Defence method and system based on active invisible defence strategy of artificial intelligence

Abstract

The application belongs to the technical field of data security, and discloses a defense method and system of an active invisible defense strategy based on artificial intelligence, which comprises the following steps: obtaining to-be-sent data of a sending end; performing sensitive identification on the to-be-sent data based on an artificial intelligence algorithm to obtain sensitive data corresponding to the to-be-sent data; converting the sensitive data based on a data conversion algorithm to obtain similar data; performing semantic confusion on the to-be-sent data by taking the similar data as confusion data based on a confusion algorithm to obtain confused data; randomly selecting an encryption algorithm from a pre-constructed encryption algorithm library, and encrypting the confused data based on the selected encryption algorithm to obtain encrypted data; compressing the encrypted data based on a compression algorithm to obtain compressed data; and sending the compressed data to a receiving end. The to-be-sent data is subjected to confusion, encryption and compression processing, so that the network security information security and digital security can be effectively guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of data security technology, specifically relating to a defense method and system based on an artificial intelligence-based proactive stealth defense strategy. Background Technology

[0002] With the popularization of the internet and the arrival of the big data era, network data security has become a major issue concerning the security and development of individuals, businesses, and even nations. Ensuring network data security means protecting corporate trade secrets and personal privacy from infringement.

[0003] Currently, network data security protection mainly adopts a passive defense strategy. The passive defense strategy mainly involves monitoring, recording and analyzing to identify potential threats and vulnerabilities, as well as taking countermeasures after a security incident occurs.

[0004] However, passive defense strategies still have at least the following problems in practical applications:

[0005] 1. Post-incident response: Passive defense strategies rely primarily on monitoring and logging, which only come into play after a security incident has occurred, rather than preventing attacks before they happen. This means that the attack may have already caused damage.

[0006] 2. Lack of initiative: Since it does not actively intervene in network traffic or data transmission, passive defense strategies cannot stop ongoing attacks and can only provide analysis and reporting after an attack occurs.

[0007] 3. Reliance on manual analysis: Passive defense systems typically require security experts to analyze data and logs, which can lead to delayed responses, especially when resources are limited;

[0008] 4. High false alarm rate: Some passive defense systems may generate a large number of false alarms, which consumes the security team's time and resources and may cause real threats to be overlooked;

[0009] 5. Inability to cope with unknown threats: Passive defense strategies are usually based on known security rules and patterns, so they may not be able to effectively identify and defend against new or unknown attack methods.

[0010] In summary, existing passive defense strategies suffer from problems such as high false alarm rates, lack of initiative, and high dependence on human intervention. Summary of the Invention

[0011] The purpose of this invention is to provide a defense method and system based on an artificial intelligence-based active stealth defense strategy, in order to solve the problems of high false alarm rate, lack of initiative and high dependence on human intervention in existing passive defense strategies.

[0012] To achieve the above objectives, the present invention adopts the following technical solution:

[0013] In a first aspect, the present invention provides a defense method based on an active stealth defense strategy using artificial intelligence, the method comprising:

[0014] Get the data to be sent from the sender;

[0015] Based on artificial intelligence algorithms, sensitive data is obtained by identifying the sensitive data corresponding to the data to be sent.

[0016] Sensitive data is transformed using data transformation algorithms to obtain similar data;

[0017] Based on the obfuscation algorithm, similar data is used as obfuscation data to perform semantic obfuscation on the data to be sent, resulting in obfuscated data.

[0018] Randomly select an encryption algorithm from a pre-built encryption algorithm library, and encrypt the obfuscated data based on the selected encryption algorithm to obtain the encrypted data;

[0019] The encrypted data is compressed using a compression algorithm to obtain compressed data.

[0020] The compressed data is sent to the receiving end.

[0021] Preferably, the encryption algorithms in the encryption algorithm library include at least the SM2 algorithm, SM3 algorithm, SM4 algorithm, and ZUK algorithm.

[0022] Preferably, the artificial intelligence algorithm is a deep learning algorithm; based on the artificial intelligence algorithm, sensitive data is obtained by performing sensitive identification on the data to be sent, including:

[0023] Initialize the model parameters of the deep learning algorithm;

[0024] The model parameters of the deep learning algorithm are trained based on the pre-built training set to obtain a trained deep learning model.

