可能造成关键数据泄漏的问题。It solves the problem in the existing technology of using rule engines and signature detection methods to identify known attack patterns, which cannot effectively identify external supplies. At the same time, traditional data sharing methods may directly share original data, which may cause key data leakage. question.
附图说明Description of the drawings
图1为本发明的数字平台的数据安全管理方法的流程图。Figure 1 is a flow chart of the data security management method of the digital platform of the present invention.
具体实施方式Detailed ways
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整的描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.
实施例Example
请参阅图1，本发明提供一种数字平台的数据安全管理方法，包括如下步骤：Please refer to Figure 1. The present invention provides a data security management method for a digital platform, which includes the following steps:
S1、搭建深度学习模型，使用历史数据进行训练，并持续监控网络流量和用户行为，识别异常活动，实时监控网络流量和用户行为，自动检测并警报异常和潜在威胁，提供高精度的入侵检测；S1. Build a deep learning model, use historical data for training, and continuously monitor network traffic and user behavior, identify abnormal activities, monitor network traffic and user behavior in real time, automatically detect and alert anomalies and potential threats, and provide high-precision intrusion detection;
搭建深度模型并进行训练的具体步骤为：The specific steps to build a deep model and train it are:
收集网络流量和用户行为的历史数据作为训练数据集，进行数据清洗、特征提取和标签标注；Collect historical data of network traffic and user behavior as training data sets for data cleaning, feature extraction and labeling;
在入侵检测任务中，选用深度学习算法包括卷积神经网络CNN、递归神经网络RNN、长短期记忆网络LSTM、变换器Transformer，构建模型架构，包括输入层、隐藏层和输出层，设置激活函数、损失函数和优化算法；In the intrusion detection task, deep learning algorithms are selected, including convolutional neural network CNN, recursive neural network RNN, long short-term memory network LSTM, and transformer Transformer. The model architecture is constructed, including input layer, hidden layer and output layer, and the activation function is set. Loss functions and optimization algorithms;
将数据集划分为训练集、验证集和测试集，训练集用于模型训练，验证集用于调整超参数和避免过拟合，测试集用于评估模型性能；Divide the data set into a training set, a validation set and a test set. The training set is used for model training, the validation set is used to adjust hyperparameters and avoid overfitting, and the test set is used to evaluate model performance;
使用训练集对深度学习模型进行训练，迭代优化模型参数，最小化损失函数，优化算法选用梯度下降法GradientDescent和其变种，包括Adam、RMSprop；Use the training set to train the deep learning model, iteratively optimize the model parameters, and minimize the loss function. The optimization algorithm uses the gradient descent method GradientDescent and its variants, including Adam and RMSprop;
根据验证集的性能，调整模型的超参数，包括学习率、正则化系数、隐藏层节点数，优化模型的性能和泛化能力；According to the performance of the verification set, adjust the hyperparameters of the model, including learning rate, regularization coefficient, and number of hidden layer nodes, to optimize the performance and generalization ability of the model;
使用测试集对训练好的深度学习模型进行评估，计算性能指标，包括准确率、召回率、F1值；Use the test set to evaluate the trained deep learning model and calculate performance indicators, including accuracy, recall, and F1 value;
将训练好的深度学习模型部署到数字平台上，持续监控网络流量和用户行为，实时输入数据样本到模型中进行预测，识别出异常活动并触发相应的响应措施；Deploy the trained deep learning model to the digital platform, continuously monitor network traffic and user behavior, input data samples into the model in real time for prediction, identify abnormal activities and trigger corresponding response measures;
具体实现过程为：The specific implementation process is:
深度学习模型以卷积神经网络建立：The deep learning model is built with a convolutional neural network:
输入：XInput:
隐藏层h＝f(W*X+b)Hidden layer h=f(W*X+b)
输出层y＝g(V*h+c)Output layer y=g(V*h+c)
其中X为数据样本的特征向量，f为激活函数1，g为激活函数2；Among them, X is the feature vector of the data sample, f is the activation function 1, and g is the activation function 2;
损失函数定义：Loss function definition:
损失函数L(y，y’)Loss function L(y, y’)
其中y为实际标签，y’为预测标签；where y is the actual label and y’ is the predicted label;
算法优化，参数更新规则：Algorithm optimization, parameter update rules:
其中α为学习率，为梯度向量；where α is the learning rate and is the gradient vector;
S2、差分隐私数据共享，采用差分隐私技术对敏感数据进行加密和加噪处理；S2. Differential privacy data sharing, using differential privacy technology to encrypt and add noise to sensitive data;
差份隐私数据共享步骤具体包括：Differential privacy data sharing steps include:
对预处理后的数据添加噪声，选用拉普拉斯噪声来进行加噪处理，具体噪声添加公式为：To add noise to the preprocessed data, Laplace noise is used for noise processing. The specific noise addition formula is:
其中，表示拉普拉斯分布的噪声，epsilon是隐私预算，sensitivity是敏感度；in, Represents the noise of Laplace distribution, epsilon is the privacy budget, and sensitivity is the sensitivity;
设置差分隐私的隐私预算epsilon，将经过加密和加噪处理后的数据共享给授权的数据使用方；Set the privacy budget epsilon for differential privacy, and share the encrypted and denoised data with authorized data users;
在接收到数据使用方的查询时，将加密和加噪后的数据进行解密和处理，然后返回响应结果；When receiving a query from the data user, the encrypted and noisy data will be decrypted and processed, and then the response result will be returned;
进行隐私保护分析，评估差分隐私技术的效果，确保共享数据满足隐私保护要求；Conduct privacy protection analysis, evaluate the effect of differential privacy technology, and ensure that shared data meets privacy protection requirements;
隐私保护分析步骤具体包括：Privacy protection analysis steps specifically include:
对于共享的敏感数据，计算其敏感度；For shared sensitive data, calculate its sensitivity;
根据隐私预算epsilon的设置，确定噪声的大小；Determine the size of the noise according to the setting of the privacy budget epsilon;
使用差分隐私的数学定义来评估共享数据的隐私保护效果，具体为：Use the mathematical definition of differential privacy to evaluate the privacy protection effect of shared data, specifically:
对于任意相邻的数据集D和D’，以及任意查询Q，对于所有可能的查询结果S，满足以下条件：For any adjacent data sets D and D', and any query Q, for all possible query results S, the following conditions are satisfied:
Pr[Q(D)∈S]<＝exp(epsilon)*Pr[Q(D’)∈S]Pr[Q(D)∈S]<=exp(epsilon)*Pr[Q(D’)∈S]
其中，epsilon表示隐私预算，Q(D)表示在数据集D上查询D的结果，exp(epsilon)表示隐私预算的指数化；Among them, epsilon represents the privacy budget, Q(D) represents the result of querying D on the data set D, and exp(epsilon) represents the indexation of the privacy budget;
使用差分隐私失真来评估共享数据的隐私保护效果；Use differential privacy distortion to evaluate the privacy protection effect of shared data;
在进行隐私保护处理后，对共享数据进行性能评估，包括模型准确性、数据可用性和查询响应时间；After privacy-preserving processing, perform performance evaluation on shared data, including model accuracy, data availability, and query response time;
根据隐私保护分析的结果，调整差分隐私技术中的参数；隐私预算epsilon和噪声大小，以平衡隐私保护和数据准确性；According to the results of privacy protection analysis, adjust the parameters in differential privacy technology; privacy budget epsilon and noise size to balance privacy protection and data accuracy;
确保共享数据在进行差分隐私处理后，能够在保护隐私的同时仍具有可用性和有效性，通过设置隐私预算和噪声大小，以及进行性能评估和优化参数，使共享数据在差分隐私技术的保护下达到较好的隐私保护效果；Ensure that shared data can still be usable and effective while protecting privacy after differential privacy processing. By setting the privacy budget and noise size, as well as performing performance evaluation and optimizing parameters, the shared data can be protected by differential privacy technology. Better privacy protection effect;
S3、生成对抗网络防御，引入生成对抗网络，生成对抗性样本来测试和加强传统机器学习模型的安全性；有效检测和防御对抗性攻击，提高模型的鲁棒性和安全性；S3. Generative adversarial network defense: introduce generative adversarial networks to generate adversarial samples to test and enhance the security of traditional machine learning models; effectively detect and defend against adversarial attacks, and improve the robustness and security of the model;
引入生成对抗网络来测试和加强传统机器学习模型的安全性步骤具体包括：The specific steps to introduce generative adversarial networks to test and enhance the security of traditional machine learning models include:
采用生成器网络和判别器网络，并准备用于训练GAN的数据集，包括真实数据和噪声数据；使用真实数据和噪声数据训练GAN，生成器网络试图生成接近真实数据的样本，而判别器网络则试图区分真实数据和生成器生成的数据；Adopt a generator network and a discriminator network, and prepare a data set for training GAN, including real data and noise data; use real data and noise data to train GAN, the generator network attempts to generate samples close to real data, and the discriminator network then attempts to distinguish between real data and data generated by the generator;
使用训练好的生成器网络，生成对抗性样本；对抗性样本是指对原始输入样本进行微小扰动后得到的样本；Use the trained generator network to generate adversarial samples; adversarial samples refer to samples obtained after slight perturbations of the original input samples;
使用生成的对抗性样本对传统机器学习模型进行测试，将对抗性样本作为输入输入到传统模型中，观察模型的输出结果；如果模型在对抗性样本上表现不佳，可能表明模型对抗性攻击的鲁棒性较差；Use the generated adversarial samples to test the traditional machine learning model, input the adversarial samples as input into the traditional model, and observe the output results of the model; if the model performs poorly on the adversarial samples, it may indicate that the model is vulnerable to adversarial attacks. Poor robustness;
根据对传统模型的测试结果，选择性进行对抗性U型捏脸和防御方法改进：Based on the test results of traditional models, adversarial U-shaped face pinching and defense method improvements are selectively carried out:
对抗性训练：将生成的对抗性样本与原始训练数据混合在一起，重新训练传统模型；Adversarial training: Mix the generated adversarial samples with the original training data to retrain the traditional model;
通过反复迭代训练，生成器网络逐渐学会生成接近真实数据的样本，而判别器网络逐渐提高区分真实数据和生成数据的能力，通过利用生成对抗网络生成对抗性样本，评估和加强传统机器学习模型的鲁棒性，提高其在面对未知攻击和对抗性样本时的性能；Through repeated iterative training, the generator network gradually learns to generate samples that are close to real data, while the discriminator network gradually improves its ability to distinguish real data from generated data. By using the generative adversarial network to generate adversarial samples, it evaluates and enhances the performance of traditional machine learning models. Robustness, improving its performance in the face of unknown attacks and adversarial samples;
S4、联邦学习，采用联邦学习方法，对多个数据源在本地训练模型，仅共享模型参数而非原始数据；S4, federated learning, uses federated learning methods to train models locally on multiple data sources, sharing only model parameters rather than original data;
在不集中数据的情况下进行模型训练，减少数据泄露风险，同时实现更好的模型泛化能力；Conduct model training without centralizing data, reducing the risk of data leakage while achieving better model generalization capabilities;
采用联邦学习方法本地训练步骤具体包括：The local training steps using the federated learning method include:
将需要参与联邦学习的多个数据源分别收集起来；每个数据源在本地对自己的数据进行预处理、特征提取和标签标注，确保数据的一致性和可用性；Collect multiple data sources that need to participate in federated learning separately; each data source performs local preprocessing, feature extraction, and labeling of its own data to ensure data consistency and availability;
在开始联邦学习之前，随机初始化联邦学习模型的参数；Before starting federated learning, randomly initialize the parameters of the federated learning model;
在每次联邦学习迭代中，数据源进行顺序具体为：In each federated learning iteration, the order of data sources is specifically:
每个数据源使用本地数据在本地训练模型；Each data source trains the model locally using local data;
在本地训练完成后，每个数据源将本地训练得到的模型参数上传到中央服务器；After local training is completed, each data source uploads the model parameters obtained by local training to the central server;
中央服务器将收集到的模型参数进行聚合，并将中央服务器将聚合后的模型参数发送回各个数据源，更新各自的本地模型参数；The central server aggregates the collected model parameters, and the central server sends the aggregated model parameters back to each data source to update their respective local model parameters;
重复进行上述联邦学习迭代，直到模型收敛；Repeat the above federated learning iterations until the model converges;
联邦学习过程中的参数聚合公式为：The parameter aggregation formula in the federated learning process is:
其中，ω_avg是参数的平均值，N是数据源的数量，ω_i是第i个数据源的本地模型参数；Among them, ω_avg is the average value of the parameters, N is the number of data sources, and ω_i is the local model parameter of the i-th data source;
联邦学习允许多个数据源在本地训练模型，仅共享模型参数而不共享原始数据，从而保护用户的隐私和数据安全，通过联邦学习，不同数据源可以共同训练一个全局模型，而无需将数据集集中在一处；Federated learning allows multiple data sources to train models locally, sharing only model parameters but not original data, thereby protecting user privacy and data security. Through federated learning, different data sources can jointly train a global model without the need to separate the data set. concentrated in one place;
S5、安全增强学习，采用安全增强学习技术，使系统与环境交互，自主学习和调整防御策略；提高系统对不断变化的安全威胁的适应能力，增强安全防御效果；S5. Security-enhanced learning, using security-enhanced learning technology, allows the system to interact with the environment, learn and adjust defense strategies independently; improve the system's adaptability to changing security threats and enhance security defense effects;
安全增强学习中，使系统与环境交互，自主学习和调整防御策略的具体步骤为：In security enhancement learning, the specific steps to enable the system to interact with the environment, learn independently and adjust defense strategies are:
对系统环境进行建模，包括将系统运行环境、网络结构、攻击者行为抽象为数学模型；Model the system environment, including abstracting the system operating environment, network structure, and attacker behavior into mathematical models;
定义奖励函数，用于评估系统在不同状态下的表现；Define a reward function to evaluate the performance of the system in different states;
采用Q-learning算法，将建立的增强学习模型与系统的防御部分进行连接，使系统能够与环境交互；The Q-learning algorithm is used to connect the established reinforcement learning model with the defense part of the system so that the system can interact with the environment;
根据增强学习算法进行学习和优化，持续与环境交互和学习；Learn and optimize according to reinforcement learning algorithms, and continuously interact and learn with the environment;
增强学习中的强化学习算法的更新规则公式为：The update rule formula of the reinforcement learning algorithm in reinforcement learning is:
Q(s,a)＝Q(s,a)+a*(r+γ*max(Q(s’,a’))-Q(s,a))Q(s,a)＝Q(s,a)+a*(r+γ*max(Q(s’,a’))-Q(s,a))
其中，Q(s,a)表示在状态s下执行动作a的预期回报，a是学习率，r是在状态s下执行动作a后获得的奖励，γ是折扣因子，s’是执行动作a后的新状态，a’是在新状态s’下选择的最优动作，更新规则能够使系统根据环境的反馈不断优化Q值，从而找到最优的防御策略；通过安全增强学习在不断与环境交互和学习的过程中，自主学习和调整防御策略，提高系统对安全威胁的适应能力，从而增强系统的安全性；Among them, Q(s,a) represents the expected return of executing action a in state s, a is the learning rate, r is the reward obtained after executing action a in state s, γ is the discount factor, and s' is the execution of action a. After the new state, a' is the optimal action selected under the new state s'. The update rules can enable the system to continuously optimize the Q value based on feedback from the environment, thereby finding the optimal defense strategy; through security enhancement learning, it continuously interacts with the environment. During the process of interaction and learning, it learns and adjusts defense strategies independently to improve the system's adaptability to security threats, thereby enhancing system security;
S6、边缘智能，在终端设备上部署边缘智能技术，实现实时的安全监控和处理；S6, edge intelligence, deploys edge intelligence technology on terminal devices to achieve real-time security monitoring and processing;
减少对中心服务器的依赖，增强数据安全性和即时响应性；Reduce dependence on central servers and enhance data security and instant responsiveness;
在终端设备上部署边缘智能技术具体步骤包括：Specific steps to deploy edge intelligence technology on terminal devices include:
采用深度学习模型的轻量化的边缘智能技术，在终端设备上采集安全监控所需的数据，包括传感器数据、日志数据，将采集到的数据传输到边缘节点；Lightweight edge intelligence technology using deep learning models collects data required for security monitoring on terminal devices, including sensor data and log data, and transmits the collected data to edge nodes;
在边缘节点上部署深度学习模型的轻量化智能算法进行安全事件的检测和分析，包括实时的数据处理和安全监控模型；Deploy lightweight intelligent algorithms of deep learning models on edge nodes to detect and analyze security events, including real-time data processing and security monitoring models;
在边缘节点上进行实时的安全监控和处理，对采集到的数据进行分析和处理，检测可能的安全威胁，并触发相应的响应措施；Perform real-time security monitoring and processing on edge nodes, analyze and process the collected data, detect possible security threats, and trigger corresponding response measures;
当检测到安全威胁时，边缘节点触发安全响应机制，发送警报、阻断攻击流量；When a security threat is detected, the edge node triggers the security response mechanism, sends alerts, and blocks attack traffic;
S7、可解释性AI，使用可解释性的人工智能模型，对模型决策过程进行解释和可视化；帮助安全团队理解模型的行为，快速识别异常情况和安全事件；S7. Interpretable AI uses interpretable artificial intelligence models to explain and visualize the model decision-making process; help the security team understand the behavior of the model and quickly identify anomalies and security events;
使用可解释性的人工智能模型对模型决策过程进行解释和可视化，的具体步骤包括：Specific steps to use explainable artificial intelligence models to explain and visualize the model decision-making process include:
使用预处理后的数据对可解释性模型进行训练；Use the preprocessed data to train the interpretability model;
通过特征重要性分析、局部解释和全局解释，解释可解释性模型的决策过程；Explain the decision-making process of the interpretability model through feature importance analysis, local explanation and global explanation;
局部解释中，使用LIME法构建局部线性模型来解释模型在特定样本上的预测结果；In local explanation, the LIME method is used to build a local linear model to explain the model's prediction results on specific samples;
全局解释中，采用SHAP法计算特征对模型预测结果的贡献；In the global explanation, the SHAP method is used to calculate the contribution of features to the model prediction results;
将解释的结果进行可视化；Visualize the results of the interpretation;
S8、自动漏洞修复，利用机器学习技术自动检测系统中的漏洞，并实时生成修复策略，提高系统的安全性，自动修复漏洞，减少安全风险；S8. Automatic vulnerability repair, using machine learning technology to automatically detect vulnerabilities in the system and generate repair strategies in real time to improve system security, automatically repair vulnerabilities, and reduce security risks;
采用逻辑回归模型，利用机器学习技术自动检测系统中的漏洞，并实时生成修复策略的具体步骤为：The specific steps for using a logistic regression model, using machine learning technology to automatically detect vulnerabilities in the system, and generating repair strategies in real time are:
将数据集划分为训练集和测试集，用于模型的训练和评估；Divide the data set into a training set and a test set for model training and evaluation;
使用训练集对选定的机器学习模型进行训练；Train the selected machine learning model using the training set;
使用测试集对训练好的模型进行评估，评估模型的准确性和性能；Use the test set to evaluate the trained model to evaluate the accuracy and performance of the model;
在系统运行时，实时检测系统中的漏洞，利用训练好的机器学习模型，对实时收集的数据进行预测，判断是否存在漏洞，如果检测到漏洞，根据模型的预测结果和漏洞的特征，生成相应的修复策略；When the system is running, vulnerabilities in the system are detected in real time, and the trained machine learning model is used to predict the data collected in real time to determine whether there is a vulnerability. If a vulnerability is detected, a corresponding response is generated based on the prediction results of the model and the characteristics of the vulnerability. repair strategies;
将生成的修复策略应用到系统中，修复检测到的漏洞；Apply the generated remediation strategy to the system to repair the detected vulnerabilities;
逻辑回归的机器学习模型的预测公式为：The prediction formula of the logistic regression machine learning model is:
其中，y是模型的预测输出，z是输入数据的线性组合。where y is the predicted output of the model and z is a linear combination of the input data.
综合以上内容，在本申请中：Based on the above content, in this application:
本发明所提供的数字平台的数据安全管理方法使用深度学习模型进行入侵检测，搜集网络流量和用户行为的历史数据作为训练数据集，在入侵检测任务中构建模型架构，包括输入层、隐藏层和输出层，并设置激活函数、损失函数和优化算法，将数据集划分为训练集、验证集和测试集，用于模型的训练、超参数调整和评估，使用训练集对深度学习模型进行训练，通过迭代优化模型参数，最小化损失函数，采用梯度下降法进行优化，并根据验证集的性能，调整模型的超参数，包括学习率、正则化系数和隐藏层节点数，以提高模型的性能和泛化能力。The data security management method of the digital platform provided by the present invention uses a deep learning model for intrusion detection, collects historical data of network traffic and user behavior as a training data set, and builds a model architecture in the intrusion detection task, including an input layer, a hidden layer and Output layer, and set activation function, loss function and optimization algorithm, divide the data set into training set, verification set and test set for model training, hyperparameter adjustment and evaluation, use the training set to train the deep learning model, By iteratively optimizing the model parameters, minimizing the loss function, using the gradient descent method for optimization, and adjusting the model's hyperparameters, including learning rate, regularization coefficient and number of hidden layer nodes, according to the performance of the validation set, to improve the performance and performance of the model. Generalization.
针对实时入侵，将训练好的深度学习模型部署到数字平台上，持续监控网络流量和用户行为，实时输入数据样本到模型中进行预测，识别出异常活动并触发相应的响应措施。For real-time intrusions, deploy the trained deep learning model to the digital platform, continuously monitor network traffic and user behavior, input data samples into the model in real time for prediction, identify abnormal activities and trigger corresponding response measures.
针对隐私数据，利用差分隐私技术对敏感数据进行加密和加噪处理，保护数据隐私，同时允许授权的数据使用方获得有限的、不可逆的洞察，根据隐私预算将加密和加噪后的数据共享给授权的数据使用方。For private data, differential privacy technology is used to encrypt and denoise sensitive data to protect data privacy, while allowing authorized data users to obtain limited and irreversible insights, and share the encrypted and denoised data with the privacy budget. Authorized Data Users.
引入生成对抗网络，生成对抗性样本来测试和加强传统机器学习模型的安全性，通过对抗性训练、改进防御策略来增强模型的安全性，使用联邦学习方法在本地设备上进行模型训练，仅共享模型参数而非原始数据，以减少数据泄露风险，提高数据安全性，同时采用安全增强学习技术，使系统与环境交互，自主学习和调整防御策略，以适应不断变化的安全威胁。Introduce generative adversarial networks, generate adversarial samples to test and enhance the security of traditional machine learning models, enhance the security of models through adversarial training and improved defense strategies, use federated learning methods to train models on local devices, and only share Model parameters instead of raw data to reduce the risk of data leakage and improve data security. At the same time, security enhancement learning technology is used to enable the system to interact with the environment, learn and adjust defense strategies autonomously to adapt to changing security threats.
在本发明的实施例的描述中，对于本领域的普通技术人员而言，可以理解在不脱离本发明的原理和精神的情况下可以对这些实施例进行多种变化、修改、替换和变型，本发明的范围由所附权利要求及其等同物限定。In the description of the embodiments of the present invention, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and modifications can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the invention is defined by the appended claims and their equivalents.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311004874.XACN117056951B (en) | 2023-08-09 | 2023-08-09 | Data security management method for digital platform |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311004874.XACN117056951B (en) | 2023-08-09 | 2023-08-09 | Data security management method for digital platform |
| Publication Number | Publication Date |
|---|---|
| CN117056951Atrue CN117056951A (en) | 2023-11-14 |
| CN117056951B CN117056951B (en) | 2024-06-25 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311004874.XAActiveCN117056951B (en) | 2023-08-09 | 2023-08-09 | Data security management method for digital platform |
| Country | Link |
|---|---|
| CN (1) | CN117056951B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117289141A (en)* | 2023-11-22 | 2023-12-26 | 深圳市麦迪瑞科技有限公司 | Electric bicycle charging state monitoring method based on artificial intelligence |
| CN117557270A (en)* | 2024-01-08 | 2024-02-13 | 深圳合纵富科技有限公司 | Mobile terminal secure payment management method and system |
| CN117931827A (en)* | 2024-03-21 | 2024-04-26 | 中国标准化研究院 | Dynamic data updating method and system based on service open platform |
| CN118101287A (en)* | 2024-03-04 | 2024-05-28 | 北京邮电大学 | A method for abnormal network traffic detection based on bidirectional generative adversarial network |
| CN118353667A (en)* | 2024-04-22 | 2024-07-16 | 云仓库(广东)信息科技有限公司 | Network security early warning method and system based on deep learning |
| CN118468988A (en)* | 2024-07-09 | 2024-08-09 | 浙江大学 | Terminal data leakage event prediction method and system based on horizontal federated learning |
| CN118503937A (en)* | 2024-07-18 | 2024-08-16 | 南京长深科技有限公司 | Educational administration management interaction method and system based on AI intelligent campus |
| CN118643511A (en)* | 2024-06-18 | 2024-09-13 | 深圳市润迅数智科技有限公司 | A federated learning optimization method based on privacy protection of neural network models |
| CN119293670A (en)* | 2024-12-10 | 2025-01-10 | 南京同力宇科技有限公司 | Data security management method and system applied to big data management platform |
| CN119311422A (en)* | 2024-12-12 | 2025-01-14 | 北京华鲲振宇智能科技有限责任公司 | A server data processing method based on deep learning model |
| CN119398341A (en)* | 2024-10-31 | 2025-02-07 | 河海大学 | Deep learning model analysis method and system for renewable energy power generation output prediction |
| CN119577604A (en)* | 2025-01-20 | 2025-03-07 | 浙江久婵物联科技有限公司 | Training and optimization method of intelligent prediction model for information transmission attack events |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111245793A (en)* | 2019-12-31 | 2020-06-05 | 西安交大捷普网络科技有限公司 | Method and device for analyzing abnormity of network data |
