CN117592114A - Network parallel simulation oriented data desensitization method, system and readable storage medium - Google Patents
Network parallel simulation oriented data desensitization method, system and readable storage medium Download PDFInfo
- Publication number
- CN117592114A CN117592114A CN202410078407.XA CN202410078407A CN117592114A CN 117592114 A CN117592114 A CN 117592114A CN 202410078407 A CN202410078407 A CN 202410078407A CN 117592114 A CN117592114 A CN 117592114A
- Authority
- CN
- China
- Prior art keywords
- network
- simulation
- generator
- sample
- real
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004088 simulation Methods 0.000 title claims abstract description 52
- 238000000586 desensitisation Methods 0.000 title claims abstract description 41
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000003860 storage Methods 0.000 title claims abstract description 14
- 238000004140 cleaning Methods 0.000 claims abstract description 5
- 238000000605 extraction Methods 0.000 claims abstract description 5
- 238000010606 normalization Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000013256 Gubra-Amylin NASH model Methods 0.000 claims description 2
- 238000009826 distribution Methods 0.000 abstract description 6
- 238000007781 pre-processing Methods 0.000 abstract description 2
- 238000011160 research Methods 0.000 abstract description 2
- 230000035945 sensitivity Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 9
- 238000013461 design Methods 0.000 description 6
- 230000006399 behavior Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004883 computer application Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011076 safety test Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0475—Generative networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/094—Adversarial learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Bioethics (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Computer Security & Cryptography (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computer Hardware Design (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a network parallel simulation-oriented data desensitization method, a system and a readable storage medium. The invention carries out preprocessing operations such as data cleaning, feature extraction, normalization and the like on the original data by collecting the original data of the parallel wired network, and trains and generates a generator and a discriminator model of the countermeasure network. The input random noise is converted into an approximately real analog sample through a generator, and a discriminator is responsible for distinguishing the real sample from the generated analog sample; and then, inputting random noise by using the trained generator model to generate synthesized simulation samples, wherein the simulation samples can represent the distribution characteristics of the original wired network data. The finally generated simulation sample can be used for sharing, research and other purposes, and the sensitivity and risk of a real sample are reduced.
Description
Technical Field
The invention relates to the field of information security, in particular to a data desensitization method and system for network parallel simulation and a corresponding computer readable storage medium.
Background
A network parallel simulation system is a tool for simulating and evaluating computer network performance, behavior, and interactions. It provides a controllable simulation environment by simulating various factors and elements in the actual network environment so as to perform network design, optimization and security assessment. Considering that the network parallel simulation system is designed and realized by a third party manufacturer in many cases, the simulated data needs to be desensitized for the safety of the real network data.
The traditional data desensitization method mainly comprises anonymization, encryption, generalization of data, desensitization replacement and the like, and mainly adopts the principle that sensitive information in original data is removed by modifying or deleting the sensitive information, and rules or algorithms are used for transforming the data and the like, so that risks of data distortion, information loss and privacy disclosure exist. In addition, in order to ensure the accuracy of the simulation effect in the network parallel simulation system, the data needs to be decrypted once after the data is input, and a certain leakage risk exists.
Disclosure of Invention
The invention aims at: aiming at all or part of the problems, the data desensitization solution for network parallel simulation is provided, and the simulation is performed by directly generating a simulation sample to replace a real sample, so that the protection of sensitive information is ensured, the unified characteristics of the simulation data are consistent with those of the real sample, namely, the problems of data distortion and information loss of simulation data are avoided.
The technical scheme adopted by the invention is as follows:
a network parallel simulation oriented data desensitization method, the method comprising:
acquiring network parameter information of a real network;
sub-graph coding is carried out on the network parameter information, and upper triangle adjacency matrix representation is carried out on the sub-graph;
training a pre-constructed network model by utilizing the subgraph, the upper triangle adjacency matrix and random noise to obtain a network model of input random noise and output simulation samples, wherein the pre-constructed network model comprises a generator and a discriminator;
and (3) inputting random noise to generate a simulation sample by using the trained network model, so as to obtain desensitization data.
Preferably, the acquiring network parameter information of the real network includes:
collecting raw data representing network behavior and characteristics of a real network;
and sequentially carrying out data cleaning, feature extraction and normalization on the original data.