[0025] The data to be sent is input into a trained deep learning model for sensitive prediction, and sensitive data is obtained.

[0026] Preferably, the data transformation algorithm is an improved generative adversarial network algorithm, which transforms sensitive data to obtain similar data, including:

[0027] Random noise is acquired, and a generator based on an improved generative adversarial network algorithm maps the random noise to the sample space of sensitive data, generating discrete and continuous data.

[0028] A discriminator based on an improved generative adversarial network algorithm optimizes discrete and continuous data so that the distribution of discrete and continuous data matches the true distribution of sensitive data, thus obtaining optimized discrete and continuous data.

[0029] Similar data are constructed based on optimized discrete and continuous data.

[0030] Preferably, the generator of the improved generative adversarial network algorithm includes: a first generator, the first generator being used for:

[0031] Extract the categories of sensitive data, and create initial discrete features based on the categories of sensitive data and random noise;

[0032] The initial discrete features are encoded to obtain an encoding vector;

[0033] The encoded vector is simulated based on a pre-constructed polynomial to obtain a parameterized vector;

[0034] Calculate the probability distribution of the parameterized vector and sample the probability distribution of the parameterized vector to obtain the probability vector;

[0035] Convert the probability vector into a continuous vector;

[0036] Decoding continuous vectors yields the final discrete features;

[0037] Discrete data is generated based on the final discrete features.

[0038] Preferably, the generator of the improved generative adversarial network algorithm includes: a second generator; wherein the second generator is used for:

[0039] Extract the categories of sensitive data, and directly map random noise into the sample space of sensitive data based on the categories of sensitive data to generate continuous data.

[0040] Preferably, similar data is used as obfuscated data to semantically obfuscate the data to be sent, resulting in obfuscated data, including:

[0041] By using similar data to replace sensitive data in the data to be sent, obfuscated data is obtained.

[0042] Secondly, the present invention provides a defense system based on an artificial intelligence-based active stealth defense strategy, used to implement the aforementioned defense method based on an artificial intelligence-based active stealth defense strategy, the system comprising:

[0043] The data acquisition module is used to acquire the data to be sent from the sending end;

[0044] The sensitive identification module is used to identify the sensitive data to be sent based on artificial intelligence algorithms, and obtain the sensitive data corresponding to the data to be sent.

[0045] The data transformation module is used to transform sensitive data based on data transformation algorithms to obtain similar data;

[0046] The data obfuscation module is used to semantically obfuscate the data to be sent based on obfuscation algorithms and similar data as obfuscated data, so as to obtain obfuscated data.

[0047] The data encryption module is used to randomly select an encryption algorithm from a pre-built encryption algorithm library, and encrypt the obfuscated data based on the selected encryption algorithm to obtain the encrypted data;

[0048] The data compression module is used to compress encrypted data based on a compression algorithm to obtain compressed data;

[0049] The data sending module is used to send compressed data to the receiving end.

[0050] Thirdly, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned defense method based on an artificial intelligence-driven active stealth defense strategy.

[0051] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned defense method based on an artificial intelligence-based active stealth defense strategy.

[0052] Beneficial effects:

[0053] 1. This invention employs proactive defense strategies such as obfuscation, encryption, and compression of the data to be sent after acquiring it from the sending end. This prevents attackers from detecting the device, the protected device, and the network (sending end), and from stealing the transmitted data, thus effectively protecting network security, information security, and digital security.

[0054] 2. The proactive defense strategy of this invention can reduce manual analysis and reduce false alarm rate. Attached Figure Description

[0055] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0056] Figure 1This is a flowchart of a defense method based on an artificial intelligence-based active stealth defense strategy provided in one embodiment of the present invention;

[0057] Figure 2 This is a block diagram of a defense system based on an artificial intelligence-based active stealth defense strategy provided in one embodiment of the present invention. Detailed Implementation

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the accompanying drawings is only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0059] Example 1