| CN112866185A (en)* | 2019-11-28 | 2021-05-28 | 海信集团有限公司 | Network traffic monitoring device and abnormal traffic detection method |
| CN113468521A (en)* | 2021-07-01 | 2021-10-01 | 哈尔滨工程大学 | Data protection method for federal learning intrusion detection based on GAN |
| CN113536382A (en)* | 2021-08-09 | 2021-10-22 | 北京理工大学 | Blockchain-based medical data sharing privacy-preserving method using federated learning |
| US20220138626A1 (en)* | 2020-11-02 | 2022-05-05 | Tsinghua University | System For Collaboration And Optimization Of Edge Machines Based On Federated Learning |
| CN116319061A (en)* | 2023-04-18 | 2023-06-23 | 天津市职业大学 | Intelligent control network system |
| CN116527362A (en)* | 2023-05-06 | 2023-08-01 | 北京邮电大学 | A Data Protection Method Based on LayerCFL Intrusion Detection |
| CN116541006A (en)* | 2023-06-28 | 2023-08-04 | 壹仟零壹艺网络科技(北京)有限公司 | Graphic processing method and device for computer man-machine interaction interface |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112866185A (en)* | 2019-11-28 | 2021-05-28 | 海信集团有限公司 | Network traffic monitoring device and abnormal traffic detection method |
| CN111245793A (en)* | 2019-12-31 | 2020-06-05 | 西安交大捷普网络科技有限公司 | Method and device for analyzing abnormity of network data |
| US20220138626A1 (en)* | 2020-11-02 | 2022-05-05 | Tsinghua University | System For Collaboration And Optimization Of Edge Machines Based On Federated Learning |
| CN113468521A (en)* | 2021-07-01 | 2021-10-01 | 哈尔滨工程大学 | Data protection method for federal learning intrusion detection based on GAN |
| CN113536382A (en)* | 2021-08-09 | 2021-10-22 | 北京理工大学 | Blockchain-based medical data sharing privacy-preserving method using federated learning |
| CN116319061A (en)* | 2023-04-18 | 2023-06-23 | 天津市职业大学 | Intelligent control network system |
| CN116527362A (en)* | 2023-05-06 | 2023-08-01 | 北京邮电大学 | A Data Protection Method Based on LayerCFL Intrusion Detection |
| CN116541006A (en)* | 2023-06-28 | 2023-08-04 | 壹仟零壹艺网络科技(北京)有限公司 | Graphic processing method and device for computer man-machine interaction interface |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117289141A (en)* | 2023-11-22 | 2023-12-26 | 深圳市麦迪瑞科技有限公司 | Electric bicycle charging state monitoring method based on artificial intelligence |
| CN117557270A (en)* | 2024-01-08 | 2024-02-13 | 深圳合纵富科技有限公司 | Mobile terminal secure payment management method and system |
| CN117557270B (en)* | 2024-01-08 | 2024-05-07 | 深圳合纵富科技有限公司 | Mobile terminal secure payment management method and system |
| CN118101287A (en)* | 2024-03-04 | 2024-05-28 | 北京邮电大学 | A method for abnormal network traffic detection based on bidirectional generative adversarial network |
| CN117931827A (en)* | 2024-03-21 | 2024-04-26 | 中国标准化研究院 | Dynamic data updating method and system based on service open platform |
| CN118353667A (en)* | 2024-04-22 | 2024-07-16 | 云仓库(广东)信息科技有限公司 | Network security early warning method and system based on deep learning |
| CN118643511A (en)* | 2024-06-18 | 2024-09-13 | 深圳市润迅数智科技有限公司 | A federated learning optimization method based on privacy protection of neural network models |
| CN118468988A (en)* | 2024-07-09 | 2024-08-09 | 浙江大学 | Terminal data leakage event prediction method and system based on horizontal federated learning |
| CN118468988B (en)* | 2024-07-09 | 2024-10-08 | 浙江大学 | Terminal data leakage event prediction method and system based on horizontal federated learning |
| CN118503937A (en)* | 2024-07-18 | 2024-08-16 | 南京长深科技有限公司 | Educational administration management interaction method and system based on AI intelligent campus |
| CN119398341A (en)* | 2024-10-31 | 2025-02-07 | 河海大学 | Deep learning model analysis method and system for renewable energy power generation output prediction |
| CN119293670A (en)* | 2024-12-10 | 2025-01-10 | 南京同力宇科技有限公司 | Data security management method and system applied to big data management platform |
| CN119311422A (en)* | 2024-12-12 | 2025-01-14 | 北京华鲲振宇智能科技有限责任公司 | A server data processing method based on deep learning model |
| CN119577604A (en)* | 2025-01-20 | 2025-03-07 | 浙江久婵物联科技有限公司 | Training and optimization method of intelligent prediction model for information transmission attack events |
| Publication number | Publication date |
|---|---|
| CN117056951B (en) | 2024-06-25 |
| Publication | Publication Date | Title |
|---|---|---|
| CN117056951B (en) | Data security management method for digital platform | |
| US11347867B2 (en) | Methods and apparatuses to evaluate cyber security risk by establishing a probability of a cyber-attack being successful | |
| Thuraisingham et al. | A data driven approach for the science of cyber security: Challenges and directions | |
| Al-Janabi | Pragmatic miner to risk analysis for intrusion detection (PMRA-ID) | |
| US20240372882A1 (en) | Systems and methods for detecting malicious network traffic using multi-domain machine learning | |
| US20230421582A1 (en) | Cybersecurity operations case triage groupings | |
| Abdullahi et al. | Comparison and investigation of AI-based approaches for cyberattack detection in cyber-physical systems | |
| Gill et al. | A systematic review on game-theoretic models and different types of security requirements in cloud environment: Challenges and opportunities | |
| Alhassan et al. | A fuzzy classifier-based penetration testing for web applications | |
| Kritika | A comprehensive literature review on ransomware detection using deep learning | |
| Sahu et al. | Federated LSTM Model for Enhanced Anomaly Detection in Cyber Security: A Novel Approach for Distributed Threat. | |
| Alotaibi | Network Intrusion Detection Model Using Fused Machine Learning Technique. | |
| Rane et al. | Adversarial Machine Learning for Cybersecurity Resilience and Network Security Enhancement | |
| Srivastava et al. | An effective computational technique for taxonomic position of security vulnerability in software development | |
| Gilbert et al. | The Impact of AI on Cybersecurity Defense Mechanisms: Future Trends and Challenges | |
| Bhaskara et al. | Causal effect analysis-based intrusion detection system for IoT applications | |
| Rani et al. | An optimized neural network with AdaHessian for cryptojacking attack prediction for Securing Crypto Exchange Operations of MEC applications | |
| Patel | Enhanced Network Security: Real-Time Malicious Traffic Detection in SD-WAN Using LSTM-GRU Hybrid Model | |
| Ye et al. | Zero-day vulnerability risk assessment and attack path analysis using security metric | |
| Tang et al. | An advanced data analytics approach to a cognitive cyber-physical system for the identification and mitigation of cyber threats in the medical internet of things (MIoT) | |
| Wei | Application of Bayesian algorithm in risk quantification for network security | |
| Hassan | Real-Time Risk Assessment in SaaS Payment Infrastructures: Examining Deep Learning Models and Deployment Strategies | |
| Li et al. | A threat recognition solution of edge data security in industrial internet | |
| Luo et al. | Harmless Backdoor-based Client-side Watermarking in Federated Learning | |
| Reynaud et al. | Review of eXplainable artificial intelligence for cybersecurity systems |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right | Effective date of registration:20240819 Address after:No. 102, Group 6, Yuechi Village, Xindian Town, Rudong County, Nantong City, Jiangsu Province, 226400 Patentee after:Nantong Haoxin Haoyi Intelligent Technology Co.,Ltd. Country or region after:China Address before:201210 room 302-09, building 10, No. 75, Zhangjiang Road, China (Shanghai) pilot Free Trade Zone, Pudong New Area, Shanghai Patentee before:Shanghai Haoxin Haoyi Intelligent Technology Co.,Ltd. Country or region before:China | |
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right | Effective date of registration:20241030 Address after:Room 210-2, Zone A, 2nd Floor, Building 1, No. 95, Lane 85, Cailun Road, China (Shanghai) Pilot Free Trade Zone, Pudong New Area, Shanghai, 200120 Patentee after:Shanghai Haoxin Haoyi Intelligent Technology Co.,Ltd. Country or region after:China Address before:No. 102, Group 6, Yuechi Village, Xindian Town, Rudong County, Nantong City, Jiangsu Province, 226400 Patentee before:Nantong Haoxin Haoyi Intelligent Technology Co.,Ltd. Country or region before:China |