Preferably, the network parameter information includes node location, link bandwidth, node type, link type, and delay.
Preferably, the sub-picture encoding of the network parameter information includes:
creating the network parameter information into a plurality of closed subgraphs by using a full connected neural network;
traversing each sub-graph, and acquiring a node set of each sub-graph according to the connection relation and the topological structure between the nodes;
node encoding was performed using the Weisfeiler-Lehman algorithm in WL-GAN.
Preferably, the pre-constructed network model is a WL-GAN model.
Preferably, the loss function of the generator is:
gloss= -mean (D (G (z))); wherein,
z is random noise input by the generator;
g (z) is a simulated sample generated by the generator;
d (G (z)): the discriminator judges the result of the simulation sample generated by the generator;
mean: and an operator for averaging, which is used for calculating the average judging result of the judging device on the plurality of simulation samples generated by the generator.
Preferably, the loss function of the discriminator is:
d_loss=mean (D (y)) -mean (D (G (z))); wherein,
y is a real sample;
d (y): representing the discrimination result of the discriminator on the real sample y.
The invention also provides a computer readable storage medium storing a computer program which is run to perform the above-described network parallel simulation oriented data desensitization method.
The invention also provides a data desensitization system facing the network parallel simulation, which comprises a processor and the computer readable storage medium, wherein the processor is configured to run a computer program stored in the computer readable storage medium so as to execute a data desensitization method facing the network parallel simulation.
In summary, due to the adoption of the technical scheme, the beneficial effects of the invention are as follows:
1. the data desensitization method based on the generation of the countermeasure network can generate the synthesized data similar to the real sample, and maintains the distribution characteristics and the statistical characteristics of the original data. This helps to improve the accuracy and authenticity of the simulation results.
2. The invention reduces the risk of sensitive information leakage. An attacker cannot infer sensitive information in the original data through the synthesized data, so that the privacy protection effect is improved.
3. The synthesized data generated by the invention keeps the characteristics and distribution of the original data, can be used for data analysis, modeling and prediction more widely, and improves the understanding of network flow behaviors and the functions of a simulation system.
Drawings
The invention will now be described by way of example and with reference to the accompanying drawings in which:
FIG. 1 is a flow chart of a method of data desensitization.
Detailed Description
All of the features disclosed in this specification, or all of the steps in a method or process disclosed, may be combined in any combination, except for mutually exclusive features and/or steps.
Any feature disclosed in this specification (including any accompanying claims, abstract) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. That is, each feature is one example only of a generic series of equivalent or similar features, unless expressly stated otherwise.
The invention carries out preprocessing operations such as data cleaning, feature extraction, normalization and the like on the original data by collecting the original data of the parallel wired network, and trains and generates a generator and a discriminator model of the countermeasure network. The input random noise is converted into an approximately real analog sample through the generator, and the discriminator is responsible for distinguishing the real sample from the generated analog sample. And then, inputting random noise by using the trained generator model to generate synthesized simulation samples, wherein the simulation samples can represent the distribution characteristics of the original wired network data. The finally generated simulation sample can be used for sharing, research and other purposes, and the sensitivity and risk to the real sample are reduced.
Specifically, as shown in fig. 1, the method of the present invention comprises the steps of:
(1) Training data set preparation step
The method comprises the steps of acquiring network parameter information from a real wired network as a real network data set, converting the network parameter information into a plurality of closed subgraphs based on the structural characteristics of a full-connected neural network learning graph by utilizing a subgraph coding thought and based on the subgraph adjacency matrix, coding the subgraphs, and establishing an upper triangular adjacency matrix for each node to be used as a typical network data set.
1) And acquiring network parameter information of the real network.
The set of network parameter information collected/obtained from the real wired network is typically typical configuration parameters including node location, link bandwidth, node type, link type, delay, sensitive content, etc.
The method comprises the steps of monitoring network flow, recording transmission data, acquiring equipment state and the like, collecting original data representing real network behaviors and characteristics, and then sequentially carrying out data cleaning, characteristic extraction and normalization processing to obtain the network parameter information.
2) And converting the network parameter information into a plurality of closed subgraphs, and encoding the closed subgraphs.
Network parameter information is created as a number of closed subgraphs using the structural characteristics of the full-connected neural network learning graph.
Traversing each sub-graph, and acquiring a node set of each sub-graph according to the connection relation and the topological structure between the nodes.
And (3) performing node coding by using a Weisfeiler-Lehman algorithm in the WL-GAN, and extracting the topological structure and the local information of the node by iteratively updating the label of the node to obtain a coding set for representing the sub-graph node and the characteristics thereof. Alternatively, one-Hot encoding may be used to encode each node, and the node attributes may be represented numerically based on information such as the node type and position.
For example, the node is One-Hot encoded according to the node type. The node type comprises a server, a router, a switch and terminal equipment, and is respectively represented by binary codes with the length of 6, wherein only one element is 1, and the rest elements are 0.
Example data set:
sub-graph 1:
nodes 1, 2, 3, 4, 7
Sub-graph 2:
nodes 5, 6, 8, 9, 10
Example subgraph coding:
sub-picture 1 coding:
node 1 encodes: [1, 0, 0, 0, 0, 0] # server node
Node 2 encodes: [0, 1, 0, 0, 0, 0] # router node
Node 3 encodes: [0, 1, 0, 0, 0, 0] # router node
Node 4 encodes: [0, 1, 0, 0, 0, 0] # router node
Node 7 encodes: [0, 1, 0, 0, 0, 0] # router node
Sub-picture 2 coding:
node 5 encodes: [0, 0, 1, 0, 0, 0] # switch node
Node 6 encodes: [0, 0, 1, 0, 0, 0] # switch node
Node 8 encodes: [0, 0, 1, 0, 0, 0] # switch node
Node 9 encodes: [0, 0, 1, 0, 0, 0] # switch node
Node 10 encodes: [0, 0, 0, 0, 1] # end device node
3) Upper triangle adjacency matrix representation subgraph
The upper triangle adjacency matrix is a symmetric matrix used to represent the edge connection in the subgraph, wherein the elements of the matrix represent whether an edge connection exists between two nodes. And constructing an upper triangular adjacency matrix for each coded subgraph according to adjacency relations among the nodes, wherein the upper triangular adjacency matrix is used for representing the topological structure of the subgraph.
The upper triangular adjacency matrix of some of the nodes in the example sub-graph 1 above is represented as follows:
sub-picture 1 coding:
node 1 encodes: [0, 1, 1, 1, 1] # node 1 is connected to nodes 2, 3, 4, 7
Node 2 encodes: [0, 0, 1, 0, 0] # node 2 is connected to node 3
Node 3 encodes: [0, 0, 0, 0] # node 3 has no connected nodes
Node 4 encodes: [0, 0, 0, 0, 0] # node 4 has no connected nodes
(2) Network model structural design step (if the step can be completed by constructing the network model in advance)
A network model structure is determined that generates the countermeasure network. WL-GAN network model structure, as used herein, includes a generator and a discriminator. The generator is responsible for generating simulated network data, and the discriminator is used to determine whether the generated data is a real sample or a simulated sample. The design of the generator and discriminator includes architectural and hierarchical designs. The following is the WL-GAN network model structural design.
1) Generator (Generator):
input: random noise and desensitization parameters (e.g., data clearance, noise level, etc.). The desensitization parameters are mainly used to control the behavior of the generator during training to balance data privacy and data quality. In particular, the desensitization parameters may be used to adjust the intensity of noise additions, data perturbations, or other desensitization techniques in the generator, by which the characteristics of the generator to generate analog samples may be changed, thereby better protecting the private information. After training is finished, no further input of desensitization parameters may be required when generating the simulated sample, because the generator has learned and adjusted the desensitization parameters by training, and a simulated sample with realistic statistical properties may be generated without input of the desensitization parameters.
And (3) outputting: the generated analog samples replace the original sensitive data.
Loss function: g_loss= -mean (D (G (z))).
Where D (G (z)) represents the discrimination result of the discriminator on the pseudo-graph generated by the generator.
mean: an operator representing the averaging for calculating an average discrimination result of the discriminator on the plurality of dummy samples generated by the generator.
D (G (z)): representing the discrimination result of the discriminator on the false sub-graph G (z) generated by the generator, i.e. scoring the generated false samples.
And G (z) is a false sub-graph generated by the generator.
And z is random noise at the input of the generator. In a wired network, it can be understood that some configuration parameters can be set according to specific requirements, so as to generate a false sample (i.e. an analog sample) similar to a real sample.
By calculating the average discrimination result D (G (z)) of the discriminator on the plurality of false samples generated by the generator, and then taking the negative sign, i.e., -mean (D (G (z))), the loss function g_loss of the generator indicates that the generator wishes to maximize the average false discrimination result of the discriminator on the generated false samples. The generator can gradually generate more realistic false samples to fool the discriminant and make it impossible to accurately distinguish between true and false samples generated.
2) Discriminator (Discriminator):
input: a real sample or a simulated sample generated by a generator.
And (3) outputting: binary values, representing the probability that the input data is a real sample (1 representing a real sample, 0 representing an analog sample).
Loss function: d_loss=mean (D (y)) -mean (D (G (z))).
When the objective function expression is combined with the wired network configuration parameters, the relevant parameters in the objective function of the WL-GAN need to be corresponding to the wired network configuration parameters.
D_loss: representing the loss function of the arbiter for measuring the performance of the arbiter. By constantly optimizing D_loss, the arbiter can accurately distinguish between a real sample and a false sample generated by the generator.
mean: an operator representing the averaging for calculating an average discrimination result of the discriminator for the plurality of dummy samples.
D (y): and representing the result of the discrimination of the discriminator on the real sub-image sample y, namely scoring or probability of the real sub-image sample. The task of the arbiter is to distinguish a real sample from a generated false sample and evaluate its authenticity.
mean (D (y)): representing the average discrimination result of the discriminator on a plurality of real samples. The overall evaluation result of the discriminator on the real sample can be obtained by calculating the score or the average value of the probability of the discriminator on the real sample.
D (G (z)): representing the discrimination result of the discriminator on the false sub-picture sample G (z) generated by the generator, i.e. scoring the generated false sample. The task of the arbiter is to judge the authenticity of the generated false samples and to give a corresponding score.
And G (z) is a false sub-graph generated by the generator. The object of the generator is to generate a false sample similar to the real sample, and the generated false sample is input to the discriminator for evaluation, so that the discrimination result of the discriminator on the generated false sample can be obtained.
D (G (z)) the discrimination result of the discriminator on the false sub-graph generated by the generator.
y is the true sample. In a wired network, the actual network configuration or topology can be understood.
And z is random noise at the input of the generator. In a wired network, it is understood that some configuration parameters may be set according to specific requirements.
(3) Network training
The network training step is to train the network model designed in the step (2) to the process of parameter convergence. And (2) using the subgraph and the upper triangle adjacency matrix which are coded in the step (1) as real samples, inputting the real samples, the random noise and the desensitization parameters into a network model designed in the step (2) for training, and when the distribution of the simulation samples (false samples) generated by the generator and the real network parameter information (real samples) tends to be consistent, the generation result of the generator can meet the expected wired network requirements.
1) And (3) taking the coded subgraph and the upper triangle adjacency matrix in the step (1) as real samples, and inputting the real samples into a discriminator of the network model in the step (2) for training.
2) The generator of the network model in step (2) generates a large number of simulation samples by inputting random noise (random variable) and desensitization parameters.
3) Inputting the simulation sample generated in the substep 2) into the discriminator of the network model in the step (2), and optimizing the model until the model converges with the true sample input in the step 1).
In the process of training the network model by using the real sample and the random noise, the super parameters of training are required to be set, and the super parameters refer to important parameters in the process of generating the anti-network model, and the super parameters can influence the training effect, convergence speed and stability of the model, including the learning rate, the batch size and the training iteration number. And a model optimization process comprising:
a fake analog sample is generated by the generator given the random noise and the desensitization parameters.
The real sample and the generated simulated sample are mixed, and the discriminator is trained so that the discriminator can distinguish the real sample from the simulated sample.
Based on the feedback from the discriminator, the parameters of the generator are updated to make the generated counterfeit data more realistic.
This process is repeated until the model converges.
(4) Data desensitization processing
And (3) inputting random noise by using the trained generator in the step (3) to obtain a simulation sample for parallel simulation, thereby achieving the aim of data desensitization.
And (3) inputting random noise (vector) through the trained network model in the step (3), obtaining a large number of simulation samples which tend to be consistent with the characteristic distribution of the real network parameter information, and then carrying out network parallel simulation by using the simulation samples. The method of the invention is used for carrying out network parallel simulation, which not only ensures the safety of safety test data, but also can continuously generate a large amount of sample data which accords with the real network structural characteristics and input the sample data into a network parallel simulation system, thereby achieving the aim of data desensitization.
Furthermore, the method logic of the present invention may be configured into a computer application, for example, into a computer readable storage medium, and the configuration of the algorithm is accomplished by writing a computer program in the computer readable storage medium, and the computer program is executed to perform the data desensitization method for network parallel oriented simulation.
In addition, the computer readable storage medium can be configured into a computer system, and the processor is used for executing the computer program in the computer readable storage medium, so as to further execute the data desensitization method for network parallel simulation.
In general, the data desensitization method based on the generation of the countermeasure network has better data authenticity and privacy protection effect compared with the traditional method, can more accurately simulate the real network environment in a network parallel simulation system, and provides valuable support for network design, optimization, security and the like.
The invention is not limited to the specific embodiments described above. The invention extends to any novel one, or any novel combination, of the features disclosed in this specification, as well as to any novel one, or any novel combination, of the steps of the method or process disclosed.
Claims (9)
1. A network parallel simulation oriented data desensitization method, comprising:
acquiring network parameter information of a real network;
sub-graph coding is carried out on the network parameter information, and upper triangle adjacency matrix representation is carried out on the sub-graph;
training a pre-constructed network model by utilizing the subgraph, the upper triangle adjacency matrix and random noise to obtain a network model of input random noise and output simulation samples, wherein the pre-constructed network model comprises a generator and a discriminator;
and (3) inputting random noise to generate a simulation sample by using the trained network model, so as to obtain desensitization data.
2. The network parallel simulation oriented data desensitization method according to claim 1, wherein said obtaining network parameter information of a real network comprises:
collecting raw data representing network behavior and characteristics of a real network;
and sequentially carrying out data cleaning, feature extraction and normalization on the original data.
3. The network parallel emulation oriented data desensitization method of claim 1, wherein said network parameter information comprises node location, link bandwidth, node type, link type, and delay.
4. A network parallel emulation oriented data desensitization method according to claim 3, wherein said sub-picture coding said network parameter information comprises:
creating the network parameter information into a plurality of closed subgraphs by using a full connected neural network;
traversing each sub-graph, and acquiring a node set of each sub-graph according to the connection relation and the topological structure between the nodes;
node encoding was performed using the Weisfeiler-Lehman algorithm in WL-GAN.
5. The network parallel simulation oriented data desensitization method according to claim 4, wherein said pre-built network model is WL-GAN model.
6. The network parallel simulation oriented data desensitization method according to claim 5, wherein said generator's loss function is:
gloss= -mean (D (G (z))); wherein,
z is random noise input by the generator;
g (z) is a simulated sample generated by the generator;
d (G (z)): the discriminator judges the result of the simulation sample generated by the generator;
mean: and an operator for averaging, which is used for calculating the average judging result of the judging device on the plurality of simulation samples generated by the generator.
7. The network parallel emulation oriented data desensitization method of claim 6, wherein said discriminator's loss function is:
d_loss=mean (D (y)) -mean (D (G (z))); wherein,
y is a real sample;
d (y): representing the discrimination result of the discriminator on the real sample y.
8. A computer readable storage medium storing a computer program, characterized in that the computer program is run to perform a network parallel emulation oriented data desensitization method according to any of claims 1-7.
9. A network parallel simulation oriented data desensitization system comprising a processor and a computer readable storage medium according to claim 8, said processor configured to run a computer program stored in said computer readable storage medium to perform a network parallel simulation oriented data desensitization method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410078407.XA CN117592114B (en) | 2024-01-19 | 2024-01-19 | Network parallel simulation oriented data desensitization method, system and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410078407.XA CN117592114B (en) | 2024-01-19 | 2024-01-19 | Network parallel simulation oriented data desensitization method, system and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117592114A true CN117592114A (en) | 2024-02-23 |
| CN117592114B CN117592114B (en) | 2024-04-19 |
Family
ID=89920566
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410078407.XA Active CN117592114B (en) | 2024-01-19 | 2024-01-19 | Network parallel simulation oriented data desensitization method, system and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117592114B (en) |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200134415A1 (en) * | 2018-10-30 | 2020-04-30 | Huawei Technologies Co., Ltd. | Autoencoder-Based Generative Adversarial Networks for Text Generation |
| CN111475838A (en) * | 2020-04-02 | 2020-07-31 | 中国人民解放军国防科技大学 | Graph data anonymizing method, device and storage medium based on deep neural network |
| CN111563275A (en) * | 2020-07-14 | 2020-08-21 | 中国人民解放军国防科技大学 | Data desensitization method based on generation countermeasure network |
| US20200285944A1 (en) * | 2019-03-08 | 2020-09-10 | Adobe Inc. | Graph convolutional networks with motif-based attention |
| CN111859454A (en) * | 2020-07-28 | 2020-10-30 | 桂林慧谷人工智能产业技术研究院 | Privacy protection method for defending link prediction based on graph neural network |
| CN113297429A (en) * | 2021-05-24 | 2021-08-24 | 南京大学 | Social network link prediction method based on neural network architecture search |
| CN113469253A (en) * | 2021-07-02 | 2021-10-01 | 河海大学 | Electricity stealing detection method based on triple twin network |
| CN113489604A (en) * | 2021-06-24 | 2021-10-08 | 哈尔滨工业大学 | Network simulation method, network simulation device, terminal equipment and storage medium |
| CN113886226A (en) * | 2021-09-23 | 2022-01-04 | 中国人民解放军战略支援部队信息工程大学 | Test data generation method of confrontation generation model based on twin network |
| CN114912142A (en) * | 2022-04-22 | 2022-08-16 | 绿盟科技集团股份有限公司 | Data desensitization method and device, electronic equipment and storage medium |
| CN115101146A (en) * | 2022-07-29 | 2022-09-23 | 郑州大学 | Medicine target prediction method and system based on Weisfeiler-Lehman and deep neural network |
| CN115205196A (en) * | 2022-04-29 | 2022-10-18 | 天津大学 | No-reference image quality evaluation method based on twin network and feature fusion |
| CN116486127A (en) * | 2022-01-13 | 2023-07-25 | 中国矿业大学 | Graph self-encoder clustering method integrating structural information |
| CN117012304A (en) * | 2023-09-18 | 2023-11-07 | 河北农业大学 | Deep learning molecule generation system and method fused with GGNN-GAN |
| US20230402136A1 (en) * | 2022-06-08 | 2023-12-14 | Microsoft Technology Licensing, Llc | Transformer-based graph neural network trained with structural information encoding |
| CN117290888A (en) * | 2023-11-23 | 2023-12-26 | 江苏风云科技服务有限公司 | Information desensitization method for big data, storage medium and server |
-
2024
- 2024-01-19 CN CN202410078407.XA patent/CN117592114B/en active Active
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200134415A1 (en) * | 2018-10-30 | 2020-04-30 | Huawei Technologies Co., Ltd. | Autoencoder-Based Generative Adversarial Networks for Text Generation |
| US20200285944A1 (en) * | 2019-03-08 | 2020-09-10 | Adobe Inc. | Graph convolutional networks with motif-based attention |
| CN111475838A (en) * | 2020-04-02 | 2020-07-31 | 中国人民解放军国防科技大学 | Graph data anonymizing method, device and storage medium based on deep neural network |
| CN111563275A (en) * | 2020-07-14 | 2020-08-21 | 中国人民解放军国防科技大学 | Data desensitization method based on generation countermeasure network |
| CN111859454A (en) * | 2020-07-28 | 2020-10-30 | 桂林慧谷人工智能产业技术研究院 | Privacy protection method for defending link prediction based on graph neural network |
| CN113297429A (en) * | 2021-05-24 | 2021-08-24 | 南京大学 | Social network link prediction method based on neural network architecture search |
| CN113489604A (en) * | 2021-06-24 | 2021-10-08 | 哈尔滨工业大学 | Network simulation method, network simulation device, terminal equipment and storage medium |
| CN113469253A (en) * | 2021-07-02 | 2021-10-01 | 河海大学 | Electricity stealing detection method based on triple twin network |
| CN113886226A (en) * | 2021-09-23 | 2022-01-04 | 中国人民解放军战略支援部队信息工程大学 | Test data generation method of confrontation generation model based on twin network |
| CN116486127A (en) * | 2022-01-13 | 2023-07-25 | 中国矿业大学 | Graph self-encoder clustering method integrating structural information |
| CN114912142A (en) * | 2022-04-22 | 2022-08-16 | 绿盟科技集团股份有限公司 | Data desensitization method and device, electronic equipment and storage medium |
| CN115205196A (en) * | 2022-04-29 | 2022-10-18 | 天津大学 | No-reference image quality evaluation method based on twin network and feature fusion |
| US20230402136A1 (en) * | 2022-06-08 | 2023-12-14 | Microsoft Technology Licensing, Llc | Transformer-based graph neural network trained with structural information encoding |
| CN115101146A (en) * | 2022-07-29 | 2022-09-23 | 郑州大学 | Medicine target prediction method and system based on Weisfeiler-Lehman and deep neural network |
| CN117012304A (en) * | 2023-09-18 | 2023-11-07 | 河北农业大学 | Deep learning molecule generation system and method fused with GGNN-GAN |
| CN117290888A (en) * | 2023-11-23 | 2023-12-26 | 江苏风云科技服务有限公司 | Information desensitization method for big data, storage medium and server |
Non-Patent Citations (3)
| Title |
|---|
| 丁玥 等: "基于生成式对抗网络的链路预测方法", 计算机科学与探索, vol. 13, no. 4, 30 April 2019 (2019-04-30), pages 555 - 557 * |
| 杜炜, 王行刚: "一种网络仿真工具的系统设计", 计算机工程与应用, no. 21, 21 July 2003 (2003-07-21) * |
| 郝志峰;柯妍蓉;李烁;蔡瑞初;温雯;王丽娟;: "基于图编码网络的社交网络节点分类方法", 计算机应用, no. 01, 30 January 2020 (2020-01-30) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117592114B (en) | 2024-04-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109685647B (en) | Credit fraud detection method and training method and device of model thereof, and server | |
| CN112311780B (en) | Method for generating multi-dimensional attack path and attack graph | |
| CN109005145A (en) | A kind of malice URL detection system and its method extracted based on automated characterization | |
| CN111651762A (en) | Convolutional neural network-based PE (provider edge) malicious software detection method | |
| CN110110529B (en) | Software network key node mining method based on complex network | |
| CN112492059A (en) | DGA domain name detection model training method, DGA domain name detection device and storage medium | |
| CN106960017A (en) | E-book is classified and its training method, device and equipment | |
| CN112613599A (en) | Network intrusion detection method based on generation countermeasure network oversampling | |
| CN114090402A (en) | User abnormal access behavior detection method based on isolated forest | |
| CN108718291A (en) | A kind of malice URL detection methods based on big data | |
| CN110868312A (en) | Industrial behavior anomaly detection method based on genetic algorithm optimization | |
| CN113553624A (en) | WGAN-GP privacy protection system and method based on improved PATE | |
| CN112884121A (en) | Traffic identification method based on generation of confrontation deep convolutional network | |
| CN116167002A (en) | Industrial control network anomaly detection method based on optimized random forest | |
| CN113259388B (en) | Network flow abnormity detection method, electronic equipment and readable storage medium | |
| KR20190028880A (en) | Method and appratus for generating machine learning data for botnet detection system | |
| CN104680017A (en) | Stability analysis system and method of time-varying power system | |
| CN117592114B (en) | Network parallel simulation oriented data desensitization method, system and readable storage medium | |
| CN110191081A (en) | The Feature Selection system and method for network flow attack detecting based on learning automaton | |
| CN111967003A (en) | Automatic wind control rule generation system and method based on black box model and decision tree | |
| CN116545764A (en) | Abnormal data detection method, system and equipment of industrial Internet | |
| CN109636194B (en) | Multi-source cooperative detection method and system for major change of power transmission and transformation project | |
| CN114615052A (en) | Intrusion detection method and system based on knowledge compilation | |
| Čisar et al. | Application of artificial immune networks in continuous function optimizations | |
| CN112884069A (en) | Method for detecting confrontation network sample |
Legal Events
| 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 |