CN107104978A - A kind of network risks method for early warning based on deep learning - Google Patents
A kind of network risks method for early warning based on deep learning Download PDFInfo
- Publication number
- CN107104978A CN107104978A CN201710375043.1A CN201710375043A CN107104978A CN 107104978 A CN107104978 A CN 107104978A CN 201710375043 A CN201710375043 A CN 201710375043A CN 107104978 A CN107104978 A CN 107104978A
- Authority
- CN
- China
- Prior art keywords
- risk
- sample data
- training
- database
- forecast model
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/20—Network architectures or network communication protocols for network security for managing network security; network security policies in general
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of network risks method for early warning based on deep learning, this method is comprised the following steps：A1. collection the whole network section cyberspace asset risk sample data, and be stored in database；A2. data are extracted from database, convolutional neural networks distribution training study is carried out, forms initial risks forecast model；A3. creation data is inputted into risk forecast model, assesses its value-at-risk, such as reached threshold value of warning, then alarm.By this method and equipment, security risk assessment and early warning can be carried out to multiple objective networks or in the absence of the target of obvious leak, the safe condition of a network can be assessed on the whole；And response speed is improved, quickly find risk point；Maintenance cost is reduced simultaneously, saves manpower.
Description
Technical field
It is directed to the present invention relates to network risks early warning technology, more particularly to one kind in regional extent, based on machine depth
The network risks method for early warning and system of study.
Background technology
Current network security field, one target of detection whether safety, be all the tradition such as port scan by vulnerability scanning
Mode is carried out, and this mode is for single target and to there is obvious leak effective, for batch target or in the absence of obvious
The target of leak then quickly can not comprehensively obtain its safe condition.
The content of the invention
To solve the above problems, the present invention provides a kind of network risks method for early warning and equipment based on deep learning, its
It can carry out quickly comprehensively obtaining safe condition to batch target or in the absence of the target of obvious leak.
The present invention provides a kind of network risks method for early warning based on deep learning, it is characterised in that comprise the following steps：
A1. collection the whole network section cyberspace asset risk sample data, and be stored in database；A2. data are extracted from database, are carried out
Convolutional neural networks (CNN) distribution training study, forms risk forecast model；A3. creation data is inputted into risk profile mould
Type, assesses its value-at-risk, such as reaches threshold value of warning, then alarms.
Preferably, the step A1 includes：A11. risk elements are determined, to the cyberspace asset risk sample of the whole network section
Data are acquired；A12. the risk sample data collected is subjected to vulnerability scanning, and divides level of security.
Further preferably, the risk elements include：Target IP, open port, server system type and version, clothes
Be engaged in device application type and version, the leak existed, type of database and version, weak passwurd, whether accelerated using CDN and fire wall
In one or more.
Further preferably, the security classification is：High-risk, middle danger, low danger, four level of securitys of safety, it compares
Example is 1:1:1:1, quantity >=5000 of every kind of level of security.
Further preferably, the step A1 also includes：A13. by the cyberspace asset risk sample data after collection
It is converted into the recognizable binary sample data of deep learning.
Further preferably, the step A13 includes：A131. sample is subjected to pictured processing, be cut to unified big
It is small；A132. whitening processing is carried out to the picture after cutting.
Preferably, the distributed training study of the step A2 is by the way of gradient is successively decreased, and its Initial Gradient is
10-4。
Preferably, the step A2 includes：A21. training environment is prepared, training environment uses Tensorflow GPU patterns
Carry out；A22. training sample data are extracted from database, model training is carried out with reference to convolutional neural networks, obtains risk profile
Model；A23. test sample data are extracted from database, test is estimated to risk forecast model.
Further preferably, the step A22 includes：A221. prototype network structure, using 3 convolutional layers, first
With 3*3 convolution kernel, behind two convolution kernels using 2*2, each convolutional layer back is with maximum pond layer, afterwards again with two
Individual hidden layer and an output layer, the characteristic pattern of each convolutional layer is respectively with 32,64,128；A222. entered using softmax functions
Row is returned, and last output layer does not need softmax to return；A223. it is trained using training sample data, obtains initial wind
Dangerous forecast model.
The present invention also provides a kind of computer-readable recording medium for including computer program, and the computer program is counted
Calculation machine performs to realize method as described above.
Beneficial effects of the present invention：The whole network section cyberspace asset risk sample is acquired, and combines convolutional Neural
Network (CNN) carries out distributed training study, and self is carried out by comprehensive all local results, and with reference to analysis of neural network
Study and adjustment, the risk forecast model that one obtained integrates comprehensively.The risk forecast model can be to multiple objective networks
Or security risk assessment and early warning are carried out in the absence of the target of obvious leak, the safe shape of a network can be assessed on the whole
State；And response speed is improved, risk point is quickly found, the treatment effeciency of network safety situation analysis prediction is improved and accurate
Property；Maintenance cost is reduced simultaneously, saves manpower.
More advantages can also be obtained in further preferred scheme：Network security assessment and early warning are carried out using CNN
Maximum resistance be：The structure of application scenarios learning sample.The present invention is by the way that the risk elements of risk sample are defined to：Target
IP, open port, server system type and version, server application type and version, the leak existed, type of database and
Version, weak passwurd, the one or more for whether being accelerated using CDN, whether being used in fire wall, so that it is distributed both to have saved CNN
The time of training, the accuracy of security evaluation and early warning result is improved again.
Brief description of the drawings
Fig. 1 is the network risks method for early warning schematic flow sheet based on deep learning of the specific embodiment of the invention.
Fig. 2 is convolutional neural networks distribution training learning process schematic diagram in this specific embodiment of the invention.
Embodiment
With reference to embodiment and compare accompanying drawing the present invention be described in further detail, it should be emphasised that,
What the description below was merely exemplary, the scope being not intended to be limiting of the invention and its application.
As shown in figure 1, the present embodiment provides a kind of network risks method for early warning based on deep learning, it includes following step
Suddenly：
Step 1. collection the whole network section cyberspace asset risk sample data.
Step 1-1, storehouse is built to cyberspace asset risk sample data, is identified by the risk point to assets, really
Determine risk elements, risk elements include：Target IP, open port, server system type and version, server application type and
Version, the leak existed, type of database and version, weak passwurd, if accelerated using CDN, if use fire wall.According to wind
Strategically located and difficult of access element, the cyberspace asset risk sample data to the whole network section is acquired.
It is likely to result in the risk elements of serious consequence to network security as described above by extracting, forms sample data,
The real reliability that post depth study predicts the outcome can be ensured.Some risk elements appear not to danger, but combine
To together, it is possible to cause fatal leak.
Cyberspace risk sample data acquisition methods：Use " service type and the version letter that objective network main frame is run
Breath " detection techniques, " information such as operating system and device type " detection techniques, " destination host Security Vulnerability " identification technology,
" CDN, fire wall " identification technology carrys out the collection work of completed sample evidence, and the sample tool of collection is ensured using distributed computing technology
There is real-time.
The risk sample data collected is carried out vulnerability scanning by step 1-2., and is divided into four level of securitys：High-risk,
Middle danger, low danger, safety.Wherein high-risk, middle danger, low danger, the ratio of four level of securitys of safety are 1:1:1:1, every kind of safe level
Other quantity >=5000.
By the division of safe class, each danger grade all includes the unsafe factor specified and this leakage
The maximum extent of damage that hole may be brought, the loss that user can tentatively grasp the affiliated classification of leak and be likely to result in,
The formulation of specific defensive measure is carried out, lowers the risk that user faces Dangerous Internet with this.
Cyberspace asset risk sample data is converted into the recognizable binary sample number of deep learning by step 1-3.
According to.Acquisition node data summarization is put in storage in control server, data after data cleansing.
The task of data cleansing is to filter out undesirable data, predominantly incomplete data, the data of mistake
And the data repeated etc..
In view of most of result data in the database is text or numeral, and combined situation is various, in quantized samples
There is very big difficulty in parameter, it is difficult to form the learning model of deep learning, therefore sample data is fabricated to picture.
1) samples pictures are handled：Samples pictures are uniformly cropped to 100x100 pixel sizes, cutting middle section is used to comment
Estimate or random cropping is used to train.
2) approximate whitening processing is carried out to picture, makes model insensitive for the dynamic range change of picture.Step 2.
Data are extracted from database, convolutional neural networks distribution training study is carried out, forms risk forecast model.
Good learning model, can not only improve the speed of study, can more improve the accuracy of learning outcome, while to examine
Consider the quantity of sample, in terms of comprehensive, CNN models are current optimal deep learning models.This step uses picture training
The characteristics of mode combination convolutional neural networks are good at solution picture recognition is trained.
Learning model is divided into two class samples：Training sample and test sample.Training sample is for needed for debugging, training stage
Sample data, learn function used, method as percentage regulation, guiding final result leads to correct direction；Test sample
The effect of Network Risk Assessment and early warning whether is met as checking accuracy, for evaluation stage.Training sample is training mould
The sample data that the type stage uses；Test sample is sample data used in the assessment models stage.Two class samples all it is known from
Variable and dependent variable.
According to the sample of division, training machine is formed into risk forecast model, its process is as shown in Figure 2.
Step 2-1. prepares training environment.Training environment is carried out using Tensorflow GPU patterns, GPU calculating speed
It is faster than CPU, the time cost of training process can be reduced.
The step 2-2. training pattern stages.The ready training sample combination convolution god of step 2 is used in training environment
Model training is carried out through network.Training process is as follows：
1) the prototype network structure defined, using 3 convolutional layers, first convolution kernel with 3*3, behind two use 2*
2 convolution kernel, each convolutional layer back is with maximum pond layer, afterwards again with two hidden layers and an output layer, Mei Gejuan
The characteristic pattern of lamination is respectively with 32,64,128.
2) returned using softmax functions, last output layer does not need softmax function regressions.
3) after prototype network structure is defined, it is trained, training obtains initial risks forecast model.
Degree of accuracy optimization is carried out with gradient decreasing fashion, its Initial Gradient is 10-4；Trained using CNN distributions, pass through ladder
Spending the mode successively decreased makes training data carry out linear regression, reaches poised state, so that find out influences larger to training result
Factor, then carry out distributed training using data as CNN inputs.Each work of the distributed training of data parallel formula in GPU
Make all to store the backup of a model, the different piece of processing data on each node, each working node of recombinant on node
Result, and synchronistic model parameter among the nodes；It can accelerate data training and model sets up efficiency.
The step 2-3. assessment models stages.The ready test sample of step 2 is used in training environment, to step 2-2
Obtained initial risks forecast model is estimated test, confirms whether accuracy is qualified.Method of testing：Input test sample is arrived
In initial risks forecast model, see whether match with expection after result to be output.The flow that then puts into production is matched, for network
The early warning of risk.It is unqualified, return to step 2-2 and carry out algorithm optimization, until output result matches with expection.Output result
There are four kinds:Safety, low-risk, risk, excessive risk.
The risk forecast model set up by as above method, according to the difference for the risk elements for forming sample data, its wind
The accuracy rate of dangerous forecast model is different.According to risk elements selected in sample data, its formed risk forecast model when
Between it is different, and the accuracy rate of risk forecast model is also different, its result such as following table：
It was found from from upper table：When risk elements include " Target IP, open port, server system type and version, service
Device application type and version, the leak existed, type of database and version, weak passwurd, if accelerated using CDN, if use
During fire wall ", the accuracy rate of its risk forecast model is high, and learning time is short, and when risk elements lack it is therein some
When, its result lacks accuracy.
When risk elements it is unnecessary these when, test result indicates that：Its learning time is long, forms the time of risk forecast model
Long, the cost of consuming is high.The appropriate risk elements by choosing：" Target IP, open port, server system type and version,
Server application type and version, the leak existed, type of database and version, weak passwurd, if accelerated using CDN, if
Using fire wall " so that learning time is short, and the accuracy rate of the risk forecast model formed is high, i.e., fast and accurate.
Creation data is inputted risk forecast model by step 3., assesses its value-at-risk, is such as reached threshold value of warning, is then alarmed.
Above content is to combine specific/preferred embodiment made for the present invention be further described, it is impossible to recognized
The specific implementation of the fixed present invention is confined to these explanations.For general technical staff of the technical field of the invention,
Without departing from the inventive concept of the premise, it can also make some replacements or modification to the embodiment that these have been described,
And these are substituted or variant should all be considered as belonging to protection scope of the present invention.
Claims (10)
1. a kind of network risks method for early warning based on deep learning, it is characterised in that comprise the following steps：
A1. collection the whole network section cyberspace asset risk sample data, and be stored in database；
A2. data are extracted from database, convolutional neural networks distribution training study is carried out, forms risk forecast model；
A3. creation data is inputted into risk forecast model, assesses its value-at-risk, such as reached threshold value of warning, then alarm.
2. the method as described in claim 1, it is characterised in that the step A1 includes：
A11. risk elements are determined, the cyberspace asset risk sample data to the whole network section is acquired；
A12. the risk sample data collected is subjected to vulnerability scanning, and divides level of security.
3. method as claimed in claim 2, it is characterised in that the risk elements include：Target IP, open port, service
Device system type and version, server application type and version, the leak existed, type of database and version, weak passwurd, whether
Accelerated and the one or more in fire wall using CDN.
4. method as claimed in claim 2, it is characterised in that the security classification is：High-risk, middle danger, low danger, safety
Four level of securitys, its ratio is 1:1:1:1, quantity >=5000 of every kind of level of security.
5. method as claimed in claim 2, it is characterised in that the step A1 also includes：
A13. cyberspace asset risk sample data is converted into the recognizable binary sample data of deep learning.
6. method as claimed in claim 5, it is characterised in that the step A13 includes：
A131. sample is subjected to pictured processing, is cut to unified size；
A132. whitening processing is carried out to the picture after cutting.
7. the method as described in claim 1, it is characterised in that the distributed training study of the step A2 uses gradient
The mode successively decreased, its Initial Gradient is 10-4。
8. the method as described in claim 1, it is characterised in that the step A2 includes：
A21. training environment is prepared, training environment is carried out using Tensorflow GPU patterns；
A22. training sample data are extracted from database, model training is carried out with reference to convolutional neural networks, obtains initial risks
Forecast model；
A23. test sample data are extracted from database, test is estimated to initial risks forecast model.
9. the method as described in claim 1, it is characterised in that the step A22 includes：
A221. prototype network structure, using 3 convolutional layers, first convolution kernel with 3*3, behind two convolution using 2*2
Core, each convolutional layer back is with maximum pond layer, afterwards again with two hidden layers and an output layer, the spy of each convolutional layer
Figure is levied respectively with 32,64,128；
A222. returned using softmax functions, last output layer does not need softmax to return；
A223. it is trained using training sample data, obtains initial risks forecast model.
10. a kind of computer-readable recording medium for including computer program, it is characterised in that the computer program is calculated
Machine performs to realize any described methods of claim 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710375043.1A CN107104978B (en) | 2017-05-24 | 2017-05-24 | Network risk early warning method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710375043.1A CN107104978B (en) | 2017-05-24 | 2017-05-24 | Network risk early warning method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107104978A true CN107104978A (en) | 2017-08-29 |
| CN107104978B CN107104978B (en) | 2019-12-24 |
Family
ID=59669379
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710375043.1A Active CN107104978B (en) | 2017-05-24 | 2017-05-24 | Network risk early warning method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107104978B (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107707553A (en) * | 2017-10-18 | 2018-02-16 | 北京启明星辰信息安全技术有限公司 | Weak passwurd scan method, device and computer-readable storage medium |
| CN107948587A (en) * | 2017-11-15 | 2018-04-20 | 中国联合网络通信集团有限公司 | Methods of risk assessment, the apparatus and system of monitoring device |
| CN108364106A (en) * | 2018-02-27 | 2018-08-03 | 平安科技（深圳）有限公司 | A kind of expense report Risk Forecast Method, device, terminal device and storage medium |
| CN108897614A (en) * | 2018-05-25 | 2018-11-27 | 福建天晴数码有限公司 | A kind of memory method for early warning and server-side based on convolutional neural networks |
| CN109472396A (en) * | 2018-10-17 | 2019-03-15 | 四川佳联众合企业管理咨询有限公司 | Mountain fire prediction technique based on depth e-learning |
| CN109993412A (en) * | 2019-03-01 | 2019-07-09 | 百融金融信息服务股份有限公司 | The construction method and device of risk evaluation model, storage medium, computer equipment |
| CN110399252A (en) * | 2019-07-19 | 2019-11-01 | 广东浪潮大数据研究有限公司 | A kind of data back up method, device, equipment and computer readable storage medium |
| CN110543565A (en) * | 2019-08-30 | 2019-12-06 | 广西电网有限责任公司南宁供电局 | Auditing method, system and readable storage medium based on convolutional neural network model |
| CN110598959A (en) * | 2018-05-23 | 2019-12-20 | 中国移动通信集团浙江有限公司 | Asset risk assessment method and device, electronic equipment and storage medium |
| CN110875912A (en) * | 2018-09-03 | 2020-03-10 | 中移（杭州）信息技术有限公司 | Network intrusion detection method, device and storage medium based on deep learning |
| CN111400572A (en) * | 2020-02-28 | 2020-07-10 | 开普云信息科技股份有限公司 | Content safety monitoring system and method for realizing image feature recognition based on convolutional neural network |
| CN112001565A (en) * | 2020-09-08 | 2020-11-27 | 清华大学合肥公共安全研究院 | Earthquake disaster loss prediction and evaluation method and system based on Softmax regression model |
| CN112565255A (en) * | 2020-12-04 | 2021-03-26 | 广东电网有限责任公司珠海供电局 | Electric power Internet of things equipment safety early warning method based on BP neural network |
| CN113065666A (en) * | 2021-05-11 | 2021-07-02 | 海南善沙网络科技有限公司 | Distributed computing method for training neural network machine learning model |
| CN113225358A (en) * | 2021-07-09 | 2021-08-06 | 四川大学 | Network security risk assessment system |
| CN113361855A (en) * | 2021-05-07 | 2021-09-07 | 浙江警官职业学院 | Short, medium and long-term risk warning method and device |
| CN113506039A (en) * | 2021-08-03 | 2021-10-15 | 杭银消费金融股份有限公司 | Risk prediction method and device based on intelligent AI machine learning |
| CN113542278A (en) * | 2021-07-16 | 2021-10-22 | 北京源堡科技有限公司 | Network security assessment method, system and device |
| CN115396242A (en) * | 2022-10-31 | 2022-11-25 | 江西神舟信息安全评估中心有限公司 | Data identification method and network security vulnerability detection method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104873186A (en) * | 2015-04-17 | 2015-09-02 | 中国科学院苏州生物医学工程技术研究所 | Wearable artery detection device and data processing method thereof |
| CN204708828U (en) * | 2015-04-17 | 2015-10-21 | 中国科学院苏州生物医学工程技术研究所 | A kind of wearable noinvasive arterial health checkout gear |
| US20160098646A1 (en) * | 2014-10-06 | 2016-04-07 | Seagate Technology Llc | Dynamically modifying a boundary of a deep learning network |
| CN105894372A (en) * | 2016-06-13 | 2016-08-24 | 腾讯科技（深圳）有限公司 | Method and device for predicting group credit |
-
2017
- 2017-05-24 CN CN201710375043.1A patent/CN107104978B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160098646A1 (en) * | 2014-10-06 | 2016-04-07 | Seagate Technology Llc | Dynamically modifying a boundary of a deep learning network |
| CN104873186A (en) * | 2015-04-17 | 2015-09-02 | 中国科学院苏州生物医学工程技术研究所 | Wearable artery detection device and data processing method thereof |
| CN204708828U (en) * | 2015-04-17 | 2015-10-21 | 中国科学院苏州生物医学工程技术研究所 | A kind of wearable noinvasive arterial health checkout gear |
| CN105894372A (en) * | 2016-06-13 | 2016-08-24 | 腾讯科技（深圳）有限公司 | Method and device for predicting group credit |
Non-Patent Citations (1)
| Title |
|---|
| 张海舰等: "基于深度学习技术的恶意APP检测方案", 《网络安全技术与应用》 * |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107707553A (en) * | 2017-10-18 | 2018-02-16 | 北京启明星辰信息安全技术有限公司 | Weak passwurd scan method, device and computer-readable storage medium |
| CN107707553B (en) * | 2017-10-18 | 2020-02-07 | 北京启明星辰信息安全技术有限公司 | Weak password scanning method and device and computer storage medium |
| CN107948587A (en) * | 2017-11-15 | 2018-04-20 | 中国联合网络通信集团有限公司 | Methods of risk assessment, the apparatus and system of monitoring device |
| CN107948587B (en) * | 2017-11-15 | 2019-12-27 | 中国联合网络通信集团有限公司 | Risk assessment method, device and system for monitoring equipment |
| CN108364106A (en) * | 2018-02-27 | 2018-08-03 | 平安科技（深圳）有限公司 | A kind of expense report Risk Forecast Method, device, terminal device and storage medium |
| CN110598959A (en) * | 2018-05-23 | 2019-12-20 | 中国移动通信集团浙江有限公司 | Asset risk assessment method and device, electronic equipment and storage medium |
| CN108897614A (en) * | 2018-05-25 | 2018-11-27 | 福建天晴数码有限公司 | A kind of memory method for early warning and server-side based on convolutional neural networks |
| CN110875912A (en) * | 2018-09-03 | 2020-03-10 | 中移（杭州）信息技术有限公司 | Network intrusion detection method, device and storage medium based on deep learning |
| CN109472396A (en) * | 2018-10-17 | 2019-03-15 | 四川佳联众合企业管理咨询有限公司 | Mountain fire prediction technique based on depth e-learning |
| CN109472396B (en) * | 2018-10-17 | 2023-06-20 | 成都卡普数据服务有限责任公司 | Mountain fire prediction method based on deep network learning |
| CN109993412A (en) * | 2019-03-01 | 2019-07-09 | 百融金融信息服务股份有限公司 | The construction method and device of risk evaluation model, storage medium, computer equipment |
| CN110399252A (en) * | 2019-07-19 | 2019-11-01 | 广东浪潮大数据研究有限公司 | A kind of data back up method, device, equipment and computer readable storage medium |
| CN110543565A (en) * | 2019-08-30 | 2019-12-06 | 广西电网有限责任公司南宁供电局 | Auditing method, system and readable storage medium based on convolutional neural network model |
| CN111400572A (en) * | 2020-02-28 | 2020-07-10 | 开普云信息科技股份有限公司 | Content safety monitoring system and method for realizing image feature recognition based on convolutional neural network |
| CN112001565A (en) * | 2020-09-08 | 2020-11-27 | 清华大学合肥公共安全研究院 | Earthquake disaster loss prediction and evaluation method and system based on Softmax regression model |
| CN112565255A (en) * | 2020-12-04 | 2021-03-26 | 广东电网有限责任公司珠海供电局 | Electric power Internet of things equipment safety early warning method based on BP neural network |
| CN113361855A (en) * | 2021-05-07 | 2021-09-07 | 浙江警官职业学院 | Short, medium and long-term risk warning method and device |
| CN113065666A (en) * | 2021-05-11 | 2021-07-02 | 海南善沙网络科技有限公司 | Distributed computing method for training neural network machine learning model |
| CN113225358A (en) * | 2021-07-09 | 2021-08-06 | 四川大学 | Network security risk assessment system |
| CN113225358B (en) * | 2021-07-09 | 2021-09-03 | 四川大学 | Network security risk assessment system |
| CN113542278A (en) * | 2021-07-16 | 2021-10-22 | 北京源堡科技有限公司 | Network security assessment method, system and device |
| CN113542278B (en) * | 2021-07-16 | 2023-04-25 | 北京源堡科技有限公司 | Network security assessment method, system and device |
| CN113506039A (en) * | 2021-08-03 | 2021-10-15 | 杭银消费金融股份有限公司 | Risk prediction method and device based on intelligent AI machine learning |
| CN115396242A (en) * | 2022-10-31 | 2022-11-25 | 江西神舟信息安全评估中心有限公司 | Data identification method and network security vulnerability detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107104978B (en) | 2019-12-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107104978A (en) | A kind of network risks method for early warning based on deep learning | |
| CN105677791B (en) | For analyzing the method and system of the operation data of wind power generating set | |
| CN103761748B (en) | Anomaly detection method and device | |
| CN109460574A (en) | A kind of prediction technique of aero-engine remaining life | |
| CN109902018B (en) | Method for acquiring test case of intelligent driving system | |
| CN109379379A (en) | Based on the network inbreak detection method for improving convolutional neural networks | |
| CN105897714A (en) | Botnet detection method based on DNS (Domain Name System) flow characteristics | |
| Lai et al. | Online pattern matching and prediction of incoming alarm floods | |
| CN105072214B (en) | C＆C domain name recognition methods based on domain name feature | |
| CN106060008B (en) | A kind of network intrusions method for detecting abnormality | |
| CN109818798A (en) | A kind of wireless sensor network intruding detection system and method merging KPCA and ELM | |
| CN112001110B (en) | Structural damage identification monitoring method based on vibration signal space real-time recurrent graph convolutional neural network | |
| CN107491752A (en) | Ship board character recognition method, device in a kind of natural scene based on deep learning | |
| CN111723367B (en) | Method and system for evaluating service scene treatment risk of power monitoring system | |
| CN109145030B (en) | Abnormal data access detection method and device | |
| Anwar et al. | A data-driven approach to distinguish cyber-attacks from physical faults in a smart grid | |
| CN105354198A (en) | Data processing method and apparatus | |
| CN112333128B (en) | Web attack behavior detection system based on self-encoder | |
| CN111126820A (en) | Electricity stealing prevention method and system | |
| CN116167010A (en) | Rapid identification method for abnormal events of power system with intelligent transfer learning capability | |
| CN115660262A (en) | Intelligent engineering quality inspection method, system and medium based on database application | |
| CN104143088B (en) | Face identification method based on image retrieval and feature weight learning | |
| CN115795353B (en) | Underground metal target classification method and system based on unbalanced data set | |
| CN117152528A (en) | Insulator state recognition method, insulator state recognition device, insulator state recognition apparatus, insulator state recognition program, and insulator state recognition program | |
| CN116884192A (en) | Power production operation risk early warning method, system and equipment |
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 |