[0060] Figure 1 This is a flowchart of a defense method based on an artificial intelligence-based proactive stealth defense strategy, provided by one embodiment of the present invention. The application scenario of this AI-based proactive stealth defense method includes: a user terminal and a defense terminal. The user terminal includes, but is not limited to, mobile phones, PCs, industrial control terminals, and application servers. Each user terminal deploys one defense terminal, and the user terminal communicates with its corresponding defense terminal. All defense terminals are connected via the Internet. The AI-based proactive stealth defense method of this embodiment runs on the defense terminal. The execution steps of the method on the defense terminal are as follows: Figure 1 As shown, the execution steps include the following:

[0061] Step S10: Obtain the data to be sent from the sending end. In this embodiment, when two user terminals need to communicate, one user terminal acts as the sending end and the other user terminal acts as the receiving end. The sending end sends the corresponding data to the locally deployed defense terminal. After receiving the data, the defense terminal treats it as data to be sent and performs subsequent obfuscation, encryption, and compression processing on the data to be sent.

[0062] Step S20: Based on artificial intelligence algorithms, perform sensitivity identification on the data to be sent to obtain the sensitive data corresponding to the data to be sent.

[0063] In this embodiment, the data to be sent contains a large amount of sensitive data, such as personal identification information, corporate financial information, and trade secrets. If this data is illegally attacked or stolen, it can lead to privacy violations, economic losses, and other consequences. Therefore, it is necessary to protect this sensitive data. In this embodiment, an artificial intelligence algorithm is used to identify the sensitive data in the data to be sent for subsequent processing. The artificial intelligence algorithm is a deep learning algorithm; therefore, based on the artificial intelligence algorithm, sensitive data identification is performed on the data to be sent to obtain the sensitive data corresponding to the data to be sent, including:

[0064] Step a10: Initialize the model parameters of the deep learning algorithm; the main model parameters of the deep learning algorithm include: weights, biases, number of layers, and number of neurons in each layer.

[0065] In a neural network, each connection (edge) has a corresponding weight, which determines how information is transmitted in the network.

[0066] Each neuron typically also has a bias term, which is added to the weights and affects the neuron's activation state.

[0067] The number of layers in a neural network is also a hyperparameter, which determines the depth of the model.

[0068] Step a20: Train the model parameters of the deep learning algorithm based on the pre-built training set to obtain a trained deep learning model. The training set consists of a large amount of sensitive sample data and corresponding labels, and is used to train the model parameters of the deep learning algorithm.

[0069] Step a30: Input the data to be sent into the trained deep learning model for sensitive prediction to obtain sensitive data; the trained deep learning model is mainly used to classify the data to be sent into sensitive data and non-sensitive data, so as to achieve sensitive prediction of the data to be sent.

[0070] Step S30: Transform the sensitive data based on the data transformation algorithm to obtain similar data.

[0071] In this embodiment, sensitive data is transformed into data with a certain degree of similarity to the sensitive data, also known as fake data. This fake data can be mixed and recombined with the sensitive data to obfuscate the sensitive data and effectively improve the security of the sensitive data.

[0072] In this embodiment, the method of converting sensitive data into data with a certain degree of similarity can employ a generative adversarial network (GAN) algorithm. The GAN algorithm includes a generator and a discriminator. The generator aims to generate sufficiently realistic data to deceive the discriminator, i.e., it receives a random noise vector as input and outputs a data sample with a distribution similar to the real data. The discriminator aims to distinguish between real data and fake data generated by the generator, i.e., it receives data samples from the generator or the real dataset and outputs a scalar representing the probability that the sample is real data.

[0073] Because sensitive data contains both discrete and continuous numerical features, and discrete numerical features suffer from non-differentiable processing issues, the gradient during backpropagation from the discriminator to the generator is often zero. Therefore, discrete features can lead to noise and low-quality data generated by traditional generative adversarial networks (GANs), and can also cause instability in discriminator network training, potentially resulting in model collapse. Therefore, the data transformation algorithm described in this embodiment is an improved GAN algorithm. This improved GAN algorithm has two generators: a first generator and a second generator. The first generator transforms the discrete numerical features in the sensitive data to obtain discrete data; the second generator transforms the continuous numerical features in the sensitive data to obtain continuous data.

[0074] Therefore, sensitive data is transformed using data transformation algorithms to obtain similar data, including:

[0075] Step b10: Obtain random noise. The generator based on the improved generative adversarial network algorithm maps the random noise into the sample space of the sensitive data to generate discrete and continuous data.

[0076] The first generator is used to: extract the categories of sensitive data, and create initial discrete features based on the categories of sensitive data and random noise; wherein the random noise is Gaussian white noise; and the categories of sensitive data are, for example, financial, trade secret, identity information, etc.

[0077] The initial discrete features are encoded to obtain an encoding vector. In this embodiment, the one-hot encoding algorithm is used to convert the initial discrete features into a one-hot encoded form, so that the initial discrete features are converted into a d-dimensional one-hot vector, i.e., the encoding vector.

[0078] The encoded vector is simulated based on a pre-constructed polynomial to obtain a parameterized vector; that is, a parameterized polynomial distribution is used to simulate a one-hot vector. The parameters of this distribution can be obtained through a sigmoid function, the input of which is a d-dimensional vector.

[0079] The functional expression of the parameterized vector is:

[0080] ;

[0081] In the formula, This represents the encoded vector of the i-th dimension. This represents the i-th dimension parameterized vector corresponding to the i-th dimension encoded vector. Represents trainable parameters, when The closer the value is to 0, the more similar the generated discrete data is to the original data; exp() represents the exponential function. It represents the continuous probability distribution probability of the i-th dimension encoding vector, which is calculated using the continuous probability distribution function (Gumbel).

[0082] Calculate the probability distribution of the parameterized vector, and sample the probability distribution of the parameterized vector to obtain a probability vector; where the functional expression of the probability distribution of the parameterized vector is:

[0083] ;

[0084] In the formula, This is the sigmoid function.

[0085] The probability vector is converted into a continuous vector, the continuous vector is decoded to obtain the final discrete features, and discrete data is generated based on the final discrete features.

[0086] In this embodiment, the first generator can solve the problem that discrete numerical features are not differentiable, which can easily lead to the gradient of backpropagation from the discriminator to the generator always being zero. This improves the stability of the discriminator network training and the quality of similar data generation.

[0087] The data processing steps of the second generator are the same as those of the generator in the traditional generative adversarial network algorithm. That is, the second generator is used to: extract the category of sensitive data, and directly map random noise to the sample space of sensitive data based on the category of sensitive data to generate continuous data.

[0088] Step b20: The discriminator based on the improved generative adversarial network algorithm optimizes the discrete and continuous data so that the distribution of the discrete and continuous data reaches the true distribution of the sensitive data, thus obtaining the optimized discrete and continuous data.

[0089] Step b30: Construct similar data based on the optimized discrete and continuous data.

[0090] In this embodiment, by concatenating the optimized discrete data with the optimized continuous data, similar data can be obtained.

[0091] Step S40: Based on the obfuscation algorithm, use similar data as obfuscation data to perform semantic obfuscation on the data to be sent, and obtain the obfuscated data.

[0092] As a further optimization of this embodiment, similar data is used as obfuscated data to semantically obfuscate the data to be sent, and obfuscated data is obtained. This includes replacing sensitive data in the data to be sent with similar data to obtain obfuscated data.

[0093] Step S50: Randomly select an encryption algorithm from the pre-built encryption algorithm library, and encrypt the obfuscated data based on the selected encryption algorithm to obtain the encrypted data. In this embodiment, the encryption algorithm library contains multiple encryption algorithms, such as, but not limited to, SM2, SM3, SM4, and ZUK algorithms. Randomly selecting an encryption algorithm to encrypt the obfuscated data each time has the following advantages:

[0094] 1. Enhanced Security: Randomly selecting an encryption algorithm reduces an attacker's ability to predict encryption strategies, thereby improving data security. If an attacker doesn't know which algorithm is being used, they would need to try multiple methods to crack it, significantly increasing the difficulty of the attack.

[0095] 2. Preventing targeted attacks: Some encryption algorithms may have known vulnerabilities that attackers may exploit. Randomly selecting algorithms can reduce the risk of targeted attacks.

[0096] 3. Increase diversity: Using different encryption algorithms can increase the diversity of the system, so that even if one algorithm is cracked, other data will still remain secure.

[0097] 4. Preventing pattern recognition: If the same encryption algorithm is always used, identifiable patterns may form in the encrypted data, which may be exploited by attackers. Randomly selecting an algorithm can reduce this risk.

[0098] 5. Increased cracking costs: In order to crack the data, attackers may need to try different cracking methods for each possible encryption algorithm, which will significantly increase their time and resource costs.

[0099] In this embodiment, the SM2 (ShangMi, abbreviated as SM, representing "commercial cryptography") algorithm is a public-key cryptography algorithm standard issued by the State Cryptography Administration of China. It is based on elliptic curve cryptography (ECC). The main components of the SM2 algorithm are as follows:

[0100] Elliptic curve parameters: The SM2 algorithm defines a standard set of elliptic curve parameters, including the prime number p, the coefficients a and b of the elliptic curve equation, and the coordinates of the base point G;

[0101] Key generation: The user selects a random number as the private key, and then generates the corresponding public key through dot product operation on an elliptic curve.

[0102] Digital signature: A message is signed using a private key, and the signature can be verified using a public key;

[0103] Key exchange: Two users can securely exchange keys using dot product on an elliptic curve.

[0104] Encryption and decryption: Encrypt data using the public key and decrypt data using the private key.

[0105] In this embodiment, the SM3 algorithm is a cryptographic hash algorithm standard developed in China, and the SM4 algorithm is a block cipher algorithm standard developed in China, and the SM4 block cipher algorithm is one of a series of cryptographic standards issued by the State Cryptography Administration (SCAM) of the People's Republic of China. Together with the SM2 elliptic curve public key cryptography algorithm and the SM4 block cipher algorithm, they constitute China's commercial cryptography system.

[0106] Step S60: Compress the encrypted data based on the compression algorithm to obtain compressed data.

[0107] In this embodiment, the compression algorithm can be bzip2, Huffman coding, or Run-Length Encoding (RLE) to reduce the data size and improve the efficiency of data transmission.

[0108] Step S70: Send the compressed data to the receiving end.

[0109] In this embodiment, when the compressed data is sent to the receiving end, the defense terminal of the receiving end performs decompression, decryption and deobfuscation operations on the compressed data in sequence after receiving the compressed data to obtain the original data, that is, the data to be sent by the sending end, and then sends the original data to the corresponding user end.

[0110] This invention employs proactive defense strategies, such as obfuscation, encryption, and compression, after acquiring the data to be sent from the sending end. This prevents attackers from probing the device, the protected device, and the network (sending end), and from stealing the transmitted data, effectively safeguarding network security, information security, and digital security. Furthermore, this proactive defense strategy reduces manual analysis and false alarm rates.

[0111] Example 2

[0112] Figure 2 This is a block diagram of a defense system based on an artificial intelligence-driven active stealth defense strategy, provided in one embodiment of the present invention. Figure 2 As shown, this embodiment provides a defense system based on an artificial intelligence-based proactive stealth defense strategy, used to implement the defense method based on the artificial intelligence-based proactive stealth defense strategy in Embodiment 1. The system includes:

[0113] The data acquisition module is used to acquire the data to be sent from the sending end;

[0114] The sensitive identification module is used to identify the sensitive data to be sent based on artificial intelligence algorithms, and obtain the sensitive data corresponding to the data to be sent.

[0115] The data transformation module is used to transform sensitive data based on data transformation algorithms to obtain similar data;

[0116] The data obfuscation module is used to semantically obfuscate the data to be sent based on obfuscation algorithms and similar data as obfuscated data, so as to obtain obfuscated data.

[0117] The data encryption module is used to randomly select an encryption algorithm from a pre-built encryption algorithm library, and encrypt the obfuscated data based on the selected encryption algorithm to obtain the encrypted data;

[0118] The data compression module is used to compress encrypted data based on a compression algorithm to obtain compressed data;

[0119] The data sending module is used to send compressed data to the receiving end.

[0120] This embodiment also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the defense method of the active stealth defense strategy based on artificial intelligence in Embodiment 1.

[0121] This embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the defense method of the artificial intelligence-based active stealth defense strategy in Embodiment 1.

[0122] This invention obtains the data to be sent from the sending end and then obfuscates, encrypts, and compresses it, making it impossible for attackers to probe the device, the protected device, and the network (sending end), and to steal the transmitted data. This effectively protects network security, information security, and digital security.

[0123] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0124] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0125] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A defense method based on an artificial intelligence-driven active stealth defense strategy, characterized in that, The method includes: Get the data to be sent from the sender; Based on artificial intelligence algorithms, sensitive data is obtained by identifying the sensitive data corresponding to the data to be sent. Sensitive data is transformed using a data transformation algorithm to obtain similar data. The data transformation algorithm is an improved generative adversarial network algorithm, including: Random noise is acquired, and a generator based on an improved generative adversarial network algorithm maps the random noise to the sample space of sensitive data, generating discrete and continuous data. A discriminator based on an improved generative adversarial network algorithm optimizes discrete and continuous data so that the distribution of discrete and continuous data matches the true distribution of sensitive data, thus obtaining optimized discrete and continuous data. Similar data are constructed based on optimized discrete and continuous data; Based on the obfuscation algorithm, similar data is used as obfuscation data to perform semantic obfuscation on the data to be sent, resulting in obfuscated data. Randomly select an encryption algorithm from a pre-built encryption algorithm library, and encrypt the obfuscated data based on the selected encryption algorithm to obtain the encrypted data; The encrypted data is compressed using a compression algorithm to obtain compressed data. The compressed data is sent to the receiving end.

2. The defense method based on artificial intelligence-driven active stealth defense strategy according to claim 1, characterized in that, The encryption algorithms in the encryption algorithm library include at least the SM2 algorithm, SM3 algorithm, SM4 algorithm, and ZUK algorithm.

3. The defense method based on artificial intelligence-driven active stealth defense strategy according to claim 1, characterized in that, The artificial intelligence algorithm is a deep learning algorithm. Based on the artificial intelligence algorithm, sensitive data is identified in the data to be sent to obtain the sensitive data corresponding to the data to be sent, including: Initialize the model parameters of the deep learning algorithm; The model parameters of the deep learning algorithm are trained based on the pre-built training set to obtain a trained deep learning model. The data to be sent is input into a trained deep learning model for sensitive prediction, and sensitive data is obtained.

4. The defense method based on artificial intelligence-driven active stealth defense strategy according to claim 1, characterized in that, The generator of the improved generative adversarial network algorithm includes: a first generator, which is used for: Extract the categories of sensitive data, and create initial discrete features based on the categories of sensitive data and random noise; The initial discrete features are encoded to obtain an encoding vector; The encoded vector is simulated based on a pre-constructed polynomial to obtain a parameterized vector; Calculate the probability distribution of the parameterized vector and sample the probability distribution of the parameterized vector to obtain the probability vector; Convert the probability vector into a continuous vector; Decoding continuous vectors yields the final discrete features; Discrete data is generated based on the final discrete features.

5. The defense method based on artificial intelligence-driven active stealth defense strategy according to claim 1, characterized in that, The generator of the improved generative adversarial network algorithm further includes: a second generator, which is used for: Extract the categories of sensitive data, and directly map random noise into the sample space of sensitive data based on the categories of sensitive data to generate continuous data.

6. The defense method based on the active stealth defense strategy according to claim 1, characterized in that, Using similar data as obfuscation data, semantic obfuscation is performed on the data to be sent, resulting in obfuscated data, including: By using similar data to replace sensitive data in the data to be sent, obfuscated data is obtained.

7. A defense system based on an artificial intelligence-based active stealth defense strategy, used to implement the defense method based on the artificial intelligence-based active stealth defense strategy according to any one of claims 1-6, characterized in that, The system includes: The data acquisition module is used to acquire the data to be sent from the sending end; The sensitive identification module is used to identify the sensitive data to be sent based on artificial intelligence algorithms, and obtain the sensitive data corresponding to the data to be sent. The data transformation module is used to transform sensitive data based on data transformation algorithms to obtain similar data; The data obfuscation module is used to semantically obfuscate the data to be sent based on obfuscation algorithms and similar data as obfuscated data, so as to obtain obfuscated data. The data encryption module is used to randomly select an encryption algorithm from a pre-built encryption algorithm library, and encrypt the obfuscated data based on the selected encryption algorithm to obtain the encrypted data; The data compression module is used to compress encrypted data based on a compression algorithm to obtain compressed data; The data sending module is used to send compressed data to the receiving end.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the defense method of the active stealth defense strategy based on artificial intelligence as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the defense method of the active stealth defense strategy based on artificial intelligence as described in any one of claims 1-6.

Images