CN106355250A - Optimization method and device for judging convert channels based on neural network - Google Patents
Optimization method and device for judging convert channels based on neural network Download PDFInfo
- Publication number
- CN106355250A CN106355250A CN201610777841.2A CN201610777841A CN106355250A CN 106355250 A CN106355250 A CN 106355250A CN 201610777841 A CN201610777841 A CN 201610777841A CN 106355250 A CN106355250 A CN 106355250A
- Authority
- CN
- China
- Prior art keywords
- communication channel
- private communication
- neural network
- network model
- pseudo
- 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
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The invention provides an optimization method and device for judging convert channels based on a neural network. The method comprises steps as follows: a BP (back-propagation) neural network model is established; the BP neural network model is subjected to learning training by the aid of a preset convert channel set, and an optimal weighting parameter is obtained through the training, wherein the preset convert channel set comprises pseudo convert channels and real convert channels; the optical weighting parameter obtained through the training is added to the BP neural network model, and an updated BP neutral network model is obtained; the updated BP neutral network model is used for judging an actual convert channel set comprising pseudo convert channels, so that real convert channels are found out. Based on a feedback type learning mode of the BP neural network, the real convert channels and the pseudo convert channels can be found out from a suspicious convert channel set.
Description
Technical field
The present invention relates to technical field of data security, excellent particularly to a kind of judgement private communication channel based on neutral net
Change method and device.
Background technology
Private communication channel is the communication port that a kind of permission process transmits information in the form of running counter to System Security Policy.Simply
For, private communication channel is exactly the communication port being not intended to for transmitting information.Private communication channel has been widely used now
In network information data safe transmission.
Start most, private communication channel is considered as the security threat of one-of-a-kind system, most of research with regard to private communication channel is all
It is for Multilevel Security Systems.With the prosperity and development of computer system, the focus of private communication channel gradually transfers to computer
In procotol.The carrier being provided by private communication channel by computer network overt channel, with the development of network, network is hidden
The applied research covering channel is also growing.
Private communication channel is the extension of Information Hiding Techniques, and ciphertext is exposed to attacker unlike encryption method by it, and
It is that information is snugly delivered to from one section by the other end by the method for Communication hiding passage.Therefore, private communication channel is to ensure that
One of important method of safe information transmission, high performance private communication channel can resist attack and the destruction of the third party, and has
Larger channel capacity and the rate of information throughput.
In prior art, when finding the channel set that a group is probably private communication channel, it is difficult to by effective method
Therefrom filter out pseudo- private communication channel, this is probably that the small probability unusual operation of user causes.
Content of the invention
The purpose of the present invention is intended at least solve one of described technological deficiency.
For this reason, it is an object of the invention to proposing a kind of optimization method of the judgement private communication channel based on neutral net and dress
Put, can reaction type learning style based on bp neutral net, the set of suspicious private communication channel is found out true private communication channel
With pseudo- private communication channel.
To achieve these goals, the embodiment of one aspect of the present invention provides a kind of hidden letter of judgement based on neutral net
The optimization method in road, comprises the steps:
Step s1, sets up bp neural network model;
Step s2, carries out learning training using default private communication channel set to described bp neural network model, training obtains
Optimal weights parameter, wherein, described default private communication channel set includes: pseudo- private communication channel and true private communication channel;
Step s3, the optimal weights parameter that training is obtained is added to described bp neural network model, after being updated
Bp neural network model;
Step s4, using the bp neural network model after updating to the actual private communication channel set including pseudo- private communication channel
Judged, to find out true private communication channel and pseudo- private communication channel.
Further, in described step s1, set up described bp neural network model, comprising: the definition number of setting input
According to outfan defines data, the definition of weighting function.
Further, in described step s2, pseudo- private communication channel in described default private communication channel set and truly hidden letter
Road, is preset and be would know that by user, to realize using default private communication channel set, described bp neural network model being carried out
There is the learning training of supervision.
The optimization device of the judgement private communication channel based on neutral net of the embodiment of the present invention, comprising: model building module,
For setting up bp neural network model;Learning training module, described learning training module is connected with described model building module, uses
In described bp neural network model being carried out with learning training using default private communication channel set, training obtains optimal weights parameter,
Wherein, described default private communication channel set includes: pseudo- private communication channel and true private communication channel, and the optimal weights that training is obtained
Parameter is added to described bp neural network model, with the bp neural network model after being updated;Private communication channel judge module, uses
In being judged to the actual private communication channel set including pseudo- private communication channel using the bp neural network model after updating, to look into
Find out true private communication channel and pseudo- private communication channel.
Further, described model building module sets up described bp neural network model, comprising: the definition number of setting input
According to outfan defines data, the definition of weighting function.
Further, the pseudo- private communication channel in described default private communication channel set and true private communication channel, are carried out pre- by user
And if would know that, to realize instructing using the study that default private communication channel set carries out having supervision to described bp neural network model
Practice.
The optimization method of the judgement private communication channel based on neutral net according to embodiments of the present invention, using bp neutral net
Reaction type learning style, to comprising the study that the channel set that true private communication channel is with pseudo- private communication channel carries out having supervision, from
And train optimum weight parameter, and then optimum weight parameter is added in bp neural network model, using this model
True private communication channel and pseudo- private communication channel is found out in the new set comprising suspicious private communication channel.
The aspect that the present invention adds and advantage will be set forth in part in the description, and partly will become from the following description
Obtain substantially, or recognized by the practice of the present invention.
Brief description
The above-mentioned and/or additional aspect of the present invention and advantage will become from reference to the description to embodiment for the accompanying drawings below
Substantially and easy to understand, wherein:
Fig. 1 is the flow chart of the optimization method according to the embodiment of the present invention based on the judgement private communication channel of neutral net;
Fig. 2 is the structure chart according to the embodiment of the present invention based on the optimization device of the judgement private communication channel of neutral net.
Specific embodiment
Embodiments of the invention are described below in detail, the example of described embodiment is shown in the drawings, wherein from start to finish
The element that same or similar label represents same or similar element or has same or like function.Below with reference to attached
The embodiment of figure description is exemplary it is intended to be used for explaining the present invention, and is not considered as limiting the invention.
The embodiment of the present invention proposes a kind of optimization method and device of the judgement private communication channel based on neutral net, is based on
The optimized algorithm of the judgement private communication channel of neutral net can be concentrated from private communication channel and find out pseudo- private communication channel and truly hidden
Channel.
First below the neutral net that the present invention is suitable for is introduced.
Artificial neural network (artificial neural networks, be abbreviated as anns) is also referred to as neutral net
(nns) or referred to as link model (connection model), it be a kind of imitate animal nerve network behavior feature, carry out point
The algorithm mathematics model that cloth parallel information is processed.This network relies on the complexity of system, by the internal a large amount of sections of adjustment
Interconnective relation between point, thus reach the purpose of processing information.A kind of important neutral net is bp nerve net
Network.
Bp (back propagation) neutral net is 1986 by the science headed by rumelhart and mccelland
Group of family proposes, and is a kind of Multi-layered Feedforward Networks by Back Propagation Algorithm training, is most widely used nerve net at present
One of network model.Bp network can learn and store substantial amounts of input-output mode map relation, and need not disclose in advance describe this
Plant the math equation of mapping relations.Its learning rules are to use steepest descent method, constantly adjust network by back propagation
Weights and threshold value, make the error sum of squares of network minimum.Bp neural network model topological structure include input layer (input),
Hidden layer (hidden layer) and output layer (output layer).Using bp neutral net, can calculate to stochastic problem
Optimal solution.
Artificial neural network is exactly the second way simulating people's thinking.This is a Kind of Nonlinear Dynamical System, and it is special
Color is that the distributed storage of information and concurrent collaborative are processed.Although the structure of single neuron is extremely simple, function is limited,
The behavior achieved by network system that a large amount of neurons are constituted but is extremely colourful.
Artificial neural network first has to be learnt with certain learning criterion, then could work.Now with artificial neuron
Network to illustrate as a example hand-written " a ", " b " two alphabetical identifications it is stipulated that when " a " input network when it should output " 1 ",
And when inputting as " b ", it is output as " 0 ".
So the criterion of e-learning should be: if the judgement that network work makes mistake, by the study of network, should make
Obtain the probability that network reduces next time and makes equally mistake.First, to random in each connection weight imparting (0,1) interval of network
Value, the image pattern corresponding to " a " is inputed to network, and input pattern weighted sum compared by network with thresholding, it is non-to carry out again
Linear operation, obtains the output of network.In the case, network be output as the probability of " 1 " and " 0 " be respectively 50% that is to say, that
It is completely random.If being at this moment output as " 1 " (result is correct), connection weight is made to increase, to make network run into again
During the input of " a " pattern, still can make correct judgement.
If being output as " 0 " (i.e. result mistake), network connection weights towards the side reducing comprehensive weighted input value
To adjustment, when its object is to make network to run into the input of " a " pattern again next time, reduce and makes equally wrong probability.So grasp
Adjust, after inputting several hand-written letter " a ", " b " by turns to network, carry out by above learning method through network some
After secondary study, the accuracy that network judges will greatly improve.This explanation network has been obtained for into the study of this two patterns
Work(, it is remembered this two pattern distributions in each connection weight of network.When network runs into any of which one again
During individual pattern, rapid, accurate judgement can be made and identify.It is, in general, that contained neuron number in network is more, then
The pattern that it can remember, identify is also more.
The optimization method and device of the judgement private communication channel based on neutral net of the embodiment of the present invention, exactly utilizes bp god
This learning capacity through network, by the private communication channel set comprising pseudo- private communication channel is carried out with the learning training having supervision,
To train optimal weights parameter, thus finding out really hidden letter in the new private communication channel set comprising pseudo- private communication channel
Road.
As shown in figure 1, the optimization method of the judgement private communication channel based on neutral net of the embodiment of the present invention, including as follows
Step:
Step s1, sets up bp neural network model.
In this step, set up bp neural network model, comprising: the definition data of setting input, outfan defines number
According to the definition of weighting function.This bp neutral net comprises reaction type learning capacity.
Step s2, carries out learning training using default private communication channel set to bp neural network model, and training obtains optimum
Weight parameter, wherein, default private communication channel set includes: pseudo- private communication channel and true private communication channel.
Specifically, the pseudo- private communication channel in default private communication channel set and true private communication channel, is carried out presetting simultaneously by user
Would know that, using default private communication channel set, bp neural network model is carried out with realization with the learning training having supervision.
Step s3, the optimal weights parameter that training is obtained is added to bp neural network model, with the bp after being updated
Neural network model.
Step s4, using the bp neural network model after updating to the actual private communication channel set including pseudo- private communication channel
Judged, to find out true private communication channel and pseudo- private communication channel.
As shown in Fig. 2 the embodiment of the present invention also proposes a kind of optimization device of the judgement private communication channel based on neutral net,
Including: model building module 1, learning training module 2 and private communication channel judge module 3.
Specifically, model building module 1 is used for setting up bp neural network model.
In one embodiment of the invention, model building module 1 sets up bp neural network model, comprising: setting input
The definition data at end, outfan defines data, the definition of weighting function.This bp neutral net comprises reaction type learning capacity.
Learning training module 2 is connected with model building module 1, for utilizing default private communication channel set to bp neutral net
Model carries out learning training, and training obtains optimal weights parameter. wherein, default private communication channel set includes: pseudo- private communication channel and
True private communication channel, and the optimal weights parameter that training is obtained is added to bp neural network model, with the bp after being updated
Neural network model.
It should be noted that the pseudo- private communication channel in default private communication channel set and true private communication channel, carried out by user
Preset and would know that, using default private communication channel set, bp neural network model is carried out with realization with the learning training having supervision.
Private communication channel judge module 3 is used for using the bp neural network model after updating to actual inclusion puppet private communication channel
Private communication channel set judged, to find out true private communication channel and pseudo- private communication channel.
The optimization method of the judgement private communication channel based on neutral net according to embodiments of the present invention, using bp neutral net
Reaction type learning style, to comprising the study that the channel set that true private communication channel is with pseudo- private communication channel carries out having supervision, from
And train optimum weight parameter, and then optimum weight parameter is added in bp neural network model, using this model
True private communication channel and pseudo- private communication channel is found out in the new set comprising suspicious private communication channel.
In the description of this specification, reference term " embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or the spy describing with reference to this embodiment or example
Point is contained at least one embodiment or the example of the present invention.In this manual, to the schematic representation of above-mentioned term not
Necessarily refer to identical embodiment or example.And, the specific features of description, structure, material or feature can be any
One or more embodiments or example in combine in an appropriate manner.
Although embodiments of the invention have been shown and described above it is to be understood that above-described embodiment is example
Property it is impossible to be interpreted as limitation of the present invention, those of ordinary skill in the art is in the principle without departing from the present invention and objective
In the case of above-described embodiment can be changed within the scope of the invention, change, replace and modification.The scope of the present invention
By claims and its equivalent limit.
Claims (6)
1. a kind of optimization method of the judgement private communication channel based on neutral net is it is characterised in that comprise the steps:
Step s1, sets up bp neural network model;
Step s2, carries out learning training using default private communication channel set to described bp neural network model, and training obtains optimum
Weight parameter, wherein, described default private communication channel set includes: pseudo- private communication channel and true private communication channel;
Step s3, the optimal weights parameter that training is obtained is added to described bp neural network model, with the bp after being updated
Neural network model;
Step s4, is carried out to the actual private communication channel set including pseudo- private communication channel using the bp neural network model after updating
Judge, to find out true private communication channel and pseudo- private communication channel.
2. the optimization method of the judgement private communication channel based on neutral net as claimed in claim 1 is it is characterised in that described
In step s1, set up described bp neural network model, comprising: the definition data of setting input, outfan defines data, weighting
The definition of function.
3. the optimization method of the judgement private communication channel based on neutral net as claimed in claim 1 is it is characterised in that described
In step s2, the pseudo- private communication channel in described default private communication channel set and true private communication channel, carried out default and can by user
Know, using default private communication channel set, described bp neural network model is carried out with realization with the learning training having supervision.
4. a kind of optimization device of the judgement private communication channel based on neutral net is it is characterised in that include:
Model building module, is used for setting up bp neural network model;
Learning training module, described learning training module is connected with described model building module, for using default private communication channel
Set carries out learning training to described bp neural network model, and training obtains optimal weights parameter, wherein, described default hidden letter
Road set includes: pseudo- private communication channel and true private communication channel, and the optimal weights parameter that training is obtained is added to described bp god
Through network model, with the bp neural network model after being updated;
Private communication channel judge module, for including the hidden of pseudo- private communication channel using the bp neural network model after updating to actual
Cover channel set to be judged, to find out true private communication channel and pseudo- private communication channel.
5. the optimization device of the judgement private communication channel based on neutral net as claimed in claim 4 is it is characterised in that described mould
Type is set up module and is set up described bp neural network model, comprising: the definition data of setting input, and outfan defines data, plus
The definition of weight function.
6. the optimization device of the judgement private communication channel based on neutral net as claimed in claim 4 is it is characterised in that described pre-
If the pseudo- private communication channel in private communication channel set and true private communication channel, preset and be would know that by user, to realize utilizing
Default private communication channel set carries out the learning training having supervision to described bp neural network model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610777841.2A CN106355250B (en) | 2016-08-31 | 2016-08-31 | The optimization method and device of judgement private communication channel neural network based |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610777841.2A CN106355250B (en) | 2016-08-31 | 2016-08-31 | The optimization method and device of judgement private communication channel neural network based |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106355250A true CN106355250A (en) | 2017-01-25 |
CN106355250B CN106355250B (en) | 2019-04-30 |
Family
ID=57856660
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610777841.2A Active CN106355250B (en) | 2016-08-31 | 2016-08-31 | The optimization method and device of judgement private communication channel neural network based |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106355250B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109214193A (en) * | 2017-07-05 | 2019-01-15 | 阿里巴巴集团控股有限公司 | Data encryption, machine learning model training method, device and electronic equipment |
CN112836214A (en) * | 2019-11-22 | 2021-05-25 | 南京聚铭网络科技有限公司 | Communication protocol hidden channel detection method |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050084032A1 (en) * | 2003-08-04 | 2005-04-21 | Lowell Rosen | Wideband holographic communications apparatus and methods |
CN101257417A (en) * | 2008-03-25 | 2008-09-03 | 浙江大学 | Method for detecting TCP/IP protocol concealed channel based on fuzzy neural network |
JP4276319B2 (en) * | 1998-12-08 | 2009-06-10 | 佳恭 武藤 | Learning method of neural network |
CN103279414A (en) * | 2013-05-23 | 2013-09-04 | 北京大学 | Covert channel detection method suitable for Xen virtualization platform |
CN104753617A (en) * | 2015-03-17 | 2015-07-01 | 中国科学技术大学苏州研究院 | Detection method of time-sequence type covert channel based on neural network |
CN105142177A (en) * | 2015-08-05 | 2015-12-09 | 西安电子科技大学 | Complex neural network channel prediction method |
CN105335790A (en) * | 2014-06-27 | 2016-02-17 | 上海电机学院 | Wind electricity power ultra-short term prediction method |
CN105469144A (en) * | 2015-11-19 | 2016-04-06 | 东北大学 | Mobile communication user loss prediction method based on particle classification and BP neural network |
-
2016
- 2016-08-31 CN CN201610777841.2A patent/CN106355250B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4276319B2 (en) * | 1998-12-08 | 2009-06-10 | 佳恭 武藤 | Learning method of neural network |
US20050084032A1 (en) * | 2003-08-04 | 2005-04-21 | Lowell Rosen | Wideband holographic communications apparatus and methods |
CN101257417A (en) * | 2008-03-25 | 2008-09-03 | 浙江大学 | Method for detecting TCP/IP protocol concealed channel based on fuzzy neural network |
CN103279414A (en) * | 2013-05-23 | 2013-09-04 | 北京大学 | Covert channel detection method suitable for Xen virtualization platform |
CN105335790A (en) * | 2014-06-27 | 2016-02-17 | 上海电机学院 | Wind electricity power ultra-short term prediction method |
CN104753617A (en) * | 2015-03-17 | 2015-07-01 | 中国科学技术大学苏州研究院 | Detection method of time-sequence type covert channel based on neural network |
CN105142177A (en) * | 2015-08-05 | 2015-12-09 | 西安电子科技大学 | Complex neural network channel prediction method |
CN105469144A (en) * | 2015-11-19 | 2016-04-06 | 东北大学 | Mobile communication user loss prediction method based on particle classification and BP neural network |
Non-Patent Citations (1)
Title |
---|
唐彰国等: "基于量子神经网络的启发式网络隐蔽信道检测模型", 《计算机应用研究》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109214193A (en) * | 2017-07-05 | 2019-01-15 | 阿里巴巴集团控股有限公司 | Data encryption, machine learning model training method, device and electronic equipment |
CN112836214A (en) * | 2019-11-22 | 2021-05-25 | 南京聚铭网络科技有限公司 | Communication protocol hidden channel detection method |
Also Published As
Publication number | Publication date |
---|---|
CN106355250B (en) | 2019-04-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9619749B2 (en) | Neural network and method of neural network training | |
KR101774844B1 (en) | Self-learning module of artificial neural network model using neuro-blockchain combination | |
KR101787613B1 (en) | Artificial intelligence platform system by self-adaptive learning technique based on deep learning | |
Yuan et al. | Resilient strategy design for cyber-physical system under DoS attack over a multi-channel framework | |
CN109711529A (en) | A kind of cross-cutting federal learning model and method based on value iterative network | |
CN106254330A (en) | A kind of software defined network intrusion detection method based on BP neutral net | |
CN103620624B (en) | For the method and apparatus causing the local competition inquiry learning rule of sparse connectivity | |
CN109375514B (en) | Design method of optimal tracking controller in presence of false data injection attack | |
Lanka et al. | Archer: Aggressive rewards to counter bias in hindsight experience replay | |
Aguilar | Adaptive random fuzzy cognitive maps | |
CN105427241A (en) | Distortion correction method for large-field-of-view display device | |
Turčaník | Packet filtering by artificial neural network | |
CN112748941A (en) | Feedback information-based target application program updating method and device | |
CN108280207A (en) | A method of the perfect Hash of construction | |
CN106355250A (en) | Optimization method and device for judging convert channels based on neural network | |
CN109800517A (en) | Improved reverse modeling method for magnetorheological damper | |
Liu et al. | An adaptive black-box defense against trojan attacks (trojdef) | |
KR101787611B1 (en) | Self-adaptive learning engine module based on deep learning | |
CN109102067A (en) | The method of increase and decrease certainly, computer equipment and the storage medium of neural network node | |
Ito et al. | Generalization in learning multiple temporal patterns using rnnpb | |
CN109697511B (en) | Data reasoning method and device and computer equipment | |
CN116127371B (en) | Multi-user model joint iteration method integrating prior distribution and homomorphic chaotic encryption | |
Petkov et al. | Generalized net model of encrypting message in an image using self organizing map neural network | |
Al-Shakarchy | Simulating DES Algorithm Using Artificial Neural Network | |
Hofmann | Deep Reinforcement Learning for Configuration of Time-Sensitive-Networking |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |