CN109785117A - Air control method, computer readable storage medium and server neural network based - Google Patents
Air control method, computer readable storage medium and server neural network based Download PDFInfo
- Publication number
- CN109785117A CN109785117A CN201811539696.XA CN201811539696A CN109785117A CN 109785117 A CN109785117 A CN 109785117A CN 201811539696 A CN201811539696 A CN 201811539696A CN 109785117 A CN109785117 A CN 109785117A
- Authority
- CN
- China
- Prior art keywords
- assessment
- data source
- client
- entry
- neural network
- 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
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 17
- 238000003062 neural network model Methods 0.000 claims abstract description 43
- 238000012502 risk assessment Methods 0.000 claims abstract description 43
- 238000011156 evaluation Methods 0.000 claims abstract description 40
- 239000011159 matrix material Substances 0.000 claims description 72
- 238000012549 training Methods 0.000 claims description 34
- 238000012545 processing Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 4
- 210000005036 nerve Anatomy 0.000 claims 2
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 5
- 238000013475 authorization Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000012854 evaluation process Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- COCAUCFPFHUGAA-MGNBDDOMSA-N n-[3-[(1s,7s)-5-amino-4-thia-6-azabicyclo[5.1.0]oct-5-en-7-yl]-4-fluorophenyl]-5-chloropyridine-2-carboxamide Chemical compound C=1C=C(F)C([C@@]23N=C(SCC[C@@H]2C3)N)=CC=1NC(=O)C1=CC=C(Cl)C=N1 COCAUCFPFHUGAA-MGNBDDOMSA-N 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to field of computer technology more particularly to a kind of air control method, computer readable storage medium and servers neural network based.The risk assessment that the method receiving terminal apparatus is sent is requested, and the identity information and evaluation type of client are extracted from risk assessment request；It determines assessment entry set corresponding with the evaluation type, includes at least one assessment entry in the assessment entry set；Data source corresponding with each assessment entry is chosen respectively from preset data source list as target data source, the data source list has recorded data source and assesses the corresponding relationship between entry, and historical record data relevant at least one assessment entry is had recorded in each data source；The historical record data of the client is obtained from each target data source according to the identity information of the client；It is handled using historical record data of the preset neural network model to the client, obtains the risk assessment grade of the client.
Description
Technical field
The invention belongs to field of computer technology more particularly to a kind of air control method neural network based, computer can
Read storage medium and server.
Background technique
In the prior art when carrying out loan risk evaluation, tool is usually directed to according to the past experience of oneself by business personnel
The scene of body carries out manual evaluation, relies primarily on the judgement of business personnel individual, subjectivity is extremely strong, and finally obtained assessment result is past
It is lower toward accuracy rate, easily resulted in significant economic losses because risk is judged by accident.
Summary of the invention
In view of this, the embodiment of the invention provides a kind of air control methods neural network based, computer-readable storage
Medium and server, subjectivity is extremely strong when solving the problems, such as manually to carry out loan risk evaluation and accuracy rate is lower.
The first aspect of the embodiment of the present invention provides a kind of air control method neural network based, may include:
The risk assessment request that receiving terminal apparatus is sent, and the identity of extraction client is believed from risk assessment request
Breath and evaluation type；
It determines assessment entry set corresponding with the evaluation type, is commented in the assessment entry set including at least one
Estimate entry；
Corresponding with each assessment entry data source is chosen respectively from preset data source list as target data source,
The data source list has recorded data source and assesses the corresponding relationship between entry, has recorded in each data source and at least one
The relevant historical record data of a assessment entry；
The historical record data of the client is obtained from each target data source according to the identity information of the client；
It is handled using historical record data of the preset neural network model to the client, obtains the client's
Risk assessment grade.
The second aspect of the embodiment of the present invention provides a kind of computer readable storage medium, the computer-readable storage
Media storage has computer-readable instruction, and the computer-readable instruction realizes following steps when being executed by processor:
The risk assessment request that receiving terminal apparatus is sent, and the identity of extraction client is believed from risk assessment request
Breath and evaluation type；
It determines assessment entry set corresponding with the evaluation type, is commented in the assessment entry set including at least one
Estimate entry；
Corresponding with each assessment entry data source is chosen respectively from preset data source list as target data source,
The data source list has recorded data source and assesses the corresponding relationship between entry, has recorded in each data source and at least one
The relevant historical record data of a assessment entry；
The historical record data of the client is obtained from each target data source according to the identity information of the client；
It is handled using historical record data of the preset neural network model to the client, obtains the client's
Risk assessment grade.
The third aspect of the embodiment of the present invention provides a kind of server, including memory, processor and is stored in institute
The computer-readable instruction that can be run in memory and on the processor is stated, the processor executes described computer-readable
Following steps are realized when instruction:
The risk assessment request that receiving terminal apparatus is sent, and the identity of extraction client is believed from risk assessment request
Breath and evaluation type；
It determines assessment entry set corresponding with the evaluation type, is commented in the assessment entry set including at least one
Estimate entry；
Corresponding with each assessment entry data source is chosen respectively from preset data source list as target data source,
The data source list has recorded data source and assesses the corresponding relationship between entry, has recorded in each data source and at least one
The relevant historical record data of a assessment entry；
The historical record data of the client is obtained from each target data source according to the identity information of the client；
It is handled using historical record data of the preset neural network model to the client, obtains the client's
Risk assessment grade.
Existing beneficial effect is the embodiment of the present invention compared with prior art: the embodiment of the present invention is receiving server hair
After the risk assessment request sent, the identity information and evaluation type of client are extracted, determines assessment corresponding with the evaluation type
Then entry set chooses data source corresponding with each assessment entry as number of targets respectively from preset data source list
According to source, and the historical record data of the client is obtained according to the identity information of the client from each target data source, most
It is handled afterwards using historical record data of the preset neural network model to the client, the risk for obtaining the client is commented
Estimate grade.Since neural network model replaces manually performing evaluation process, reduces the dependence to business personnel personal experience, assess
As a result more objective, and since these historical record datas are obtained from multiple data sources, risk assessment is carried out to be accurate
Data basis is provided, the accuracy rate of assessment result is substantially increased.
Detailed description of the invention
It to describe the technical solutions in the embodiments of the present invention more clearly, below will be to embodiment or description of the prior art
Needed in attached drawing be briefly described, it should be apparent that, the accompanying drawings in the following description is only of the invention some
Embodiment for those of ordinary skill in the art without any creative labor, can also be according to these
Attached drawing obtains other attached drawings.
Fig. 1 is a kind of one embodiment flow chart of air control method neural network based in the embodiment of the present invention；
Fig. 2 is the schematic flow diagram that the historical record data of client is obtained from each target data source；
Fig. 3 is the interaction schematic diagram for obtaining historical record data；
Fig. 4 is the schematic flow diagram being trained to neural network model；
Fig. 5 is a kind of one embodiment structure chart of wind-controlling device neural network based in the embodiment of the present invention；
Fig. 6 is a kind of schematic block diagram of server in the embodiment of the present invention.
Specific embodiment
In order to make the invention's purpose, features and advantages of the invention more obvious and easy to understand, below in conjunction with the present invention
Attached drawing in embodiment, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that disclosed below
Embodiment be only a part of the embodiment of the present invention, and not all embodiment.Based on the embodiments of the present invention, this field
Those of ordinary skill's all other embodiment obtained without making creative work, belongs to protection of the present invention
Range.
Referring to Fig. 1, a kind of one embodiment of air control method neural network based can wrap in the embodiment of the present invention
It includes:
Step S101, the risk assessment request that receiving terminal apparatus is sent, and visitor is extracted from risk assessment request
The identity information and evaluation type at family.
Client specified can be answered when borrowing money by what is installed on the terminal devices such as mobile phone, tablet computer
Risk assessment request is sent to server with program (APP).The risk assessment request in carry client identity information and
Evaluation type, the identity information of client can be that ID card No., social security number or drivers license number etc. can be with unique identification visitors
The information of family identity, evaluation type can include but is not limited to fiduciary loan, mortgage loan and consumptive loan etc..
Server can therefrom extract the identity information and assessment class of client after receiving the risk assessment request
Type.
Step S102, determination assessment entry set corresponding with the evaluation type.
It wherein, include at least one assessment entry in the assessment entry set.Each assessment entry is to carry out to client
Information in some dimension needed for risk assessment, such as: income information, existing credit information, people's row reference information, social security
Information, common reserve fund information, education background information, previous conviction information etc..
Difference between different evaluation types essentially consists in the assessment entry respectively specifically included and is different, such as:
Evaluation type 1={ assessment entry A ∪ assessment entry B ∪ assesses entry C }
Evaluation type 2={ assessment entry A ∪ assessment entry B ∪ assesses entry E }
Evaluation type 3={ assessment entry A ∪ assessment entry D ∪ assesses entry E }.
It should be noted that can be each evaluation type according to specific scene in practical application the above is only example
Corresponding assessment entry set is set.
Step S103, data source corresponding with each assessment entry is chosen respectively from preset data source list as mesh
Mark data source.
The data source list has recorded data source and assesses the corresponding relationship between entry, shown in table specific as follows:
| Assess entry | Data source |
| Assess entry A | Data source 1 |
| Assess entry B | Data source 2 |
| Assess entry C | Data source 3 |
| …… | …… |
Historical record data relevant at least one assessment entry is had recorded in each data source.For example, bank service
The income information of the client stored in device has credit information, people's row reference information, deposits in the server of management of social insurance department
The social security information of the client of storage, the common reserve fund information of the client stored in the server of public relation education department, education pipe
The criminal of the client stored in the education background information of the client stored in the server of reason department, the server of public security organ
Crime record information ... ... .. etc..
Step S104, the history note of the client is obtained from each target data source according to the identity information of the client
Record data.
Specifically, step S104 may include step as shown in Figure 2:
Step S1041, the terminal device of Xiang Suoshu client sends identity information request.
It include the device identification for executing terminal device in the identity information request, the execution terminal device is this implementation
Executing subject in example, namely carry out the server of risk assessment.
Step S1042, the identity information of the client of the terminal device feedback of the client is received.
The terminal device of the client will record setting for the lower execution terminal device after receiving identity information request
Standby mark, and feed back to the execution terminal device identity information of the client.
Step S1043, the data source conduct being not yet selected arbitrarily is chosen from each target data source
Current data source.
Step S1044, server corresponding with the current data source is chosen from preset server list as mesh
Mark server.
The server list has recorded the corresponding relationship between each data source and each server, table institute specific as follows
Show:
| Data source | Server (IP address) |
| Data source 1 | 192.168.3.56 |
| Data source 2 | 192.155.26.134 |
| Data source 3 | 192.38.80.121 |
| Data source 4 | 192.176.34.5 |
Step S1045, Xiang Suoshu destination server sends request of data.
Include the identity information of the client in the request of data, further includes the equipment mark for executing terminal device
Know.
Step S1046, the historical record data for the client that the destination server is sent is received.
After receiving the request of data, the terminal device of Xiang Suoshu client sends authorization and asks the destination server
It asks, includes the device identification for executing terminal device in the authorization requests, the terminal device of the client is to the execution
The device identification of terminal device is checked, if confirmation, sends authorized order, the target to the destination server
Server sends the historical record data of the client to the execution terminal device after receiving the authorized order.
The process of entire data interaction is as shown in Figure 3.
Step S1047, judge whether each target data source was selected.
If returning to step S1043 there is also the data source being not yet selected in each target data source,
If each target data source was selected, S1048 is thened follow the steps.
Step S1048, determine that historical record data acquisition has obtained success.
By above procedure, under the premise of obtaining user's authorization, then going through for client is obtained from each destination server
The Records of the Historian records data, ensure that the safety of customer data.It, then can be according to these after the historical record data has obtained
Historical record data is that client carries out risk assessment.
Step S105, it is handled, is obtained using historical record data of the preset neural network model to the client
The risk assessment grade of the client.
The training process of the neural network model may include step as shown in Figure 4:
Step S401, the assessment sample of preset number is chosen from historical evaluation record.
The details of passing all loan customers are had recorded in historical evaluation record, the details of each client are
For a sample, each sample is provided with label value.
For example, setting 0 for label value if the client does not occur any promise breaking after loan；If the client is providing a loan
Occur slight promise breaking afterwards, then sets 1 for label value；If conventional promise breaking occurs after loan in the client, label value is arranged
It is 2；If grave breach of contract occurs after loan in the client, 3 are set by label value.Wherein, promise breaking degree can be according to promise breaking
The amount of money and default time determine that the promise breaking amount of money is bigger, and default time is longer, then degree of breaking a contract is more serious, the promise breaking amount of money is smaller,
Default time is shorter, then degree of breaking a contract is slighter.
The assessment sample includes the sample of each risk assessment grade, and in the sample of selection, keeps various labels
The number of samples of value is consistent.For example, if 10000 assessment samples are chosen altogether, wherein each label value number of samples is
2500, guarantee the balance of training result with this.
Further, as far as possible guarantee sample in information covering it is comprehensive, for example, just income this assessment entry of information and
Speech, will cover: lower than 5000 yuan, 5000 yuan to 10,000 yuan, 10,000 yuan to 50,000 yuan, 50,000 yuan to 100,000 yuan, 100,000 yuan with first-class
It etc. each range, for existing this assessment entry of credit information, to cover: without loan, loan lower than 100,000 yuan, loan
100000 to 1,000,000, each range of 1,000,000 or more loan etc..
By the above process, the sample of equiblibrium mass distribution has been selected, these samples can be used and carry out neural network model
Training.
Step S402, numeralization processing is carried out to data of the assessment sample in each assessment entry, obtains numerical value
The assessment sample of change.
Before being trained, numeralization processing can be carried out to each assessment entry of sample.
Specifically, for assessing people's row reference information in entry, record of bad behavior number therein is counted, and be arranged not
The threshold value of good record, if being more than the threshold value, people's row reference information value turns to 1, if being less than the threshold value, calculates record of bad behavior
The ratio of number and the threshold value, using the ratio as numeralization treated result.Distinguishingly, if record of bad behavior number is 0,
People's row reference information value turns to 0.
Again by taking social security information as an example, count successive tranche social security months therein and current total value, and months threshold is set
Value and sum threshold, if successive tranche social security months are more than months threshold value, the first numerical value of social security information is 1, should if being less than
Threshold value then calculates the ratio of successive tranche social security months Yu the threshold value, using the ratio as the first numerical value, similarly, according to current
Total value and sum threshold calculate the second value of social security information, the average value of the two are finally sought, at social security Information Number value
Result after reason.
It is handled by the above numeralization, the value each assessment entry of sample being converted in [0,1] section.
Step S403, by the assessment sample composition assessment sample matrix of the numeralization, and the assessment sample moment is calculated
The covariance matrix of battle array.
It is possible, firstly, to which the assessment sample of the numeralization is formed following assessment sample matrix:
Any data line of the assessment sample matrix is corresponding with the assessment sample that one quantizes.Wherein, X is described
Assess sample matrix, xijValue of the assessment sample to quantize for i-th in j-th of assessment entry, 1≤i≤n, 1≤j≤
P, n are the sum of the assessment sample of the numeralization, and p is the number of the assessment entry.
Then, the covariance matrix of the assessment sample matrix is calculated according to the following formula:
Wherein, R is the covariance matrix of the assessment sample matrix,
Step S404, the characteristic value of the covariance matrix of the assessment sample matrix is calculated, and is selected from the characteristic value
Take the maximum characteristic value of the numerical value of preset number as dominant eigenvalue.
Firstly, solution characteristic equation | λ I-R |=0, find out eigenvalue λa, wherein I is unit matrix, and R is the assessment sample
The covariance matrix of this matrix, 1≤a≤p, P are the number of the assessment entry；
Then, the contribution rate of each characteristic value is calculated according to the following formula:
Wherein, ηaIt is characterized value λaContribution rate.
Finally, using the maximum preceding m characteristic value of the numerical value for meeting following condition as dominant eigenvalue:
And
Wherein ηthresholdFor preset contribution rate threshold value.
Step S405, the assessment sample matrix is carried out simplifying processing, the assessment sample matrix after being simplified.
The simplified assessment sample matrix can indicate are as follows:
Only retain column corresponding with the dominant eigenvalue in the simplified assessment sample matrix, and deletes other
Column.
Step S406, preset neural network model is trained using the simplified assessment sample matrix, is obtained
To trained neural network model.
Neural network model in the present embodiment may include input layer, hidden layer and output layer.The input layer is used for
Input data, including more than two input layers are received from outside, the hidden layer is used to handle data, including
More than two hidden layer nodes, the output layer is for exporting processing result, including an output node layer.
The input layer and the assessment entry correspond.For example, if a certain evaluation type shares 3 assessment items
Mesh, respectively assessment entry 1, assessment entry 2 and assessment entry 3, then the input layer of corresponding neural network model
Number also should be 3, respectively input layer 1, input layer 2 and input layer 3, wherein input layer 1 with comment
It is corresponding to estimate entry 1, input layer 2 is corresponding with assessment entry 2, and input layer 3 is corresponding with assessment entry 3.
Use fuzzy Gauss subordinating degree function to the input layer respectively in the hidden layer node of the neural network model
Node data is handled, and hidden layer node data are obtained.
In the present embodiment, the hidden layer node data can be obtained by following calculation formula:
Wherein:
I is the label of input layer, and value range is [1, n], and n is the number of input layer；
J is the label of hidden layer node, and value range is [1, h], and h is the number of hidden layer node；
ΦjIt (x) is the hidden layer node data of j-th of hidden layer node；
Gij(xi) be j-th of hidden layer node i-th of fuzzy Gauss subordinating degree function；
X is input layer data, xiFor the input layer data of i-th of input layer therein；
μijFor the mathematic expectaion of i-th of fuzzy Gauss subordinating degree function of j-th of hidden layer node；
σijFor the standard deviation of i-th of fuzzy Gauss subordinating degree function of j-th of hidden layer node.
Preferably, the hidden layer node data can also be normalized, to reduce the hidden layer node
The difference of data, specifically, maximum value and minimum value in the available hidden layer node data, then according to most
The hidden layer node data are normalized in big value and the minimum value, obtain normalized node in hidden layer
According to.
For example, the hidden layer node data can be normalized by following formula:
Wherein:
ΨjIt (x) is the normalized hidden layer node data of j-th of hidden layer node；
ΦmaxIt (x) is Φj(x) maximum value in；
ΦminIt (x) is Φj(x) minimum value in.
Summation is weighted to the hidden layer node data respectively using preset weight, obtains the risk of the client
Evaluation grade.
For the hidden layer node data not being normalized, the calculation formula of the risk assessment grade of the client can be with
Are as follows:
For normalized hidden layer node data, the calculation formula of the risk assessment grade of the client can be with are as follows:
Wherein:
ωjFor weight corresponding with the hidden layer node data of j-th of hidden layer node；
R (x) is output layer node data namely the risk assessment grade of the client.
During being trained to the neural network model, the simplified assessment sample matrix is used first
One wheel training is carried out to the neural network model, and calculates the global error of epicycle training according to the following formula:
Wherein, EtFor the training error of t-th of training sample, ztFor the training output valve of t-th of training sample, ctFor t
The theoretical output valve of a training sample, t-th of training sample are the t line number of the simplified assessment sample matrix
According to 1≤t≤n.
If the global error is greater than preset error threshold, the neural network model is adjusted, for example, can
With according to Delta learning rules between each node layer network connection weight and threshold value be adjusted, then return execute institute
The step for carrying out a wheel training to the neural network model using the simplified assessment sample matrix is stated, until described complete
Until office's error is less than the error threshold.
If the global error is less than the error threshold, current neural network model is determined as described train
Neural network model.
After training the neural network model, the neural network model may be used to the historical record of the client
Data are handled, and the risk assessment grade of the client is obtained.If risk assessment grade is 3, illustrate the promise breaking of the client
Risk is larger, can not make loans, or can only give the loan of the smaller amount of money, if risk assessment grade is 0, illustrates the visitor
The default risk at family is smaller, can preferentially give and make loans, and can give the loan of the larger amount of money.
In conclusion the embodiment of the present invention extracts the body of client after the risk assessment request that receiving terminal apparatus is sent
Part information and evaluation type determine assessment entry set corresponding with the evaluation type, then from preset data source list
It is middle to choose corresponding with each assessment entry data source respectively as target data source, and according to the identity information of the client from
The historical record data of the client is obtained in each target data source, finally using preset neural network model to the visitor
The historical record data at family is handled, and the risk assessment grade of the client is obtained.Since neural network model replaces manually
Evaluation process is executed, reduces the dependence to business personnel personal experience, assessment result is more objective, and due to these historical records
Data are obtained from multiple data sources, are provided data basis for accurate progress risk assessment, are substantially increased assessment knot
The accuracy rate of fruit.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit
It is fixed.
Implementation of the present invention is shown corresponding to a kind of air control method neural network based, Fig. 5 described in foregoing embodiments
A kind of one embodiment structure chart for wind-controlling device neural network based that example provides.
In the present embodiment, a kind of wind-controlling device neural network based may include:
Risk assessment request receiving module 501, for the risk assessment request that receiving terminal apparatus is sent, and from the wind
The identity information and evaluation type of client are extracted in the assessment request of danger；
Entry set determining module 502 is assessed, it is described for determining assessment entry set corresponding with the evaluation type
Assessing in entry set includes at least one assessment entry；
Target data source chooses module 503, for choosing and each assessment entry respectively from preset data source list
For corresponding data source as target data source, the data source list has recorded the corresponding pass between data source and assessment entry
It is that historical record data relevant at least one assessment entry is had recorded in each data source；
Historical record data obtains module 504, for the identity information according to the client from each target data source
Obtain the historical record data of the client；
Risk evaluation module 505, for use preset neural network model to the historical record data of the client into
Row processing, obtains the risk assessment grade of the client.
Further, the historical record data acquisition module may include:
Identity information request transmission unit, for sending identity information request to the terminal device of the client；
Feedback information receiving unit, the identity information for the client that the terminal device for receiving the client is fed back；
Current data source selection unit was not yet selected for arbitrarily choosing one from each target data source
Data source as current data source；
Destination server selection unit, it is corresponding with the current data source for being chosen from preset server list
For server as destination server, the server list has recorded the corresponding pass between each data source and each server
System；
Request of data transmission unit is used to send request of data to the destination server, includes in the request of data
The identity information of the client；
Historical record data receiving unit, for receiving the historical record number for the client that the destination server is sent
According to.
Further, the wind-controlling device can also include:
It assesses sample and chooses module, for choosing the assessment sample of preset number, the assessment in recording from historical evaluation
Sample includes the sample of each risk assessment grade；
Quantize processing module, for carrying out at numeralization to data of the assessment sample in each assessment entry
Reason, the assessment sample to be quantized；
Sample matrix constructing module is assessed, for the assessment sample composition of the numeralization to be assessed sample matrix, wherein
Any data line of the assessment sample matrix is corresponding with the assessment sample that one quantizes；
Covariance matrix computing module, for calculating the covariance matrix of the assessment sample matrix；
Dominant eigenvalue chooses module, the characteristic value of the covariance matrix for calculating the assessment sample matrix, and from institute
It states and chooses the maximum characteristic value of numerical value of preset number in characteristic value as dominant eigenvalue；
Simplify processing module, simplifies processing for carrying out to the assessment sample matrix, the assessment sample after being simplified
Matrix only retains column corresponding with the dominant eigenvalue in the simplified assessment sample matrix；
Model training module, for being carried out using the simplified assessment sample matrix to preset neural network model
Training, obtains trained neural network model.
Further, the dominant eigenvalue selection module may include:
Characteristic value solves unit, for solving characteristic equation | λ I-R |=0, and find out eigenvalue λa, wherein I is unit matrix,
R is the covariance matrix of the assessment sample matrix, and 1≤a≤p, P are the number of the assessment entry；
Contribution rate computing unit, for calculating the contribution rate of each characteristic value according to the following formula:
Wherein, ηaIt is characterized value λaContribution rate；
Dominant eigenvalue selection unit, the maximum preceding m characteristic value of numerical value for that will meet following condition is as main feature
Value:
AndWherein ηthresholdFor preset contribution rate threshold value.
Further, the model training module may include:
Global error computing unit, for using the simplified assessment sample matrix to the neural network model into
The wheel training of row one, and the global error of epicycle training is calculated according to the following formula:
Wherein, EtFor the training error of t-th of training sample, ztFor the training output valve of t-th of training sample, ctFor t
The theoretical output valve of a training sample, t-th of training sample are the t line number of the simplified assessment sample matrix
According to 1≤t≤n；
Model adjustment unit, if being greater than preset error threshold for the global error, to the neural network mould
Type is adjusted, and then return execution is described carries out the neural network model using the simplified assessment sample matrix
The step of one wheel training, until the global error is less than the error threshold；
Model determination unit, if being less than the error threshold for the global error, by current neural network mould
Type is determined as the trained neural network model.
It is apparent to those skilled in the art that for convenience and simplicity of description, the device of foregoing description,
The specific work process of module and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, is not described in detail or remembers in some embodiment
The part of load may refer to the associated description of other embodiments.
The schematic block diagram that Fig. 6 shows a kind of server provided in an embodiment of the present invention illustrates only for ease of description
Part related to the embodiment of the present invention.
In the present embodiment, the server 6 may include: processor 60, memory 61 and be stored in the storage
In device 61 and the computer-readable instruction 62 that can be run on the processor 60, such as execute above-mentioned neural network based
The computer-readable instruction of air control method.The processor 60 realizes above-mentioned each base when executing the computer-readable instruction 62
Step in the air control embodiment of the method for neural network, such as step S101 to S105 shown in FIG. 1.Alternatively, the processing
Device 60 realizes the function of each module/unit in above-mentioned each Installation practice, such as Fig. 5 when executing the computer-readable instruction 62
The function of shown module 501 to 505.
Illustratively, the computer-readable instruction 62 can be divided into one or more module/units, one
Or multiple module/units are stored in the memory 61, and are executed by the processor 60, to complete the present invention.Institute
Stating one or more module/units can be the series of computation machine readable instruction section that can complete specific function, the instruction segment
For describing implementation procedure of the computer-readable instruction 62 in the server 6.
The processor 60 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
The memory 61 can be the internal storage unit of the server 6, such as the hard disk or memory of server 6.
The memory 61 is also possible to the External memory equipment of the server 6, such as the plug-in type being equipped on the server 6 is hard
Disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card
(Flash Card) etc..Further, the memory 61 can also both include the internal storage unit of the server 6 or wrap
Include External memory equipment.The memory 61 is for storing needed for the computer-readable instruction and the server 6 it
Its instruction and data.The memory 61 can be also used for temporarily storing the data that has exported or will export.
The functional units in various embodiments of the present invention may be integrated into one processing unit, is also possible to each
Unit physically exists alone, and can also be integrated in one unit with two or more units.Above-mentioned integrated unit both may be used
To use formal implementation of hardware, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, technical solution of the present invention substantially or
Person says that all or part of the part that contributes to existing technology or the technical solution can body in the form of software products
Reveal and, which is stored in a storage medium, including several computer-readable instructions are used so that one
Platform computer equipment (can be personal computer, server or the network equipment etc.) executes described in each embodiment of the present invention
The all or part of the steps of method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-
Only Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can be with
Store the medium of computer-readable instruction.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations；Although referring to aforementioned reality
Applying example, invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features；And these are modified
Or replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of air control method neural network based characterized by comprising
Receiving terminal apparatus send risk assessment request, and from the risk assessment request in extract client identity information and
Evaluation type；
It determines assessment entry set corresponding with the evaluation type, includes at least one assessment item in the assessment entry set
Mesh；
Selection data source corresponding with each assessment entry is described as target data source respectively from preset data source list
Data source list has recorded data source and assesses the corresponding relationship between entry, has recorded in each data source and comments at least one
Estimate the relevant historical record data of entry；
The historical record data of the client is obtained from each target data source according to the identity information of the client；
It is handled using historical record data of the preset neural network model to the client, obtains the risk of the client
Evaluation grade.
2. air control method according to claim 1, which is characterized in that the identity information according to the client is from each
The historical record data that the client is obtained in target data source includes:
Identity information request is sent to the terminal device of the client；
Receive the identity information of the client of the terminal device feedback of the client；
Data source that one was not yet selected arbitrarily is chosen from each target data source as current data source；
Server corresponding with the current data source is chosen from preset server list as destination server, the clothes
Corresponding relationship of the device list records of being engaged between each data source and each server；
Request of data is sent to the destination server, includes the identity information of the client in the request of data；
Receive the historical record data for the client that the destination server is sent；
It returns arbitrarily to choose the data source that one was not yet selected described in executing from each target data source and be used as and work as
The step of preceding data source, until each target data source was selected.
3. air control method according to claim 1, which is characterized in that the training process of the neural network model includes:
The assessment sample of preset number is chosen from historical evaluation record, the assessment sample includes each risk assessment grade
Sample；
Numeralization processing is carried out to data of the assessment sample in each assessment entry, the assessment sample to be quantized；
By the assessment sample composition assessment sample matrix of the numeralization, and calculate the covariance square of the assessment sample matrix
Battle array, wherein any data line of the assessment sample matrix is corresponding with the assessment sample that one quantizes；
The characteristic value of the covariance matrix of the assessment sample matrix is calculated, and chooses the number of preset number from the characteristic value
It is worth maximum characteristic value as dominant eigenvalue；
The assessment sample matrix is carried out to simplify processing, the assessment sample matrix after being simplified, the simplified assessment
Only retain column corresponding with the dominant eigenvalue in sample matrix；
Preset neural network model is trained using the simplified assessment sample matrix, obtains trained nerve
Network model.
4. air control method according to claim 3, which is characterized in that the covariance for calculating the assessment sample matrix
The characteristic value of matrix, and the maximum characteristic value of numerical value of selection preset number includes: as dominant eigenvalue from the characteristic value
Solve characteristic equation | λ I-R |=0, find out eigenvalue λa, wherein I is unit matrix, and R is the association of the assessment sample matrix
Variance matrix, 1≤a≤p, P are the number of the assessment entry；
The contribution rate of each characteristic value is calculated according to the following formula:
Wherein, ηaIt is characterized value λaContribution rate；
Using the maximum preceding m characteristic value of the numerical value for meeting following condition as dominant eigenvalue:
AndWherein ηthresholdFor preset contribution rate threshold value.
5. air control method according to any one of claim 3 to 4, which is characterized in that described using described simplified
Assessment sample matrix is trained preset neural network model, and obtaining trained neural network model includes:
A wheel is carried out to the neural network model using the simplified assessment sample matrix to train, and is calculated according to the following formula
The global error of epicycle training:
Wherein, EtFor the training error of t-th of training sample, ztFor the training output valve of t-th of training sample, ctIt is instructed for t-th
Practice sample theoretical output valve, t-th of training sample be it is described it is simplified assessment sample matrix t row data, 1≤
t≤n；
If the global error is greater than preset error threshold, the neural network model is adjusted, then returns and holds
The row step for carrying out a wheel training to the neural network model using the simplified assessment sample matrix, until institute
Global error is stated less than until the error threshold；
If the global error is less than the error threshold, current neural network model is determined as the trained mind
Through network model.
6. a kind of computer readable storage medium, the computer-readable recording medium storage has computer-readable instruction, special
Sign is, the air control side as described in any one of claims 1 to 5 is realized when the computer-readable instruction is executed by processor
The step of method.
7. a kind of server, including memory, processor and storage can transport in the memory and on the processor
Capable computer-readable instruction, which is characterized in that the processor realizes following steps when executing the computer-readable instruction:
Receiving terminal apparatus send risk assessment request, and from the risk assessment request in extract client identity information and
Evaluation type；
It determines assessment entry set corresponding with the evaluation type, includes at least one assessment item in the assessment entry set
Mesh；
Selection data source corresponding with each assessment entry is described as target data source respectively from preset data source list
Data source list has recorded data source and assesses the corresponding relationship between entry, has recorded in each data source and comments at least one
Estimate the relevant historical record data of entry；
The historical record data of the client is obtained from each target data source according to the identity information of the client；
It is handled using historical record data of the preset neural network model to the client, obtains the risk of the client
Evaluation grade.
8. server according to claim 7, which is characterized in that the identity information according to the client is from each mesh
Obtaining the historical record data of the client in mark data source includes:
Identity information request is sent to the terminal device of the client；
Receive the identity information of the client of the terminal device feedback of the client；
Data source that one was not yet selected arbitrarily is chosen from each target data source as current data source；
Server corresponding with the current data source is chosen from preset server list as destination server, the clothes
Corresponding relationship of the device list records of being engaged between each data source and each server；
Request of data is sent to the destination server, includes the identity information of the client in the request of data；
Receive the historical record data for the client that the destination server is sent；
It returns arbitrarily to choose the data source that one was not yet selected described in executing from each target data source and be used as and work as
The step of preceding data source, until each target data source was selected.
9. server according to claim 7, which is characterized in that the training process of the neural network model includes:
The assessment sample of preset number is chosen from historical evaluation record, the assessment sample includes each risk assessment grade
Sample；
Numeralization processing is carried out to data of the assessment sample in each assessment entry, the assessment sample to be quantized；
By the assessment sample composition assessment sample matrix of the numeralization, and calculate the covariance square of the assessment sample matrix
Battle array, wherein any data line of the assessment sample matrix is corresponding with the assessment sample that one quantizes；
The characteristic value of the covariance matrix of the assessment sample matrix is calculated, and chooses the number of preset number from the characteristic value
It is worth maximum characteristic value as dominant eigenvalue；
The assessment sample matrix is carried out to simplify processing, the assessment sample matrix after being simplified, the simplified assessment
Only retain column corresponding with the dominant eigenvalue in sample matrix；
Preset neural network model is trained using the simplified assessment sample matrix, obtains trained nerve
Network model.
10. server according to claim 9, which is characterized in that the covariance for calculating the assessment sample matrix
The characteristic value of matrix, and the maximum characteristic value of numerical value of selection preset number includes: as dominant eigenvalue from the characteristic value
Solve characteristic equation | λ I-R |=0, find out eigenvalue λa, wherein I is unit matrix, and R is the association of the assessment sample matrix
Variance matrix, 1≤a≤p, P are the number of the assessment entry；
The contribution rate of each characteristic value is calculated according to the following formula:
Wherein, ηaIt is characterized value λaContribution rate；
Using the maximum preceding m characteristic value of the numerical value for meeting following condition as dominant eigenvalue:
AndWherein ηthresholdFor preset contribution rate threshold value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811539696.XA CN109785117A (en) | 2018-12-14 | 2018-12-14 | Air control method, computer readable storage medium and server neural network based |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811539696.XA CN109785117A (en) | 2018-12-14 | 2018-12-14 | Air control method, computer readable storage medium and server neural network based |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109785117A true CN109785117A (en) | 2019-05-21 |
Family
ID=66497179
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811539696.XA Pending CN109785117A (en) | 2018-12-14 | 2018-12-14 | Air control method, computer readable storage medium and server neural network based |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109785117A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110378388A (en) * | 2019-06-21 | 2019-10-25 | 大亚湾核电运营管理有限责任公司 | Flying object recognition methods, device, computer readable storage medium and terminal device |
| CN110458687A (en) * | 2019-07-05 | 2019-11-15 | 平安银行股份有限公司 | The automatic measures and procedures for the examination and approval of decision, device and computer readable storage medium |
| CN110490423A (en) * | 2019-07-22 | 2019-11-22 | 平安科技（深圳）有限公司 | Waste battery pollution assessment method and device, readable storage medium and server |
| CN111199476A (en) * | 2019-12-20 | 2020-05-26 | 上海淇玥信息技术有限公司 | Method and device for generating credit granting strategy based on academic prediction model and electronic equipment |
| CN112446617A (en) * | 2020-11-27 | 2021-03-05 | 平安普惠企业管理有限公司 | Risk assessment method and device, computer equipment and readable storage medium |
| CN112529429A (en) * | 2020-12-16 | 2021-03-19 | 平安科技（深圳）有限公司 | Customer information checking method and device, computer equipment and storage medium |
| CN112580917A (en) * | 2019-09-30 | 2021-03-30 | 深圳无域科技技术有限公司 | Client characteristic evaluation method and device |
| CN112990707A (en) * | 2021-03-12 | 2021-06-18 | 深圳工盟科技有限公司 | Construction risk assessment method, device, equipment and storage medium |
| CN115079738A (en) * | 2022-07-28 | 2022-09-20 | 苏州浪潮智能科技有限公司 | Method, device and equipment for regulating and controlling rotating speed of server fan and storage medium |
| CN115098508A (en) * | 2022-07-04 | 2022-09-23 | 成都秦川物联网科技股份有限公司 | Smart city check list generation method, system and device based on Internet of things |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256720A (en) * | 2017-11-07 | 2018-07-06 | 中国平安财产保险股份有限公司 | A kind of settlement of insurance claim methods of risk assessment and terminal device |
| CN108665366A (en) * | 2018-04-27 | 2018-10-16 | 平安科技（深圳）有限公司 | Determine method, terminal device and the computer readable storage medium of consumer's risk grade |
| CN108763530A (en) * | 2018-05-31 | 2018-11-06 | 中国农业银行股份有限公司 | A kind of data access method, device and electronic equipment |
-
2018
- 2018-12-14 CN CN201811539696.XA patent/CN109785117A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256720A (en) * | 2017-11-07 | 2018-07-06 | 中国平安财产保险股份有限公司 | A kind of settlement of insurance claim methods of risk assessment and terminal device |
| CN108665366A (en) * | 2018-04-27 | 2018-10-16 | 平安科技（深圳）有限公司 | Determine method, terminal device and the computer readable storage medium of consumer's risk grade |
| CN108763530A (en) * | 2018-05-31 | 2018-11-06 | 中国农业银行股份有限公司 | A kind of data access method, device and electronic equipment |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110378388B (en) * | 2019-06-21 | 2021-11-12 | 大亚湾核电运营管理有限责任公司 | Method and device for identifying flying object, computer readable storage medium and terminal equipment |
| CN110378388A (en) * | 2019-06-21 | 2019-10-25 | 大亚湾核电运营管理有限责任公司 | Flying object recognition methods, device, computer readable storage medium and terminal device |
| CN110458687A (en) * | 2019-07-05 | 2019-11-15 | 平安银行股份有限公司 | The automatic measures and procedures for the examination and approval of decision, device and computer readable storage medium |
| CN110490423A (en) * | 2019-07-22 | 2019-11-22 | 平安科技（深圳）有限公司 | Waste battery pollution assessment method and device, readable storage medium and server |
| CN110490423B (en) * | 2019-07-22 | 2024-02-27 | 平安科技（深圳）有限公司 | Waste battery pollution evaluation method and device, readable storage medium and server |
| CN112580917B (en) * | 2019-09-30 | 2024-04-05 | 深圳无域科技技术有限公司 | Customer characteristic evaluation method and device |
| CN112580917A (en) * | 2019-09-30 | 2021-03-30 | 深圳无域科技技术有限公司 | Client characteristic evaluation method and device |
| CN111199476A (en) * | 2019-12-20 | 2020-05-26 | 上海淇玥信息技术有限公司 | Method and device for generating credit granting strategy based on academic prediction model and electronic equipment |
| CN112446617A (en) * | 2020-11-27 | 2021-03-05 | 平安普惠企业管理有限公司 | Risk assessment method and device, computer equipment and readable storage medium |
| WO2022126975A1 (en) * | 2020-12-16 | 2022-06-23 | 平安科技（深圳）有限公司 | Client information verification method and apparatus, and computer device and storage medium |
| CN112529429A (en) * | 2020-12-16 | 2021-03-19 | 平安科技（深圳）有限公司 | Customer information checking method and device, computer equipment and storage medium |
| CN112529429B (en) * | 2020-12-16 | 2024-05-14 | 平安科技（深圳）有限公司 | Client information verification method, device, computer equipment and storage medium |
| CN112990707A (en) * | 2021-03-12 | 2021-06-18 | 深圳工盟科技有限公司 | Construction risk assessment method, device, equipment and storage medium |
| CN115098508A (en) * | 2022-07-04 | 2022-09-23 | 成都秦川物联网科技股份有限公司 | Smart city check list generation method, system and device based on Internet of things |
| CN115079738A (en) * | 2022-07-28 | 2022-09-20 | 苏州浪潮智能科技有限公司 | Method, device and equipment for regulating and controlling rotating speed of server fan and storage medium |
| CN115079738B (en) * | 2022-07-28 | 2024-01-19 | 苏州浪潮智能科技有限公司 | Method, device, equipment and storage medium for regulating and controlling rotation speed of server fan |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109785117A (en) | Air control method, computer readable storage medium and server neural network based | |
| US10572885B1 (en) | Training method, apparatus for loan fraud detection model and computer device | |
| US20180260891A1 (en) | Systems and methods for generating and using optimized ensemble models | |
| Fanning et al. | Neural network detection of management fraud using published financial data | |
| CN107590734A (en) | Determine method and device, terminal device and the computer-readable storage medium of the accrediting amount | |
| CN113011973B (en) | Method and equipment for financial transaction supervision model based on intelligent contract data lake | |
| CN110192217A (en) | Information processing unit, information processing method and computer program | |
| CN112926699A (en) | Abnormal object identification method, device, equipment and storage medium | |
| Berlusconi et al. | The determinants of heroin flows in Europe: A latent space approach | |
| Bukhtiarova et al. | Assessment of financial monitoring efficiency in the banking system of Ukraine | |
| CN109255747A (en) | A kind of intelligent checks method that information is declared | |
| Voican | Credit Card Fraud Detection using Deep Learning Techniques. | |
| WO2021258118A1 (en) | Systems and methods of transaction tracking and analysis for near real-time individualized credit scoring | |
| Majeske et al. | The bank loan approval decision from multiple perspectives | |
| Derevyanko et al. | Assessment of financial and economic security of the region (based on the relevant statistics of the Donetsk region) | |
| CN110796539A (en) | Credit investigation evaluation method and device | |
| CN108629685A (en) | loan product attribute determining method and server | |
| US8600894B2 (en) | Three-stage, double blind credit rating of securities | |
| Yasaka | Data mining in anti-money laundering field | |
| Kamordzhanova et al. | The impact of the digital economy on accounting, reporting and audit | |
| Iqbal | General Takaful Practice: Technical Approach to Eliminate Gharar (uncertainty), Maisir (gambling), and Riba'(usury) | |
| Brunner et al. | Hold-up in multiple banking: Evidence from SME lending | |
| DE202022104425U1 (en) | Intelligent system for secure integration of credit checks and banking systems through machine learning | |
| Constantin et al. | A new model for estimating the risk of bankruptcy of the insurance companies based on the artificial neural networks | |
| CN114048330A (en) | Risk conduction probability knowledge graph generation method, device, equipment and storage medium |
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 |