CN112149119A - Dynamic active security defense method and system for artificial intelligence system and storage medium - Google Patents
Dynamic active security defense method and system for artificial intelligence system and storage medium Download PDFInfo
- Publication number
- CN112149119A CN112149119A CN202011031638.3A CN202011031638A CN112149119A CN 112149119 A CN112149119 A CN 112149119A CN 202011031638 A CN202011031638 A CN 202011031638A CN 112149119 A CN112149119 A CN 112149119A
- Authority
- CN
- China
- Prior art keywords
- artificial intelligence
- model
- processed
- information
- dynamic active
- 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.)
- Pending
Links
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 48
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000007123 defense Effects 0.000 title claims abstract description 28
- 238000003860 storage Methods 0.000 title claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 7
- 230000010354 integration Effects 0.000 claims description 5
- 230000009545 invasion Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000001525 retina Anatomy 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
- G06F21/50—Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems
- G06F21/55—Detecting local intrusion or implementing counter-measures
Landscapes
- Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a dynamic active security defense method, a system and a storage medium for an artificial intelligence system, which comprises the following steps: s1, acquiring information to be processed and a target task to be completed corresponding to the information to be processed; s2, training at least two artificial intelligence models to form a model library, wherein each model in the model library can independently complete a target task; s3, the scheduling module selects one or more models from the model library to process the information to be processed randomly or according to a set rule; and S4, integrating and outputting the output results of each model according to the set rules. The method selects one or more models from the model library to process the information to be processed randomly or according to the set rule through the scheduling module, integrates and outputs the result output by each model according to the set rule, adds the function of dynamic active defense, can effectively reduce the probability of malicious invasion of the artificial intelligence system, and improves the safety of the artificial intelligence system.
Description
Technical Field
The invention belongs to the field of artificial intelligence application, and particularly relates to a dynamic active security defense system and a storage medium for an artificial intelligence system.
Background
With the development of modern science and technology, artificial intelligence technology is widely applied to daily life and industrial production of people. The artificial intelligence mainly relates to the aspects of machine vision, fingerprint identification, face identification, retina identification, iris identification, palm print identification, automatic planning, intelligent search, intelligent control and the like, and the artificial intelligence operation is carried out after model training.
At present, artificial intelligence systems have been widely used in a plurality of fields, including fields with high requirements for system security, such as automatically driving automobiles, etc., and thus the artificial intelligence systems are required to have higher security so as to reduce the probability of malicious attack/intrusion of the systems. The malicious attack/intrusion manner may be to combat sample attacks, model attacks (to crack the system, to modify model parameters), etc.
Therefore, it is desirable to provide a dynamic active security defense method, system and storage medium for an artificial intelligence system with high security performance.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a dynamic active security defense method, a dynamic active security defense system and a storage medium for an artificial intelligence system.
In order to solve the technical problems, the invention provides the following technical scheme:
the invention provides a dynamic active security defense method for an artificial intelligence system, which comprises the following steps:
s1, acquiring information to be processed and a target task to be completed corresponding to the information to be processed;
s2, training at least two artificial intelligence models to form a model library, wherein each model in the model library can independently complete the target task in the step S1;
s3, the scheduling module selects one or more models from the model base at random or according to a set rule to process the information to be processed in the step S1;
and S4, integrating and outputting the results output by each model in the step S3 according to the set rules.
As a preferred technical solution of the present invention, the information to be processed in step S1 includes one or more of images, videos, texts, and voices.
As a preferred technical solution of the present invention, the model library in step S2 trains and deploys the models according to different artificial intelligence tasks.
As a preferred technical solution of the present invention, the scheduling policy of the scheduling module in step S3 is to select a model randomly or according to a set rule.
As a preferable technical scheme of the invention, the method also comprises the following steps:
s1, acquiring information to be processed and a target task to be completed corresponding to the information to be processed, and dividing the target task into at least two subtasks;
s2, training at least two artificial intelligence models to form a model library, wherein each model in the model library can independently complete the subtasks in the step S1;
s3, the scheduling module selects one or more models from the model base according to each subtask randomly or according to a set rule to process the information to be processed in the step S1;
and S4, integrating the output results of each model corresponding to each subtask in the step S3 according to a set rule to obtain a subtask processing result, and integrating and outputting each subtask processing result corresponding to the target task according to the set rule.
As a preferred technical solution of the present invention, the present invention further provides a dynamic active security defense system for an artificial intelligence system, including:
the model library is used for storing each trained model for realizing the artificial intelligence task;
the acquisition module is used for acquiring the information to be processed and the corresponding target task to be completed;
the scheduling module is used for selecting a model from the model library randomly or according to a set rule to execute a target task;
the integration module is used for integrating the task results finished by each model corresponding to the same information to be processed;
and the output module is used for outputting the task result.
As a preferred embodiment of the present invention, the present invention further includes:
and the acquisition module is used for acquiring one or more information of images, videos, texts and voices.
As a preferred technical solution of the present invention, the present invention further provides a computer storage medium, in which a computer program is stored, and the computer program is used for executing a dynamic active security defense method for an artificial intelligence system.
Compared with the prior art, the invention has the following beneficial effects:
the method selects one or more models from the model library to process the information to be processed randomly or according to the set rule through the scheduling module, integrates and outputs the result output by each model according to the set rule, adds the function of dynamic active defense, can effectively reduce the probability of malicious invasion of the artificial intelligence system, and improves the safety of the artificial intelligence system.
Detailed Description
The following description of the preferred embodiments of the present invention is provided for the purpose of illustration and description, and is in no way intended to limit the invention.
Example 1
In order to achieve the object of the present invention, in one embodiment of the present invention, a dynamic active security defense method for an artificial intelligence system is provided, which includes the following steps:
s1, acquiring information to be processed and a target task to be completed corresponding to the information to be processed; the system comprises a system and a client, wherein the information to be processed comprises an acquired picture, the target task to be completed is face recognition, and a target face to be recognized is recognized from the picture by the system;
s2, training 4 artificial intelligent models to form a model library, wherein each model in the model library can independently complete the target task in the step S1;
s3, the scheduling module selects two models from the model base to process the information to be processed in the step S1 randomly or according to a set rule; the scheduling strategy of the scheduling module is to select the model randomly or according to a set rule, wherein the rule can be a rule in the prior art or a rule set according to actual requirements;
and S4, integrating and outputting the output result of each model in the step S3 according to a set rule, wherein the rule can be a rule in the prior art or a rule set according to actual requirements.
In order to further optimize the implementation effect of the present invention, the present implementation further provides a dynamic active security defense system for an artificial intelligence system, including:
the model library is used for storing each trained model for realizing the artificial intelligence task;
the acquisition module is used for acquiring the information to be processed and the corresponding target task to be completed;
the scheduling module is used for selecting a model from the model library randomly or according to a set rule to execute a target task;
the integration module is used for integrating the task results finished by each model corresponding to the same information to be processed;
and the output module is used for outputting the task result.
In order to further optimize the implementation effect of the present invention, the present embodiment further provides a computer storage medium, in which a computer program is stored, and the computer program is used for executing the artificial intelligence method for dynamic active defense.
According to the method and the device, one or more models are selected from the model base through the scheduling module randomly or according to the set rules to process the information to be processed, then the result output by each model is integrated and output according to the set rules, the function of dynamic active defense is added, the probability that the artificial intelligence system is maliciously invaded can be effectively reduced, and the safety of the artificial intelligence system is improved.
Example 2
In order to achieve the object of the present invention, in one embodiment of the present invention, a dynamic active security defense method for an artificial intelligence system is provided, which is characterized by further comprising the following steps:
s1, acquiring information to be processed and a target task to be completed corresponding to the information to be processed, and dividing the target task into two subtasks; the system adjusts the driving state of the vehicle according to various parameters such as the real-time road surface picture, the speed of the vehicle and the like, and drives the vehicle to a specified position according to the path;
s2, training four artificial intelligence models to form a model library, wherein each model in the model library can independently complete the subtask in the step S1;
s3, the scheduling module selects two models from the model base according to each subtask randomly or according to a set rule to process the information to be processed in the step S1;
and S4, integrating the output results of each model corresponding to each subtask in the step S3 according to a set rule to obtain a subtask processing result, and then integrating and outputting each subtask processing result corresponding to the target task by installing the set rule.
In order to further optimize the implementation effect of the present invention, the present implementation further provides a dynamic active security defense system for an artificial intelligence system, including:
the acquisition module is used for acquiring pictures;
the model library is used for storing each trained model for realizing the artificial intelligence task;
the acquisition module is used for acquiring the information to be processed and the corresponding target task to be completed and dividing the target task into two subtasks;
the scheduling module is used for selecting a model from the model library randomly or according to a set rule to execute each subtask;
the integration module is used for integrating the output result of each model corresponding to each subtask according to a set rule and installing the processing result of each subtask corresponding to the target task with the set rule for integration;
and the output module is used for outputting the task result.
In order to further optimize the implementation effect of the present invention, the present embodiment further provides a computer storage medium, in which a computer program is stored, and the computer program is used for executing the artificial intelligence method for dynamic active defense.
According to the method, the scheduling module selects the models from the model base randomly or according to the set rules to execute each subtask, then the output results of each model corresponding to each subtask are integrated according to the set rules to obtain the subtask processing results, then the set rules are installed on each subtask processing result corresponding to the target task to be integrated and output, and the function of dynamic active defense is added, so that the probability of malicious intrusion of the artificial intelligent system can be effectively reduced, and the safety of the artificial intelligent system is improved. The target task is divided into a plurality of subtasks, and each subtask is completed through any combination, so that the safety of artificial intelligence is further improved.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. A dynamic active security defense method for an artificial intelligence system is characterized by comprising the following steps:
s1, acquiring information to be processed and a target task to be completed corresponding to the information to be processed;
s2, training at least two artificial intelligence models to form a model library, wherein each model in the model library can independently complete the target task in the step S1;
s3, the scheduling module selects one or more models from the model base at random or according to a set rule to process the information to be processed in the step S1;
and S4, integrating and outputting the results output by each model in the step S3 according to the set rules.
2. The dynamic active security defense method for an artificial intelligence system of claim 1, wherein the information to be processed in the step S1 comprises one or more of images, video, text and voice.
3. The dynamic active security defense method for artificial intelligence system of claim 1, wherein the model library in step S2 trains and deploys models according to different artificial intelligence tasks.
4. The dynamic active security defense method for an artificial intelligence system of claim 1, wherein the scheduling policy of the scheduling module in the step S3 is to select the model randomly or according to a set rule.
5. The dynamic active security defense method for artificial intelligence systems of claim 1, further comprising the steps of:
s1, acquiring information to be processed and a target task to be completed corresponding to the information to be processed, and dividing the target task into at least two subtasks;
s2, training at least two artificial intelligence models to form a model library, wherein each model in the model library can independently complete the subtasks in the step S1;
s3, the scheduling module selects one or more models from the model base according to each subtask randomly or according to a set rule to process the information to be processed in the step S1;
and S4, integrating the output results of each model corresponding to each subtask in the step S3 according to a set rule to obtain a subtask processing result, and integrating and outputting each subtask processing result corresponding to the target task according to the set rule.
6. A dynamic active security defense system for an artificial intelligence system, comprising:
the model library is used for storing each trained model for realizing the artificial intelligence task;
the acquisition module is used for acquiring the information to be processed and the corresponding target task to be completed;
the scheduling module is used for selecting a model from the model library randomly or according to a set rule to execute a target task;
the integration module is used for integrating the task results finished by each model corresponding to the same information to be processed;
and the output module is used for outputting the task result.
7. The dynamic active security defense system for artificial intelligence systems of claim 5, further comprising:
and the acquisition module is used for acquiring one or more information of images, videos, texts and voices.
8. A computer storage medium, characterized in that the storage medium stores a computer program for executing the method for dynamic active security defense for artificial intelligence systems according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011031638.3A CN112149119A (en) | 2020-09-27 | 2020-09-27 | Dynamic active security defense method and system for artificial intelligence system and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011031638.3A CN112149119A (en) | 2020-09-27 | 2020-09-27 | Dynamic active security defense method and system for artificial intelligence system and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112149119A true CN112149119A (en) | 2020-12-29 |
Family
ID=73894826
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011031638.3A Pending CN112149119A (en) | 2020-09-27 | 2020-09-27 | Dynamic active security defense method and system for artificial intelligence system and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112149119A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023082112A1 (en) * | 2021-11-10 | 2023-05-19 | Nokia Shanghai Bell Co., Ltd. | Apparatus, methods, and computer programs |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355072A (en) * | 2016-08-19 | 2017-01-25 | 沈建国 | Implementation method and device for three-dimensional model verification code |
| CN108023876A (en) * | 2017-11-20 | 2018-05-11 | 西安电子科技大学 | Intrusion detection method and intruding detection system based on sustainability integrated study |
| CN108322445A (en) * | 2018-01-02 | 2018-07-24 | 华东电力试验研究院有限公司 | A kind of network inbreak detection method based on transfer learning and integrated study |
| CN110046709A (en) * | 2019-04-22 | 2019-07-23 | 成都新希望金融信息有限公司 | A kind of multi-task learning model based on two-way LSTM |
| CN110334814A (en) * | 2019-07-01 | 2019-10-15 | 阿里巴巴集团控股有限公司 | For constructing the method and system of risk control model |
| CN110347155A (en) * | 2019-06-26 | 2019-10-18 | 北京理工大学 | A kind of intelligent vehicle automatic Pilot control method and system |
| CN110889117A (en) * | 2019-11-28 | 2020-03-17 | 支付宝(杭州)信息技术有限公司 | Method and device for defending model attack |
| CN111489409A (en) * | 2020-04-24 | 2020-08-04 | 东软医疗系统股份有限公司 | CT image processing method and device, CT equipment and CT system |
-
2020
- 2020-09-27 CN CN202011031638.3A patent/CN112149119A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355072A (en) * | 2016-08-19 | 2017-01-25 | 沈建国 | Implementation method and device for three-dimensional model verification code |
| CN108023876A (en) * | 2017-11-20 | 2018-05-11 | 西安电子科技大学 | Intrusion detection method and intruding detection system based on sustainability integrated study |
| CN108322445A (en) * | 2018-01-02 | 2018-07-24 | 华东电力试验研究院有限公司 | A kind of network inbreak detection method based on transfer learning and integrated study |
| CN110046709A (en) * | 2019-04-22 | 2019-07-23 | 成都新希望金融信息有限公司 | A kind of multi-task learning model based on two-way LSTM |
| CN110347155A (en) * | 2019-06-26 | 2019-10-18 | 北京理工大学 | A kind of intelligent vehicle automatic Pilot control method and system |
| CN110334814A (en) * | 2019-07-01 | 2019-10-15 | 阿里巴巴集团控股有限公司 | For constructing the method and system of risk control model |
| CN110889117A (en) * | 2019-11-28 | 2020-03-17 | 支付宝(杭州)信息技术有限公司 | Method and device for defending model attack |
| CN111489409A (en) * | 2020-04-24 | 2020-08-04 | 东软医疗系统股份有限公司 | CT image processing method and device, CT equipment and CT system |
Non-Patent Citations (2)
| Title |
|---|
| 朱雪: "动态集成分类方法的研究", 《中国优秀硕士学位论文全文数据库·信息科技辑》, pages 42 - 43 * |
| 王宁燕;韩晓霞;: "聚类的动态分类器集成选择", 计算机系统应用, no. 04 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023082112A1 (en) * | 2021-11-10 | 2023-05-19 | Nokia Shanghai Bell Co., Ltd. | Apparatus, methods, and computer programs |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11727689B2 (en) | Alert directives and focused alert directives in a behavioral recognition system | |
| CN111539265A (en) | Method for detecting abnormal behaviors in elevator car | |
| CN112149119A (en) | Dynamic active security defense method and system for artificial intelligence system and storage medium | |
| CN112598677B (en) | Method and apparatus for image analysis | |
| CN110969110A (en) | Face tracking method and system based on deep learning | |
| CN112686114A (en) | Behavior detection method, device and equipment | |
| CN112153076A (en) | Computer network safety intrusion detection system | |
| CN116563327A (en) | Dynamic scene background modeling method based on box diagram | |
| CN112561076B (en) | Model processing method and device | |
| US20210287087A1 (en) | Efficient simultaneous inference computation for multiple neural networks | |
| CN112464549B (en) | Dynamic allocation method of countermeasure unit | |
| EP4217897A1 (en) | A method of training a submodule and preventing capture of an ai module | |
| EP3864546A1 (en) | Departure-based process-level detection of stealthy attacks on control systems | |
| Kharal | Towards Augmentation Based Defense Strategies Against Adversarial Attacks | |
| CN116762082A (en) | Training sub-modules and method for preventing capture of AI modules | |
| Sun et al. | Memory Efficient Corner Detection for Event-Driven Dynamic Vision Sensors | |
| CN117274769A (en) | Method, system and terminal for resisting sample attack aiming at visual target tracking | |
| CN116071253A (en) | Image content noise processing method based on pellet calculation | |
| Dominka et al. | Artificial Neural Networks for (Automotive) Embedded Control-A Comparative Study | |
| CN117103256A (en) | Intelligent inspection method, system and medium for explosion-proof robot | |
| CN116229144A (en) | Dynamic countermeasure training method and system based on image fixed features | |
| CN117314943A (en) | Image processing method and related equipment | |
| Gebrehiwot et al. | Robust Template Update Strategy for Efficient Visual Object Tracking | |
| Han et al. | A Multi-Strategy Integrated Optimized YOLOv5s Algorithm and its Application to Target Detection | |
| CN117523302A (en) | Image classification, device, storage medium and equipment for multi-scale countermeasure patches |
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 | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Kou Chaofeng Inventor before: Kou Chaofeng |
|
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20210714 Address after: 518000 software building, No.9, Gaoxin Zhongyi Road, Nanshan District, Shenzhen City, Guangdong Province Applicant after: Kou Chaofeng Address before: 215000 station A005, 3rd floor, No. 58, South Tiancheng Road, high speed rail new town, Xiangcheng District, Suzhou City, Jiangsu Province (cluster registration) Applicant before: Suzhou Xiashi Intelligent Technology Co.,Ltd. |