CN110163761A - Suspect items member recognition methods and device based on image procossing - Google Patents
Suspect items member recognition methods and device based on image procossing Download PDFInfo
- Publication number
- CN110163761A CN110163761A CN201910238911.0A CN201910238911A CN110163761A CN 110163761 A CN110163761 A CN 110163761A CN 201910238911 A CN201910238911 A CN 201910238911A CN 110163761 A CN110163761 A CN 110163761A
- Authority
- CN
- China
- Prior art keywords
- image
- case
- case audit
- similarity
- cluster
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Accounting & Taxation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Finance (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Development Economics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Technology Law (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application provides suspect items member recognition methods and device based on image procossing, suspect items member's recognition methods based on image procossing, comprising: the case for obtaining the case audit image composition of program member's submission of participation project audits image set；Extract the feature vector of case audit image described in the case audit image set；The image similarity between the case audit image is determined according to described eigenvector；Case audit image is clustered based on the similarity weight that described image similarity determines；The suspect items member in the program member is determined according to the cluster result of the cluster.Suspect items member's recognition methods based on image procossing, similitude in the case audit image set that program member submits based on case audit image determines suspect items member, and then suspicious clique is drawn a circle to approve on the basis of suspect items member, suspicious clique's recognition capability of broad sense is established, thus the probability that clique commits a crime in reduction project.
Description
Technical field
This application involves technical field of image processing, in particular to a kind of suspect items member identification based on image procossing
Method.The application is related to a kind of suspect items member's identification device based on image procossing, a kind of calculating equipment, Yi Jiyi simultaneously
Kind computer readable storage medium.
Background technique
As public anti-risk consciousness enhances, insurance becomes the investment option of more and more people.Air control is that insurance is done honest work
The important link often developed, the insurance air control grown up under internet environment is even more important, and the core compensation link insured in air control is
Finger judges whether Claims Resolution meets insurance cover clause, is one of core means of prevention and control insurance fraud.Settlement of insurance claim link can generally need
User's upload Claims Resolution material is wanted, passes through now with many illegal profit tissues and recruits volunteer's formation insurance fraud clique, and insurance fraud group
The Claims Resolution storeroom that partner uploads has certain general character mostly.
In the prior art, clique's recognition methods of air control scene is the true sexual intercourse based on the fact that sexual intercourse is judged
Refer to that the fund between the program member for participating in insurance coverage is transferred accounts relationship, correspondence, address list relationship, device relationships etc.,
By judging to achieve the purpose that determining insurance fraud clique with the presence or absence of true sexual intercourse between program member.But it much deceives now
Protect between gang member using network as contact information, thus cannot by based on the fact that clique's identifying schemes of sexual intercourse identify,
Have the defects that certain.
Summary of the invention
In view of this, the embodiment of the present application provides a kind of suspect items member's recognition methods based on image procossing, with
Solve technological deficiency existing in the prior art.The embodiment of the present application provides a kind of suspect items based on image procossing simultaneously
Member's identification device, a kind of calculating equipment and a kind of computer readable storage medium.
The embodiment of the present application discloses a kind of suspect items member's recognition methods based on image procossing, comprising:
The case for obtaining the case audit image composition of program member's submission of participation project audits image set；
Extract the feature vector of case audit image described in the case audit image set；
The image similarity between the case audit image is determined according to described eigenvector；
Case audit image is clustered based on the similarity weight that described image similarity determines；
The suspect items member in the program member is determined according to the cluster result of the cluster.
Optionally, the feature vector for extracting case audit image described in the case audit image set, comprising:
By the input of case audit image, trained deep learning model carries out image vector processing, output in advance
The feature vector of the case audit image.
Optionally, the image similarity determined according to described eigenvector between the case audit image, comprising:
It is calculated between the case audit image according to described eigenvector using word frequency inverse document frequency algorithm
Image similarity.
Optionally, the image similarity determined according to described eigenvector between the case audit image, comprising:
Vector distance between feature vector based on case audit image determines between the case audit image
Image similarity.
Optionally, the similarity weight determined based on described image similarity gathers case audit image
Class, comprising:
Connected graph is constructed, the node in case audit image and the connected graph is established into one-to-one relationship,
And the case is audited into the image similarity between image and takes logarithm as between case audit image corresponding node
Side right weight；
Connected graph input Clustering Model is clustered, the cluster mark of the case audit image is exported.
Optionally, the cluster result according to the cluster determines the suspect items member in the program member, packet
It includes:
It is suspect image that determining, which has the case audit image of identical cluster mark,；
Determine the suspect items member in the artificial program member of the submission of the suspect image.
Optionally, the cluster result according to the cluster determines the suspect items member step in the program member
After execution, comprising:
Determine that the collection of the submitter of the case audit image in the suspect image with same cluster mark is combined into
Suspicious clique.
Optionally, the cluster result according to the cluster determines the suspect items member step in the program member
After execution, further includes:
The fact that obtain suspect items member sexual intercourse；
Determined in the suspect items member based on the true sexual intercourse have the project of practical true sexual intercourse at
Member is insincere program member；
Determine the member that the case audit image with same cluster mark is submitted in the insincere program member
Collection is combined into suspicious clique.
Optionally, the case for the case audit image composition that the program member for obtaining participation project submits audits image
Before collection step executes, further includes:
The fact that obtain program member sexual intercourse；
The program member with practical true sexual intercourse is determined based on the true sexual intercourse；
Correspondingly, the case for the case audit image composition that the program member for obtaining participation project submits audits image
Collect in step, the case for obtaining the program member with practical true sexual intercourse audits image, forms the case audit
Image set.
Optionally, the fact that program member's sexual intercourse, including at least one of following:
Fund between the program member is transferred accounts relationship, correspondence, address list relationship, device relationships.
Optionally, the deep learning model is constructed based on any one following neural network:
Convolutional neural networks, deep neural network.
Optionally, the Clustering Model is based on any one following algorithm:
Label propagation algorithm, largest connected nomography.
The application provides a kind of suspect items member's identification device based on image procossing, comprising:
Case audits image set and obtains module, is configured as obtaining the case audit figure that the program member of participation project submits
As the case of composition audits image set；
Characteristic vector pickup module is configured as extracting the spy of case audit image described in the case audit image set
Levy vector；
Image similarity determining module is configured as being determined according to described eigenvector between the case audit image
Image similarity；
Cluster module is configured as the similarity weight determined based on described image similarity and audits image to the case
It is clustered；
Suspect items member's determining module is configured as being determined according to the cluster result of the cluster in the program member
Suspect items member.
The application provides a kind of calculating equipment, comprising:
Memory and processor；
For the memory for storing computer executable instructions, the processor is executable for executing the computer
Instruction:
The case for obtaining the case audit image composition of program member's submission of participation project audits image set；
Extract the feature vector of case audit image described in the case audit image set；
The image similarity between the case audit image is determined according to described eigenvector；
Case audit image is clustered based on the similarity weight that described image similarity determines；
The suspect items member in the program member is determined according to the cluster result of the cluster.
The application provides a kind of computer readable storage medium, is stored with computer instruction, which is held by processor
The step of suspect items member's recognition methods based on image procossing is realized when row.
Compared with prior art, the application has the advantages that
The application provides a kind of suspect items member's recognition methods based on image procossing, comprising: obtains participation project
The case for the case audit image composition that program member submits audits image set；Extract case described in the case audit image set
The feature vector of part audit image；The image similarity between the case audit image is determined according to described eigenvector；Base
Case audit image is clustered in the similarity weight that described image similarity determines；According to the cluster of the cluster
As a result the suspect items member in the program member is determined.
Suspect items member's recognition methods provided by the present application based on image procossing, by being audited in image set to case
Case audit image carry out image similarity calculating, case audit image is carried out figure and gathered according to image similarity calculated result
Generic operation, so that the audit image of the case with similitude is labeled like-identified, so that being identified according to mark has group
The suspect items member of partner's insurance fraud possibility reduces the probability that clique's insurance fraud occurs.
Detailed description of the invention
Fig. 1 is a kind of suspect items member's recognition methods flow chart based on image procossing provided by the embodiments of the present application；
Fig. 2 is a kind of signal of suspect items member's identification process based on image procossing provided by the embodiments of the present application
Figure；
Fig. 3 is a kind of signal of suspect items member's identification device based on image procossing provided by the embodiments of the present application
Figure；
Fig. 4 is a kind of structural block diagram for calculating equipment provided by the embodiments of the present application.
Specific embodiment
Many details are explained in the following description in order to fully understand the application.But the application can be with
Much it is different from other way described herein to implement, those skilled in the art can be without prejudice to the application intension the case where
Under do similar popularization, therefore the application is not limited by following public specific implementation.
The term used in this specification one or more embodiment be only merely for for the purpose of describing particular embodiments,
It is not intended to be limiting this specification one or more embodiment.In this specification one or more embodiment and appended claims
The "an" of singular used in book, " described " and "the" are also intended to including most forms, unless context is clearly
Indicate other meanings.It is also understood that term "and/or" used in this specification one or more embodiment refers to and includes
One or more associated any or all of project listed may combine.
It will be appreciated that though may be retouched using term first, second etc. in this specification one or more embodiment
Various information are stated, but these information should not necessarily be limited by these terms.These terms are only used to for same type of information being distinguished from each other
It opens.For example, first can also be referred to as second, class in the case where not departing from this specification one or more scope of embodiments
As, second can also be referred to as first.Depending on context, word as used in this " if " can be construed to
" ... when " or " when ... " or " in response to determination ".
The application provides a kind of suspect items member's recognition methods based on image procossing, and the application also provides one kind and is based on
Suspect items member's identification device of image procossing, a kind of calculating equipment and a kind of computer readable storage medium.Divide below
Not Jie He the attached drawing of embodiment provided by the present application be described in detail one by one, and each step of method is illustrated.
A kind of suspect items member's recognition methods embodiment based on image procossing provided by the present application is as follows:
Referring to attached drawing 1, it illustrates a kind of suspect items member identification sides based on image procossing provided in this embodiment
Method flow chart；Referring to attached drawing 2, it illustrates a kind of suspect items members based on image procossing provided in this embodiment to identify
The schematic diagram of journey.
Step S102, the case for obtaining the case audit image composition of program member's submission of participation project audit image
Collection.
Settlement of insurance claim link generally may require that user uploads Claims Resolution material, picture such as in kind, by taking vehicle insurance as an example, insurance company
It may require that user submits the appearance picture of damaged vehicle, whether the Claims Resolution application to judge that user proposes meets insurance cover item
Money.The program member of the participation project is the user for participating in insurance coverage, and it is the use for participating in insurance coverage that case, which audits image,
The picture in kind in Claims Resolution material that family is submitted, such as appearance picture, the traffic accident scene diagram piece of damaged vehicle.Practical application
In, the Claims Resolution picture that insurance fraud clique uploads has similitude to a certain extent, therefore can be known by analysis Claims Resolution picture
Other insurance fraud clique, to reduce the probability that the insurance fraud of insurance fraud clique happens in settlement of insurance claim link.
The present embodiment audits image by the case audited in image set to the case and carries out image recognition, described in acquisition
Case audits the general harmonic wavelet of all case audit images in image set, which shows as similar case audit
Image belongs to same community and is gathered for same class, to audit the submission of image by the case that retrospect belongs to same community
People identifies suspect items member, further, can also identify suspect items member on the basis of, according to suspect items at
The community of member's ownership is come the occurrence of drawing a circle to approve suspicious clique, prevent insurance fraud.
For the purpose for the calculation amount for reducing image recognition, downscaled images object can be identified before carrying out image recognition
Range, in a kind of preferred embodiment provided by the embodiments of the present application, submitted in the program member for obtaining participation project
Case audit image composition case audit image set before, the mode of being accomplished in the following manner come downscaled images identification object model
It encloses:
The fact that obtain program member sexual intercourse；
The program member with practical true sexual intercourse is determined based on the true sexual intercourse；
Correspondingly, the case audit image set for the case audit image composition submitted in the program member for obtaining participation project
During, the case for obtaining the program member with practical true sexual intercourse audits image, forms the case audit
Image set.
Whether the implementation of above-mentioned offer, being primarily based on, there is practical true sexual intercourse to sieve to the program member
Choosing, the case for then obtaining the case audit image composition that there is the program member of practical true sexual intercourse to submit audit figure
Image set, and image recognition is carried out on the basis of the case of composition audits image set, so that reference image identification is reduced,
Meanwhile a possibility that identifying suspect items member, can be larger.
Wherein, the fact that program member's sexual intercourse, including at least one of following: the fund between the program member
It transfers accounts relationship, correspondence, address list relationship, device relationships.
The gang member for participating in insurance fraud clique may carry out business transaction and intercommunication by certain medium in practical application
Information has fund money transfer transactions record, by communication apparatus communication, other side address book contact, multi-party once using together each other
One logging device etc. belongs to true sexual intercourse, described that there is practical true sexual intercourse to refer to that the program member is detected really
It is real that there is at least one of above-mentioned true sexual intercourse relationship, there is the program member of true sexual intercourse to be identified its submission
Case audit image belong to identical community in the case where, judge that it can be more reasonable for suspect items member.
Step S104 extracts the feature vector of case audit image described in the case audit image set.
In practical application, the Claims Resolution picture number that settlement of insurance claim link of insurance company during core is protected obtains compares
It greatly, is to improve image processing efficiency, it is described to extract the case and examine in a kind of preferred embodiment provided by the embodiments of the present application
The feature vector of the audit image of case described in core image set, comprising: case audit image input is trained in advance
Deep learning model carries out image vector processing, exports the feature vector of the case audit image.
During calculating the similarity of case audit image, needs image carrying out vectorization processing, image is turned
The process for being melted into vector is based on deep learning model realization, and the deep learning model is to be trained to the good mould of sample training in advance
Type inputs in the deep learning model, the output of the model is the case using case audit image as input quantity
The feature vector of part audit image.
The deep learning model is preferably based on convolutional neural networks (CNN, Convolutional Neural
Network) or deep neural network (DNN, Deep Neural Network) constructs, in addition to this it is possible to using other
Neural network, the present embodiment is it is not limited here.
Step S106 determines the image similarity between the case audit image according to described eigenvector.
It is described that the case is determined according to described eigenvector in a kind of preferred embodiment provided by the embodiments of the present application
Audit image between image similarity, comprising: according to described eigenvector using word frequency inverse document frequency (TF-IDF,
Term Frequency-Inverse Document Frequency) algorithm calculates the image between case audit image
Similarity.
The word frequency inverse document frequency algorithm is a kind of statistical method, and application is chiefly used in text, to assess
One words is for the significance level of a copy of it file in a file set or a corpus, and the present embodiment is by the case
After part audit image is converted into feature vector, image similarity calculating is carried out using the word frequency inverse document frequency algorithm.
It is described that the case is determined according to described eigenvector in a kind of preferred embodiment provided by the embodiments of the present application
Audit the image similarity between image, comprising: the vector distance between the feature vector based on case audit image, really
Image similarity between the fixed case audit image.
Wherein, the vector distance between the feature vector of case audit image, including at least one of following: it is European away from
From, COS distance；The characteristics of COS distance is cosine value close to 1, and angle tends to 0, shows that two vectors are more similar, image is similar
The characteristics of degree is higher, Euclidean distance is that Euclidean distance is smaller, and two vectors are more similar, and image similarity is higher.
Step S108 gathers case audit image based on the similarity weight that described image similarity determines
Class.
It is described based on the similar of described image similarity determination in a kind of preferred embodiment provided by the embodiments of the present application
Degree weight clusters case audit image, comprising:
Connected graph is constructed, the node in case audit image and the connected graph is established into one-to-one relationship,
And the case is audited into the image similarity between image and takes logarithm as between case audit image corresponding node
Side right weight；
Connected graph input Clustering Model is clustered, the cluster mark of the case audit image is exported.
Preferably, the Clustering Model is preferably based on label propagation algorithm (LPA, Label Propagation
Algorithm) or largest connected nomography is realized, in addition to this it is possible to which the present embodiment is not done herein using other algorithms
It limits.
By taking the Clustering Model is based on the realization of label propagation algorithm as an example, the sorting procedure is illustrated:
The input of Clustering Model is the connected graph that image and image similarity building are audited based on case, is specifically being constructed
During connected graph, the node for including in connected graph and case audit image are established into one-to-one relationship, Mei Gejie
Point separately includes the picture number of respectively corresponding case audit image, also, by the above-mentioned case audit image acquired it
Between image similarity take logarithm as the case audit image corresponding node between side right weight, complete the cluster mould
The building of connected graph needed for type.
The Clustering Model is based on label propagation algorithm, the application scenarios of label propagation algorithm are as follows: community discovery, tradition meaning
Community in justice refers to thering is biggish similitude between the group node in network, so that a kind of internal connection formed is tight
It is close, and external sparse group structure, community discovery is known as to the process that given network finds its community structure, generally
It sees, the process of community discovery is exactly a kind of process of cluster；The basic thought of label propagation algorithm is: by the neighbours of a node
Label of the most label of quantity as the node itself in the label of node, specially to each node addition label to represent
Community belonging to it, and the propagation for passing through label forms the community structure of same label.
The communication process of label propagation algorithm may be summarized to be:
1) when initial, one unique label of each node is given；
2) each node updates the label of itself using label most in the label of its neighbor node；
3) step 2) is executed repeatedly, until the label of each node is no longer changed.
The connected graph is inputted the Clustering Model operation label propagation algorithm to cluster, the case audits image
It is divided community, the label of each community of representative after output division, which is the cluster for referring to the case audit image
Mark.
Step S110 determines the suspect items member in the program member according to the cluster result of the cluster.
In a kind of preferred embodiment provided by the embodiments of the present application, the cluster result according to the cluster determines institute
State the suspect items member in program member, comprising:
It is suspect image that determining, which has the case audit image of identical cluster mark,；
Determine the suspect items member in the artificial program member of the submission of the suspect image.
In the above-mentioned building connected graph the step of, the case is audited into the image similarity between image, logarithm is taken to make
The side right weight between image corresponding node is audited for the case, allows for the case that similarity is high between the case audit image
Part audits image, and the weight for connecting the side of its corresponding node in connected graph is also high, and node connection is closer, to be more likely formed
The cluster mark of same community, the case audit image of the Clustering Model output can be used as the standard for dividing community, tool
There is the case audit image of identical cluster mark to belong to identical community, so that it is determined that it is suspect image, correspondingly, described
Suspect items member in the artificial program member of the submission of suspect image.
Preferably, the cluster result according to the cluster determines the suspect items member step in the program member
After execution, comprising: determine the submitter's of the case audit image in the suspect image with same cluster mark
Collection is combined into suspicious clique.
After determining the suspect items member, according to the cluster result, the case with same cluster mark
Part audits image and forms same community, and the group of the corresponding submitter's composition of case audit image in same community then has very much can
It can be suspicious clique.
In a kind of preferred embodiment provided by the embodiments of the present application, the item is determined according to the cluster result of the cluster
It, can also be by analyzing the fact on the basis of cluster result that cluster obtains after suspect items member in mesh member
Relationship judges that suspect items member whether from same insurance fraud clique, is implemented as follows:
1) the fact that obtain suspect items member sexual intercourse；
2) project with practical true sexual intercourse is determined in the suspect items member based on the true sexual intercourse
Member is insincere program member；
3) determine in the insincere program member submit have it is same cluster mark the case audit image at
Member's collection is combined into suspicious clique.
After determining the suspect items member, sexual intercourse screening is provided based on the fact the suspect items member of acquisition
The program member for having practical fact sexual intercourse is that insincere program member is advantageous in that, figure cluster operation is accomplished that the first
One wheel screening, filters out the case audit image for being divided into same community, and the submitter which audits image has clique
The suspicion of insurance fraud, secondly by judging whether that there is practical true sexual intercourse, which to carry out the second wheel, screens, come judge suspect items at
Whether member has practical connection in real life, screens by this two-wheeled, and the program member filtered out is provided simultaneously with submission
Case audits picture has the two conditions of true sexual intercourse between the suspicious picture and the program member, on this basis really
The accuracy that the fixed insincere program member gathers the insurance fraud clique constituted is higher.
Following combination attached drawings 2, to the suspect items member recognition methods provided by the present application based on image procossing carry out into
One step explanation, is implemented as follows:
Step S202 obtains the N Claims Resolution pictures that settlement of insurance claim link requires user to upload.
Settlement of insurance claim link generally may require that user uploads Claims Resolution material, picture of such as settling a claim, and insurance company can be to Claims Resolution material
Material is analyzed to judge whether user meets Claims Resolution condition.The present embodiment is to identify suspect items based on image processing techniques
Member, it is necessary first to obtain N Claims Resolution pictures of user's upload, N Claims Resolution pictures respectively indicate are as follows: Image1, Image2,
Image3 ... ..., ImageN.
Step S204 utilizes the deep learning model extraction N Claims Resolution respective feature vectors of picture.
Based on Claims Resolution picture number, big, occupancy calculates the big problem in space, and N Claims Resolution pictures are used uniformly model to locate
Reason, to improve image processing efficiency.This method is to carry out figure cluster operation according to the image similarity between Claims Resolution picture,
During calculating image similarity, needs that image is first carried out vectorization processing, the process that image is converted to vector is based on
The deep learning frame of open source realizes, specifically, under deep learning frame, constructed in advance using convolutional neural networks and
Deep learning model is trained, is obtained by the way that this N of Image1 to ImageN Claims Resolution pictures are inputted deep learning model
The respective feature vector of Image1 to ImageN.
Step S206 calculates the image similarity between N Claims Resolution pictures using word frequency inverse document frequency algorithm.
Vectorization is carried out to N Claims Resolution pictures to operate after obtaining its corresponding feature vector, that is, word frequency can be used against text
Frequency index algorithm calculates the image similarity between Claims Resolution picture, the foundation as figure cluster operation.
Step S208 clusters N Claims Resolution pictures using label propagation algorithm.
Specifically, being carried out to the cluster operation of Claims Resolution picture by Clustering Model, which adopts in cluster process
Algorithm is label propagation algorithm, and steps are as follows:
1) connected graph is constructed, the node in N Claims Resolution picture Image1 to ImageN and connected graph corresponds, accordingly
, node can be marked as Image1, Image2, Image3 ... ..., ImageN, by the image phase between N Claims Resolution pictures
Take logarithm as the side right weight between Claims Resolution picture corresponding node like degree；
2) connected graph is inputted into Clustering Model, the label of node is obtained using label propagation algorithm, there is higher similarity
The corresponding node of Claims Resolution picture label having the same, it is similar for being gathered.
Step S210 determines suspect items member.
By the mark of the Claims Resolution picture of Clustering Model output as the standard for dividing community, the reason with same cluster mark
Pay for picture has similarity in one aspect, is divided into same community, and it is suspicious that determining, which has the Claims Resolution picture of community attributes,
Picture determines the artificial suspect items member of the submission of suspicious picture.
The sexual intercourse of the fact that step S212, acquisition suspect items member.
After determining suspect items member, the fact that can also be by between suspect items member sexual intercourse further judges that this can
Whether doubtful program member has insurance fraud suspicion, true sexual intercourse include fund between program member transfer accounts relationship, correspondence,
Address list relationship, device relationships etc. have fund money transfer transactions record, are communicated by communication apparatus, address list joins other side each other
It is people, is once all identified as that there is true sexual intercourse using the suspect items member of the behaviors such as same logging device in many ways.
Step S214, it is insincere project that determining in suspect items member, which has the program member of practical true sexual intercourse,
Member.
After having filtered out the suspect items member that the Claims Resolution picture submitted has similitude to the cluster operation of Claims Resolution picture,
If a possibility that suspect items member goes back while having true sexual intercourse, and suspect items member participates in insurance fraud is bigger, according to
This can be based on the fact that sexual intercourse filters out the program member with true sexual intercourse as insincere in suspect items member
Program member.
Step S216 determines that the member set for the Claims Resolution image with same label submitted in insincere program member is
Suspicious clique.
Determined insincere program member has very big insurance fraud suspicion in program member, and belongs to same insurance fraud clique
Program member submit Claims Resolution picture mostly in one aspect have similitude, same label is easily identified, accordingly, according to it
Before obtained cluster result, can determine the member's group for the Claims Resolution image with same label submitted in insincere program member
At collection be combined into suspicious clique.
A kind of suspect items member's identification device embodiment based on image procossing provided by the present application is as follows:
In the above-described embodiment, a kind of suspect items member's recognition methods based on image procossing is provided, therewith phase
Corresponding, present invention also provides a kind of suspect items member's identification device based on image procossing carries out with reference to the accompanying drawing
Explanation.
Referring to attached drawing 3, it illustrates a kind of suspect items member's identification devices based on image procossing provided by the present application
The schematic diagram of embodiment.
Since Installation practice is substantially similar to embodiment of the method, so describing fairly simple, relevant part please join
The corresponding explanation of the embodiment of the method for above-mentioned offer is provided.Installation practice described below is only schematical.
The application provides a kind of suspect items member's identification device based on image procossing, comprising:
Case audits image set and obtains module 302, and the case for being configured as obtaining program member's submission of participation project is examined
The case of core image composition audits image set；
Characteristic vector pickup module 304 is configured as extracting case audit image described in the case audit image set
Feature vector；
Image similarity determining module 306, be configured as being determined according to described eigenvector the case audit image it
Between image similarity；
Cluster module 308 is configured as auditing the case based on the similarity weight that described image similarity determines
Image is clustered；
Suspect items member determining module 310, be configured as being determined according to the cluster result of the cluster project at
Suspect items member in member.
Optionally, described eigenvector extraction module 304 is specifically configured to the case auditing image input in advance
Trained deep learning model carries out image vector processing, exports the feature vector of the case audit image.
Optionally, described image similarity determining module 306 is specifically configured to utilize word frequency according to described eigenvector
Inverse document frequency algorithm calculates the image similarity between the case audit image.
Optionally, described image similarity determining module 306 is specifically configured to audit the spy of image based on the case
The vector distance between vector is levied, determines the image similarity between the case audit image.
Optionally, the cluster module 308, comprising:
Connected graph constructs submodule, is configured as building connected graph, will be in case audit image and the connected graph
Node establish one-to-one relationship, and the case is audited into the image similarity between image and takes logarithm as described in
Case audits the side right weight between image corresponding node；
Cluster mark output module is configured as clustering connected graph input Clustering Model, exports the case
Part audits the cluster mark of image.
Optionally, the suspect items member determining module 310, comprising:
Suspect image determines submodule, be configured to determine that with it is identical cluster mark the case audit image be can
Doubt image；
Suspect items member determines submodule, is configured to determine that the submission of the suspect image artificially program member
In suspect items member.
Optionally, suspect items member's identification device based on image procossing, further includes:
First suspicious clique's determining module is configured to determine that in the suspect image with the described of same cluster mark
The collection of the submitter of case audit image is combined into suspicious clique.
Optionally, suspect items member's identification device based on image procossing, further includes:
Second true sexual intercourse obtains module, is configured as the fact that obtain suspect items member sexual intercourse；
Insincere program member's determining module is configured as based on the true sexual intercourse in the suspect items member
It is insincere program member that determining, which has the program member of practical true sexual intercourse,；
Second suspicious clique's determining module is configured to determine that submit have same cluster in the insincere program member
The member set of the case audit image of mark is suspicious clique.
Optionally, suspect items member's identification device based on image procossing, further includes:
First true sexual intercourse obtains module, is configured as the fact that obtain program member sexual intercourse；
Case audits image and obtains object determining module, is configured as determining there is practical thing based on the true sexual intercourse
The program member of real sexual intercourse；
Correspondingly, the case audit image set obtains module 302, it is specifically configured to obtain described with practical true
The case of the program member of sexual intercourse audits image, forms the case audit image set.
Optionally, the fact that program member's sexual intercourse, including at least one of following:
Fund between the program member is transferred accounts relationship, correspondence, address list relationship, device relationships.
Optionally, the deep learning model is constructed based on any one following neural network:
Convolutional neural networks, deep neural network.
Optionally, the Clustering Model is based on any one following algorithm:
Label propagation algorithm, largest connected nomography.
A kind of calculating apparatus embodiments provided by the present application are as follows:
Fig. 4 is to show the structural block diagram of the calculating equipment 400 according to one embodiment of this specification.The calculating equipment 400
Component include but is not limited to memory 410 and processor 420.Processor 420 is connected with memory 410 by bus 430,
Database 450 is for saving data.
Calculating equipment 400 further includes access device 440, access device 440 enable calculate equipment 400 via one or
Multiple networks 460 communicate.The example of these networks includes public switched telephone network (PSTN), local area network (LAN), wide area network
(WAN), the combination of the communication network of personal area network (PAN) or such as internet.Access device 440 may include wired or wireless
One or more of any kind of network interface (for example, network interface card (NIC)), such as IEEE802.11 wireless local area
Net (WLAN) wireless interface, worldwide interoperability for microwave accesses (Wi-MAX) interface, Ethernet interface, universal serial bus (USB) connect
Mouth, cellular network interface, blue tooth interface, near-field communication (NFC) interface, etc..
In one embodiment of this specification, other unshowned portions in the above-mentioned component and Fig. 4 of equipment 400 are calculated
Part can also be connected to each other, such as pass through bus.It should be appreciated that calculating device structure block diagram shown in Fig. 4 merely for the sake of
Exemplary purpose, rather than the limitation to this specification range.Those skilled in the art can according to need, and increases or replaces it
His component.
Calculating equipment 400 can be any kind of static or mobile computing device, including mobile computer or mobile meter
Calculate equipment (for example, tablet computer, personal digital assistant, laptop computer, notebook computer, net book etc.), movement
Phone (for example, smart phone), wearable calculating equipment (for example, smartwatch, intelligent glasses etc.) or other kinds of shifting
Dynamic equipment, or the static calculating equipment of such as desktop computer or PC.Calculating equipment 400 can also be mobile or state type
Server.
The application provides a kind of calculating equipment, including memory 410, processor 420 and storage are on a memory and can be
The computer instruction run on processor, the processor 420 is for executing following computer executable instructions:
The case for obtaining the case audit image composition of program member's submission of participation project audits image set；
Extract the feature vector of case audit image described in the case audit image set；
The image similarity between the case audit image is determined according to described eigenvector；
Case audit image is clustered based on the similarity weight that described image similarity determines；
The suspect items member in the program member is determined according to the cluster result of the cluster.
Optionally, the feature vector for extracting case audit image described in the case audit image set, comprising:
By the input of case audit image, trained deep learning model carries out image vector processing, output in advance
The feature vector of the case audit image.
Optionally, the image similarity determined according to described eigenvector between the case audit image, comprising:
It is calculated between the case audit image according to described eigenvector using word frequency inverse document frequency algorithm
Image similarity.
Optionally, the image similarity determined according to described eigenvector between the case audit image, comprising:
Vector distance between feature vector based on case audit image determines between the case audit image
Image similarity.
Optionally, the similarity weight determined based on described image similarity gathers case audit image
Class, comprising:
Connected graph is constructed, the node in case audit image and the connected graph is established into one-to-one relationship,
And the case is audited into the image similarity between image and takes logarithm as between case audit image corresponding node
Side right weight；
Connected graph input Clustering Model is clustered, the cluster mark of the case audit image is exported.
Optionally, the cluster result according to the cluster determines the suspect items member in the program member, packet
It includes:
It is suspect image that determining, which has the case audit image of identical cluster mark,；
Determine the suspect items member in the artificial program member of the submission of the suspect image.
Optionally, the cluster result according to the cluster determines the suspect items member step in the program member
After execution, comprising:
Determine that the collection of the submitter of the case audit image in the suspect image with same cluster mark is combined into
Suspicious clique.
Optionally, the cluster result according to the cluster determines the suspect items member step in the program member
After execution, further includes:
The fact that obtain suspect items member sexual intercourse；
Determined in the suspect items member based on the true sexual intercourse have the project of practical true sexual intercourse at
Member is insincere program member；
Determine the member that the case audit image with same cluster mark is submitted in the insincere program member
Collection is combined into suspicious clique.
Optionally, the case for the case audit image composition that the program member for obtaining participation project submits audits image
Before collection step executes, further includes:
The fact that obtain program member sexual intercourse；
The program member with practical true sexual intercourse is determined based on the true sexual intercourse；
Correspondingly, the case for the case audit image composition that the program member for obtaining participation project submits audits image
Collect in step, the case for obtaining the program member with practical true sexual intercourse audits image, forms the case audit
Image set.
Optionally, the fact that program member's sexual intercourse, including at least one of following:
Fund between the program member is transferred accounts relationship, correspondence, address list relationship, device relationships.
Optionally, the deep learning model is constructed based on any one following neural network:
Convolutional neural networks, deep neural network.
One embodiment of the application also provides a kind of computer readable storage medium, is stored with computer instruction, the instruction
It is accomplished by when being executed by processor
The case for obtaining the case audit image composition of program member's submission of participation project audits image set；
Extract the feature vector of case audit image described in the case audit image set；
The image similarity between the case audit image is determined according to described eigenvector；
Case audit image is clustered based on the similarity weight that described image similarity determines；
The suspect items member in the program member is determined according to the cluster result of the cluster.
Optionally, the feature vector for extracting case audit image described in the case audit image set, comprising:
By the input of case audit image, trained deep learning model carries out image vector processing, output in advance
The feature vector of the case audit image.
Optionally, the image similarity determined according to described eigenvector between the case audit image, comprising:
It is calculated between the case audit image according to described eigenvector using word frequency inverse document frequency algorithm
Image similarity.
Optionally, the image similarity determined according to described eigenvector between the case audit image, comprising:
Vector distance between feature vector based on case audit image determines between the case audit image
Image similarity.
Optionally, the similarity weight determined based on described image similarity gathers case audit image
Class, comprising:
Connected graph is constructed, the node in case audit image and the connected graph is established into one-to-one relationship,
And the case is audited into the image similarity between image and takes logarithm as between case audit image corresponding node
Side right weight；
Connected graph input Clustering Model is clustered, the cluster mark of the case audit image is exported.
Optionally, the cluster result according to the cluster determines the suspect items member in the program member, packet
It includes:
It is suspect image that determining, which has the case audit image of identical cluster mark,；
Determine the suspect items member in the artificial program member of the submission of the suspect image.
Optionally, the cluster result according to the cluster determines the suspect items member step in the program member
After execution, comprising:
Determine that the collection of the submitter of the case audit image in the suspect image with same cluster mark is combined into
Suspicious clique.
Optionally, the cluster result according to the cluster determines the suspect items member step in the program member
After execution, further includes:
The fact that obtain suspect items member sexual intercourse；
Determined in the suspect items member based on the true sexual intercourse have the project of practical true sexual intercourse at
Member is insincere program member；
Determine the member that the case audit image with same cluster mark is submitted in the insincere program member
Collection is combined into suspicious clique.
Optionally, the case for the case audit image composition that the program member for obtaining participation project submits audits image
Before collection step executes, further includes:
The fact that obtain program member sexual intercourse；
The program member with practical true sexual intercourse is determined based on the true sexual intercourse；
Correspondingly, the case for the case audit image composition that the program member for obtaining participation project submits audits image
Collect in step, the case for obtaining the program member with practical true sexual intercourse audits image, forms the case audit
Image set.
Optionally, the fact that program member's sexual intercourse, including at least one of following:
Fund between the program member is transferred accounts relationship, correspondence, address list relationship, device relationships.
Optionally, the deep learning model is constructed based on any one following neural network:
Convolutional neural networks, deep neural network.
A kind of exemplary scheme of above-mentioned computer readable storage medium for the present embodiment.It should be noted that this is deposited
The technical solution of storage media and the technical solution of above-mentioned suspect items member's recognition methods based on image procossing belong to same
Design, the detail content that the technical solution of storage medium is not described in detail may refer to above-mentioned based on the suspicious of image procossing
The description of the technical solution of program member's recognition methods.
The computer instruction includes computer program code, the computer program code can for source code form,
Object identification code form, executable file or certain intermediate forms etc..The computer-readable medium may include: that can carry institute
State any entity or device, recording medium, USB flash disk, mobile hard disk, magnetic disk, CD, the computer storage of computer program code
Device, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory),
Electric carrier signal, telecommunication signal and software distribution medium etc..It should be noted that the computer-readable medium include it is interior
Increase and decrease appropriate can be carried out according to the requirement made laws in jurisdiction with patent practice by holding, such as in certain jurisdictions of courts
Area does not include electric carrier signal and telecommunication signal according to legislation and patent practice, computer-readable medium.
It should be noted that for the various method embodiments described above, describing for simplicity, therefore, it is stated as a series of
Combination of actions, but those skilled in the art should understand that, the application is not limited by the described action sequence because
According to the application, certain steps can use other sequences or carry out simultaneously.Secondly, those skilled in the art should also know
It knows, the embodiments described in the specification are all preferred embodiments, and related actions and modules might not all be this Shen
It please be necessary.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, it may refer to the associated description of other embodiments.
The application preferred embodiment disclosed above is only intended to help to illustrate the application.There is no detailed for alternative embodiment
All details are described, are not limited the invention to the specific embodiments described.Obviously, according to the content of this specification,
It can make many modifications and variations.These embodiments are chosen and specifically described to this specification, is in order to preferably explain the application
Principle and practical application, so that skilled artisan be enable to better understand and utilize the application.The application is only
It is limited by claims and its full scope and equivalent.
Claims (15)
1. a kind of suspect items member's recognition methods based on image procossing characterized by comprising
The case for obtaining the case audit image composition of program member's submission of participation project audits image set；
Extract the feature vector of case audit image described in the case audit image set；
The image similarity between the case audit image is determined according to described eigenvector；
Case audit image is clustered based on the similarity weight that described image similarity determines；
The suspect items member in the program member is determined according to the cluster result of the cluster.
2. image processing method according to claim 1, which is characterized in that described to extract in the case audit image set
The feature vector of the case audit image, comprising:
By the input of case audit image, trained deep learning model carries out image vector processing in advance, described in output
The feature vector of case audit image.
3. image processing method according to claim 1, which is characterized in that described according to described eigenvector determination
Case audits the image similarity between image, comprising:
Word frequency inverse document frequency algorithm is utilized to calculate the image between the case audit image according to described eigenvector
Similarity.
4. image processing method according to claim 1, which is characterized in that described according to described eigenvector determination
Case audits the image similarity between image, comprising:
Vector distance between feature vector based on case audit image determines the figure between the case audit image
As similarity.
5. image processing method according to claim 1, which is characterized in that it is described based on described image similarity determine
Similarity weight clusters case audit image, comprising:
Connected graph is constructed, the node in case audit image and the connected graph is established into one-to-one relationship, and will
Image similarity between the case audit image takes logarithm as the side between case audit image corresponding node
Weight；
Connected graph input Clustering Model is clustered, the cluster mark of the case audit image is exported.
6. image processing method according to claim 5, which is characterized in that the cluster result according to the cluster is true
Suspect items member in the fixed program member, comprising:
It is suspect image that determining, which has the case audit image of identical cluster mark,；
Determine the suspect items member in the artificial program member of the submission of the suspect image.
7. image processing method according to claim 6, which is characterized in that the cluster result according to the cluster is true
After suspect items member step in the fixed program member executes, comprising:
It is suspicious to determine that the collection of the submitter of the case audit image in the suspect image with same cluster mark is combined into
Clique.
8. image processing method according to claim 1, which is characterized in that the cluster result according to the cluster is true
After suspect items member step in the fixed program member executes, further includes:
The fact that obtain suspect items member sexual intercourse；
Determine that the program member with practical true sexual intercourse is in the suspect items member based on the true sexual intercourse
Insincere program member；
Determine the member set that the case audit image with same cluster mark is submitted in the insincere program member
For suspicious clique.
9. image processing method according to claim 1, which is characterized in that the program member for obtaining participation project mentions
Before the case audit image set step of the case audit image composition of friendship executes, further includes:
The fact that obtain program member sexual intercourse；
The program member with practical true sexual intercourse is determined based on the true sexual intercourse；
Correspondingly, the case audit image set step for the case audit image composition that the program member for obtaining participation project submits
In rapid, the case for obtaining the program member with practical true sexual intercourse audits image, forms the case audit image
Collection.
10. image processing method according to claim 8 or claim 9, which is characterized in that program member the fact property is closed
System, including at least one of following:
Fund between the program member is transferred accounts relationship, correspondence, address list relationship, device relationships.
11. image processing method according to claim 2, which is characterized in that the deep learning model is based on following
A kind of neural network of anticipating building:
Convolutional neural networks, deep neural network.
12. image processing method according to claim 5, which is characterized in that the Clustering Model, based on following any one
Kind algorithm:
Label propagation algorithm, largest connected nomography.
13. a kind of suspect items member's identification device based on image procossing characterized by comprising
Case audits image set and obtains module, is configured as obtaining the case audit image group that the program member of participation project submits
At case audit image set；
Characteristic vector pickup module, be configured as extracting the feature of case audit image described in the case audit image set to
Amount；
Image similarity determining module is configured as determining the image between the case audit image according to described eigenvector
Similarity；
Cluster module is configured as the similarity weight determined based on described image similarity and carried out to case audit image
Cluster；
Suspect items member's determining module, be configured as being determined according to the cluster result of the cluster in the program member can
Doubt program member.
14. a kind of calculating equipment characterized by comprising
Memory and processor；
The memory is for storing computer executable instructions, and for executing, the computer is executable to be referred to the processor
It enables:
The case for obtaining the case audit image composition of program member's submission of participation project audits image set；
Extract the feature vector of case audit image described in the case audit image set；
The image similarity between the case audit image is determined according to described eigenvector；
Case audit image is clustered based on the similarity weight that described image similarity determines；
The suspect items member in the program member is determined according to the cluster result of the cluster.
15. a kind of computer readable storage medium, is stored with computer instruction, which is characterized in that the instruction is held by processor
The step of claim 1 to 12 any one the method is realized when row.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910238911.0A CN110163761B (en) | 2019-03-27 | 2019-03-27 | Suspicious item member identification method and device based on image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910238911.0A CN110163761B (en) | 2019-03-27 | 2019-03-27 | Suspicious item member identification method and device based on image processing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110163761A true CN110163761A (en) | 2019-08-23 |
| CN110163761B CN110163761B (en) | 2023-05-12 |
Family
ID=67638486
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910238911.0A Active CN110163761B (en) | 2019-03-27 | 2019-03-27 | Suspicious item member identification method and device based on image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110163761B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111325265A (en) * | 2020-02-17 | 2020-06-23 | 中国银联股份有限公司 | Detection method and device for tampered image |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130101221A1 (en) * | 2011-10-25 | 2013-04-25 | International Business Machines Corporation | Anomaly detection in images and videos |
| CN107358596A (en) * | 2017-04-11 | 2017-11-17 | 阿里巴巴集团控股有限公司 | A kind of car damage identification method based on image, device, electronic equipment and system |
| US20180240194A1 (en) * | 2017-02-23 | 2018-08-23 | International Business Machines Corporation | Visual analytics based vehicle insurance anti-fraud detection |
| CN108573286A (en) * | 2018-05-10 | 2018-09-25 | 阿里巴巴集团控股有限公司 | A kind of data processing method, device, equipment and the server of Claims Resolution business |
| US20180293806A1 (en) * | 2017-04-11 | 2018-10-11 | Alibaba Group Holding Limited | Picture-based vehicle loss assessment |
| CN108734479A (en) * | 2018-04-12 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
| CN109271418A (en) * | 2018-08-14 | 2019-01-25 | 阿里巴巴集团控股有限公司 | Suspicious clique's recognition methods, device, equipment and computer readable storage medium |
-
2019
- 2019-03-27 CN CN201910238911.0A patent/CN110163761B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130101221A1 (en) * | 2011-10-25 | 2013-04-25 | International Business Machines Corporation | Anomaly detection in images and videos |
| US20180240194A1 (en) * | 2017-02-23 | 2018-08-23 | International Business Machines Corporation | Visual analytics based vehicle insurance anti-fraud detection |
| CN107358596A (en) * | 2017-04-11 | 2017-11-17 | 阿里巴巴集团控股有限公司 | A kind of car damage identification method based on image, device, electronic equipment and system |
| US20180293806A1 (en) * | 2017-04-11 | 2018-10-11 | Alibaba Group Holding Limited | Picture-based vehicle loss assessment |
| CN108734479A (en) * | 2018-04-12 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
| CN108573286A (en) * | 2018-05-10 | 2018-09-25 | 阿里巴巴集团控股有限公司 | A kind of data processing method, device, equipment and the server of Claims Resolution business |
| CN109271418A (en) * | 2018-08-14 | 2019-01-25 | 阿里巴巴集团控股有限公司 | Suspicious clique's recognition methods, device, equipment and computer readable storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111325265A (en) * | 2020-02-17 | 2020-06-23 | 中国银联股份有限公司 | Detection method and device for tampered image |
| CN111325265B (en) * | 2020-02-17 | 2023-09-01 | 中国银联股份有限公司 | Detection method and device for tampered image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110163761B (en) | 2023-05-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109918454B (en) | Method and device for embedding nodes into relational network graph | |
| US20220020026A1 (en) | Anti-money laundering methods and systems for predicting suspicious transactions using artifical intelligence | |
| Klibisz et al. | Fast, simple calcium imaging segmentation with fully convolutional networks | |
| KR102251302B1 (en) | Collaborative filtering method, device, server and storage media combined with time factor | |
| CN109873812A (en) | Method for detecting abnormality, device and computer equipment | |
| Rajagopal | Glioma brain tumor detection and segmentation using weighting random forest classifier with optimized ant colony features | |
| US11263644B2 (en) | Systems and methods for detecting unauthorized or suspicious financial activity | |
| WO2021169454A1 (en) | Graph feature processing method and device | |
| TW202009741A (en) | Feature relationship recommendation method and apparatus, computing device and storage medium | |
| WO2023179099A1 (en) | Image detection method and apparatus, and device and readable storage medium | |
| CN110163250A (en) | Image desensitization process system, method and device based on distributed scheduling | |
| CN112132279A (en) | Convolutional neural network model compression method, device, equipment and storage medium | |
| CN112347372A (en) | Method for service promotion of financial enterprise based on user portrait scheme | |
| WO2023020214A1 (en) | Retrieval model training method and apparatus, retrieval method and apparatus, device and medium | |
| CN112862585A (en) | Personal loan type bad asset risk rating method based on LightGBM decision tree algorithm | |
| CN113282623A (en) | Data processing method and device | |
| CN110717758B (en) | Abnormal transaction identification method and device | |
| CN107194815B (en) | Client segmentation method and system | |
| CN110163761A (en) | Suspect items member recognition methods and device based on image procossing | |
| CN112907308B (en) | Data detection method and device, and computer readable storage medium | |
| CN115640401A (en) | Text content extraction method and device | |
| CN113487389A (en) | Information recommendation method and device | |
| CN114331665A (en) | Training method and device for credit judgment model of predetermined applicant and electronic equipment | |
| CN111552846B (en) | Method and device for identifying suspicious relationships | |
| CN115309985A (en) | Fairness evaluation method and AI model selection method of recommendation algorithm |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200929 Address after: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Applicant before: Alibaba Group Holding Ltd. Effective date of registration: 20200929 Address after: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant after: Innovative advanced technology Co.,Ltd. Address before: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant before: Advanced innovation technology Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230404 Address after: 801-10, Section B, 8th floor, 556 Xixi Road, Xihu District, Hangzhou City, Zhejiang Province Applicant after: Ant financial (Hangzhou) Network Technology Co.,Ltd. Address before: 27 Hospital Road, George Town, Grand Cayman ky1-9008 Applicant before: Innovative advanced technology Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |