CN110009359A - Training method, update method and the device of unsupervised risk prevention system model - Google Patents
Training method, update method and the device of unsupervised risk prevention system model Download PDFInfo
- Publication number
- CN110009359A CN110009359A CN201910058892.3A CN201910058892A CN110009359A CN 110009359 A CN110009359 A CN 110009359A CN 201910058892 A CN201910058892 A CN 201910058892A CN 110009359 A CN110009359 A CN 110009359A
- Authority
- CN
- China
- Prior art keywords
- risk prevention
- target service
- control model
- model
- target
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06Q20/00—Payment architectures, schemes or protocols
- G06Q20/38—Payment protocols; Details thereof
- G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
- G06Q20/401—Transaction verification
- G06Q20/4016—Transaction verification involving fraud or risk level assessment in transaction processing
Abstract
The embodiment of the present application discloses training method, update method and the device of a kind of unsupervised risk prevention system model, this method comprises: the business datum based on prefixed time interval to on-line system in target service samples；Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations include at least the credible filter operation of data；When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target service data are had neither part nor lot in based on predetermined quantity, current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target risk prevention and control model, wherein, the current risk prevention and control model is isolated forest model；Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control model.
Description
Technical field
This application involves computer software technical field more particularly to a kind of training sides of unsupervised risk prevention system model
Method, update method and device.
Background technique
For La Xin and promote any active ues culture user viscosity, often a large amount of fund is used to seek transaction platform
Pin.The activity being related to is varied, including gives bonus, bonus, discount coupon etc..These funds can attract some black production users
Malice obtains the marketing money of platform.For this kind of risk, the source of risk is usually accountee operation, rather than by malice
It usurps, belongs to initiative risk.
Such risk is there are three feature: one, it is related to that fund is big, and a large amount of marketing money is put into a short time, once it is anti-
It controls bad, huge loss will be caused to platform.Two, without the black sample reported back, air control team is difficult to observe current wind
Dangerous water level, and targetedly prevention and control, existing air control means can only it is subsequent by rule part transaction is carried out it is qualitative, to sentence
It is disconnected whether to extract marketing money.Three, gimmick variation is fast, and entire risk resisting is the long-term moment all in variation.
In current air control system, risk prevention system model is all off-line training, most fastly can T+1 update iteration be deployed to
On line, for risk resisting, response speed is much insufficient.
Summary of the invention
The purpose of the embodiment of the present application is to provide training method, update method and the dress of a kind of unsupervised risk prevention system model
It sets, can be improved the perception velocities to business platform Risk Variation, the renewal speed of risk prevention system model is improved, to improve wind
The risk supervision efficiency of dangerous prevention and control model.
In order to solve the above technical problems, the embodiment of the present application is achieved in that
In a first aspect, a kind of update method of unsupervised risk prevention system model is proposed, this method comprises:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control
Model.
Second aspect, proposes a kind of updating device of unsupervised risk prevention system model, which includes:
Sampling module, the business datum based on prefixed time interval to on-line system in target service sample；
Resampling module, the business datum based on sampling carry out resampling to obtain target service data, the resampling
Including at least the credible filter operation of data；
Training module, when having neither part nor lot in trained target service data and reaching predetermined quantity, not joining based on predetermined quantity
With trained target service data, the current risk prevention and control model to the on-line system in the target service carries out increment instruction
Practice, to obtain target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model；
Update module, the current risk prevention and control model by the on-line system in the target service replace with the target
Risk prevention system model.
The third aspect proposes a kind of electronic equipment, which includes:
Processor；And
It is arranged to the memory of storage computer executable instructions, the executable instruction makes the place when executed
It manages device and executes following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control
Model.
Fourth aspect proposes a kind of computer readable storage medium, the computer-readable recording medium storage one
Or multiple programs, one or more of programs are when the electronic equipment for being included multiple application programs executes, so that the electricity
Sub- equipment executes following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control
Model.
5th aspect, proposes a kind of training method of unsupervised risk prevention system model, this method comprises:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
6th aspect, proposes a kind of updating device of unsupervised risk prevention system model, which includes:
Sampling module, the business datum based on prefixed time interval to on-line system in target service sample；
Resampling module, the business datum based on sampling carry out resampling to obtain target service data, the resampling
Including at least the credible filter operation of data；
Training module, when having neither part nor lot in trained target service data and reaching predetermined quantity, not joining based on predetermined quantity
With trained target service data, the current risk prevention and control model to the on-line system in the target service carries out increment instruction
Practice, to obtain target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
7th aspect, proposes a kind of electronic equipment, which includes:
Processor；And
It is arranged to the memory of storage computer executable instructions, the executable instruction makes the place when executed
It manages device and executes following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
Eighth aspect proposes a kind of computer readable storage medium, the computer-readable recording medium storage one
Or multiple programs, one or more of programs are when the electronic equipment for being included multiple application programs executes, so that the electricity
Sub- equipment executes following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
As can be seen from the technical scheme provided by the above embodiments of the present application, the embodiment of the present application scheme at least has following one kind
Technical effect:
In the embodiment of the present application, sampling and resampling are carried out by the target service data to on-line system, based on adopting again
Target service data after sample carry out incremental training to the current risk prevention and control model of on-line system, and then update on-line system
Current risk prevention and control model improves risk prevention system model so as to improve the perception velocities to business platform Risk Variation
Renewal speed, to improve the risk supervision efficiency of risk prevention system model.
In addition, the scheme of the embodiment of the present application, since the source for training the iterative data of risk prevention system model is wind
The scene of dangerous prevention and control model application, the data of the two are naturally consistent, thus the scheme of the embodiment of the present application also can avoid instruction
Practice the problem inconsistent with the data of application, improves the accuracy rate of model training.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The some embodiments recorded in application, for those of ordinary skill in the art, in the premise of not making the creative labor property
Under, it is also possible to obtain other drawings based on these drawings.
Fig. 1 is the update method flow chart of the unsupervised risk prevention system model of one embodiment of the application.
Fig. 2 is the training method flow chart of the unsupervised risk prevention system model of one embodiment of the application.
Fig. 3 is the structural schematic diagram of one embodiment electronic equipment of the application.
Fig. 4 is the structural schematic diagram of the updating device of the unsupervised risk prevention system model of one embodiment of the application.
Fig. 5 is the structural schematic diagram of one embodiment electronic equipment of the application.
Fig. 6 is the structural schematic diagram of the training device of the unsupervised risk prevention system model of one embodiment of the application.
Specific embodiment
The embodiment of the present application provides training method, update method and the device of a kind of unsupervised risk prevention system model.
In order to make those skilled in the art better understand the technical solutions in the application, below in conjunction with the application reality
The attached drawing in example is applied, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described implementation
Example is merely a part but not all of the embodiments of the present application.Based on the embodiment in the application, this field is common
The application protection all should belong in technical staff's every other embodiment obtained without creative efforts
Range.
Fig. 1 is the update method flow chart of the unsupervised risk prevention system model of one embodiment of the application.The method of Fig. 1 can
Include:
S110, the business datum based on prefixed time interval to on-line system in target service sample.
In the embodiment of the present application, can based on prefixed time interval in formal on-line system to the business number of target service
According to sampling operation is carried out, which is configurable, for example, can be set to 10 minutes, half an hour or 1 hour,
Either other time span, the embodiment of the present application to this with no restriction.
In sampling process, the candidate feature of business datum can be sampled.It might as well assume the candidate feature number of acquisition
For m, the sample number of acquisition is n, then can acquire the data matrix for generating a n*m.Certainly, data can also be used in the data of acquisition
Representation other than matrix, the embodiment of the present application to this with no restriction.
S120, the business datum based on sampling carry out re-sampling operations to obtain target service data, the resampling behaviour
Make to include at least the credible filter operation of data.
It should be understood that due in the data of sampling both include white sample data, also include black sample data, and abnormality detection this
One task, which belongs to especially unbalanced black few white more task, can greatly reduce trained sample by filtering believable white sample
This quantity, to can at least save the training cost more than 50%.In addition, by filtering believable sample, it can also be slight
The performance of lift scheme.
It should be understood, of course, that may also include other filter operations when carrying out re-sampling operations.For example, may filter that described adopt
The filtering of data of the characteristic dimension other than preset range in the business datum of sample, i.e. bad data.Due to real-time source sample without
Evitable occasional is influenced by dirty data, is in the data for monitoring each characteristic dimension using minute/hour rank
When the especially abnormal fluctuation of no presence, need for there are the data of special unusual fluctuations to be rejected, to the data of fuctuation within a narrow range
Retained, with the effect of lift scheme training.Optionally, the definition of fluctuation is characterized average value more than the moment this feature
3 σ confidence intervals.Wherein, Gaussian Profile of the sample data in this feature in some predetermined amount of time before σ is current time
Standard deviation, for example, the Gaussian Profile standard deviation, etc. of 7 days this feature data in the past.It should be understood, of course, that bad data filter operation
Generally before the credible filter operation of data.
For convenient for recording, in the embodiment of the present application, the target service data acquisition system that might as well will complete re-sampling operations is recorded
For D.
S130 has neither part nor lot in instruction based on predetermined quantity when having neither part nor lot in trained target service data and reaching predetermined quantity
Experienced target service data, the current risk prevention and control model to the on-line system in the target service carry out incremental training,
To obtain target risk prevention and control model.
Wherein, the current risk prevention and control model is isolated forest model.
In the embodiment of the present application, to the risk prevention system model progress as isolated forest model by the way of incremental training
Training.Target risk prevention and control model after current risk prevention and control model and training is all isolated forest model.
It, can be from set when the sample number of the target service data in set D reaches predetermined quantity by taking step S120 as an example
The sample of predetermined quantity is taken out in D, and takes out the risk prevention system model being being currently used from on-line system, is then increased
Amount training.For example, taking out 500,000 samples when the sample size in set D reaches 500,000, being instructed to risk prevention system model
Practice.When the sample of taking-up, the time sequencing that can be generated based on sample is carried out.
It should be understood, of course, that the sample of trained mistake, it is necessary to be removed from set D.
It should be understood that current risk prevention and control model is allowed in the presence of the maximum height allowed
Maximum height be pre-configured.
It should be understood that the average path length of current risk prevention and control model is based on leaf in the current risk prevention and control model
Node number weighting determination.
S140, the current risk prevention and control model by the on-line system in the target service replace with the target risk
Prevention and control model.
In the embodiment of the present application, sampling and resampling are carried out by the target service data to on-line system, based on adopting again
Target service data after sample carry out incremental training to the current risk prevention and control model of on-line system, and then update on-line system
Current risk prevention and control model improves risk prevention system model so as to improve the perception velocities to business platform Risk Variation
Renewal speed, to improve the risk supervision efficiency of risk prevention system model.
In addition, the scheme of the embodiment of the present application, since the source for training the iterative data of risk prevention system model is wind
The scene of dangerous prevention and control model application, the data of the two are naturally consistent, thus the scheme of the embodiment of the present application also can avoid instruction
Practice the problem inconsistent with the data of application, improves the accuracy rate of model training.
It should be understood that the training step in step S130 for ease of understanding, this be further described below.
The training process of conventional isolated forest model is as follows:
Step 1: randomly choosing the sample point of Ψ point from training sample data as subsample, be put into the root of tree
Node.
Step 2: a specified dimension (attribute) of training sample data is randomly assigned, in present node data
A cut point p is randomly generated, which results from present node data between the maximum value and minimum value of specified dimension.
Step 3: a hyperplane being generated with this cut point, it is empty that present node data space is then divided into 2 sons
Between: the sample data the specified dimension less than p is placed on the left child of present node, which is more than or equal to the sample of p
Notebook data is placed on the right child of present node.
Step 4: the recursion step 2 and 3 in child nodes constantly constructs new child nodes, until in child nodes
The restriction height for having data (can not be further continued for cutting) or child nodes to arrived isolated forest model (allows
Maximum height).
After obtaining t iTree, the training of isolated forest model (iForest) just finishes.At this point it is possible to life
At iForest assess test data.For a training sample data x, it can be enabled to traverse each iTree, then counted
It calculates x and finally falls in each which layer of tree (height of the x in tree), to obtain x in the height average of each tree.If x falls in one
Contain multiple training sample data in a node, can be used one, only relevant formula corrects the height of x with node size size
It calculates.
On the basis of conventional isolated forest model, the incremental training to isolated forest model of step S130 can divide
The other each iTree to current isolated forest model is updated, and the steps include:
Step 10: first to the data in the D of step S120, applying step 1 obtains subsample D_s.
Step 20: D_s being cut according to the iTree structure of current isolated forest model, is updated after D_s is included in
The counting of each node of iTree.
Due to the cut ring section in step 20 be substantially D_s is predicted, and in existing model line every friendship
Yi Douhui marking, need to only retain the log of prediction, at this time without computing repeatedly.
So far, the data increment training of step S130 is completed.
Optionally, as a preferred solution, step S140 specifically can be achieved are as follows:
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control
Model, comprising:
AB test is carried out in the target service to the on-line system based on the target risk prevention and control model；
When the first evaluation index of AB test meets the first preset condition, by the on-line system in the target
The current risk prevention and control model of business replaces with the target risk prevention and control model.
By carrying out AB test, formal empirical flow of the flow as new departure in part is cut, to verify the new side under confrontation
Whether case can be used.If finding that the index of new departure is to be significantly better than old scheme in statistical significance after AB test, then executing
Models switching.
In the embodiment of the present application, after target risk prevention and control model is tested by AB, just by the current wind of on-line system
Dangerous prevention and control model replaces with target risk prevention and control model, so as to guarantee the model in on-line system using preferably detection mould
Type avoids being introduced directly into the model without actual test and bringing immesurable loss.
It should be understood, of course, that in the embodiment of the present application, first preset condition can include: described in the AB test
First evaluation index of target risk prevention and control model is greater than the first evaluation index of the current risk prevention and control model.
In general, when the related evaluation index of only target risk prevention and control model is greater than current risk prevention and control model,
It needs to be updated risk prevention system model.Particularly, the only related evaluation index of target risk prevention and control model is beyond current
It when risk prevention system model certain proportion, just needs to be updated risk prevention system model, to avoid frequent updating risk prevention system mould
Type causes detection efficiency to reduce.
Should immediately, since currently employed risk prevention system model is unsupervised model, which can be friendship
More businessman visitor group degrees of overlapping etc. reflect the side operational indicator of abnormality detection result indirectly after Yi Hou trade company returned money rate, transaction.When
So, also other indexs are also not excluded for.
Further, AB survey is carried out in the target service to the on-line system based on the target risk prevention and control model
Examination, comprising:
When the second evaluation index of the target risk prevention and control model meets the second preset condition, it is based on the target wind
Dangerous prevention and control model carries out AB test in the target service to the on-line system.
After the completion of training, 1,000,000 samples occurred after training set is taken to be assessed.Due to using unsupervised mould
Type, thus assessment herein is indirect.For example, side operational indicator such as trade company's returned money rate after trading of transaction can be used, hand over
The indexs such as more businessman visitor group degrees of overlapping are measured after easily.
In the embodiment of the present application, it is pre- that the second evaluation index of the target risk prevention and control model after incremental training meets second
If condition just carries out AB test, exempts to introduce the obvious poor risk prevention system model progress AB test of effect, can be further improved
The update efficiency of risk prevention system model, while also can avoid abnormality detection efficiency and being decreased obviously in the more new stage.
It should be understood, of course, that the second preset condition, for example, at least may include one of following:
Second evaluation index of the target risk prevention and control model is greater than preset threshold；
Second evaluation index of the target risk prevention and control model is greater than the second assessment of the current risk prevention and control model
The preset ratio of index.
It should be understood that second evaluation index includes following at least one:
Trade company's remittance rate after the transaction of the target service；
More businessman visitor group degrees of overlapping after the transaction of the target service.
It should be understood, of course, that these second evaluation indexes are applicable to the evaluation of the marketing result to transaction platform, for example,
The evaluation for being applicable to the new client culture to transaction platform, promoting the marketing activities such as any active ues, culture user's viscosity.Marketing
Effect is better, and the abnormality detection effect for also reflecting risk prevention system model indirectly is better.
It should be understood that the second evaluation index may be the same or different with the first evaluation index.
Optionally, as one embodiment, step S140 specifically be can be achieved are as follows:
When the second evaluation index of the target risk prevention and control model meets the second preset condition, by the on-line system
The target risk prevention and control model is replaced in the current risk prevention and control model of the target service.
Fig. 2 is the training method of the unsupervised risk prevention system model of one embodiment of the application, comprising:
S210, the business datum based on prefixed time interval to on-line system in target service sample；
S220, the business datum based on sampling carry out resampling to obtain target service data, and the resampling is at least wrapped
Include the credible filter operation of data；
S230 has neither part nor lot in instruction based on predetermined quantity when having neither part nor lot in trained target service data and reaching predetermined quantity
Experienced target service data, the current risk prevention and control model to the on-line system in the target service carry out incremental training,
To obtain target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
In the embodiment of the present application, sampling and resampling are carried out by the target service data to on-line system, based on adopting again
Target service data after sample carry out incremental training to the current risk prevention and control model of on-line system, anti-so as to shorten risk
The cycle of training of model is controlled, renewal speed is improved.
In addition, the scheme of the embodiment of the present application, since the source for training the iterative data of risk prevention system model is wind
The scene of dangerous prevention and control model application, the data of the two are naturally consistent, thus the scheme of the embodiment of the present application also can avoid instruction
Practice the problem inconsistent with the data of application, improves the accuracy rate of model training.
The specific implementation of each step of the embodiment of the present application, can refer to corresponding step in embodiment illustrated in fig. 1, no longer superfluous
It states.
It is above-mentioned that this specification specific embodiment is described.Other embodiments are in the scope of the appended claims
It is interior.In some cases, the movement recorded in detail in the claims or step can be come according to the sequence being different from embodiment
It executes and desired result still may be implemented.In addition, process depicted in the drawing not necessarily require show it is specific suitable
Sequence or consecutive order are just able to achieve desired result.In some embodiments, multitasking and parallel processing be also can
With or may be advantageous.
Fig. 3 is the structural schematic diagram of one embodiment electronic equipment of the application.Referring to FIG. 3, in hardware view, the electricity
Sub- equipment includes processor, optionally further comprising internal bus, network interface, memory.Wherein, memory may be comprising interior
It deposits, such as high-speed random access memory (Random-Access Memory, RAM), it is also possible to further include non-volatile memories
Device (non-volatile memory), for example, at least 1 magnetic disk storage etc..Certainly, which is also possible that other
Hardware required for business.
Processor, network interface and memory can be connected with each other by internal bus, which can be ISA
(Industry Standard Architecture, industry standard architecture) bus, PCI (Peripheral
Component Interconnect, Peripheral Component Interconnect standard) bus or EISA (Extended Industry Standard
Architecture, expanding the industrial standard structure) bus etc..The bus can be divided into address bus, data/address bus, control always
Line etc..Only to be indicated with a four-headed arrow in Fig. 3, it is not intended that an only bus or a type of convenient for indicating
Bus.
Memory, for storing program.Specifically, program may include program code, and said program code includes calculating
Machine operational order.Memory may include memory and nonvolatile memory, and provide instruction and data to processor.
Processor is from the then operation into memory of corresponding computer program is read in nonvolatile memory, in logical layer
Model modification device is formed on face.Processor executes the program that memory is stored, and is specifically used for executing following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control
Model.
The method that the updating device of unsupervised risk prevention system model disclosed in the above-mentioned embodiment illustrated in fig. 1 such as the application executes
It can be applied in processor, or realized by processor.Processor may be a kind of IC chip, the place with signal
Reason ability.During realization, each step of the above method can by the integrated logic circuit of the hardware in processor or
The instruction of software form is completed.Above-mentioned processor can be general processor, including central processing unit (Central
Processing Unit, CPU), network processing unit (Network Processor, NP) etc.；It can also be Digital Signal Processing
Device (Digital Signal Processor, DSP), specific integrated circuit (Application Specific Integrated
Circuit, ASIC), field programmable gate array (Field-Programmable Gate Array, FPGA) or other can
Programmed logic device, discrete gate or transistor logic, discrete hardware components.It may be implemented or execute the application implementation
Disclosed each method, step and logic diagram in example.General processor can be microprocessor or the processor can also be with
It is any conventional processor etc..The step of method in conjunction with disclosed in the embodiment of the present application, can be embodied directly in hardware decoding
Processor executes completion, or in decoding processor hardware and software module combination execute completion.Software module can position
In random access memory, flash memory, read-only memory, programmable read only memory or electrically erasable programmable memory, register
In the storage medium of equal this fields maturation.The storage medium is located at memory, and processor reads the information in memory, in conjunction with it
Hardware completes the step of above method.
The method that the electronic equipment can also carry out Fig. 1, and realize the function of embodiment illustrated in fig. 1, the embodiment of the present application exists
This is repeated no more.
Certainly, other than software realization mode, other implementations are not precluded in the electronic equipment of the application, for example patrol
Collect device or the mode of software and hardware combining etc., that is to say, that the executing subject of following process flow is not limited to each patrol
Unit is collected, hardware or logical device are also possible to.
The embodiment of the present application also proposed a kind of computer readable storage medium, the computer-readable recording medium storage one
A or multiple programs, the one or more program include instruction, and the instruction is when by the portable electronic including multiple application programs
When equipment executes, the method that the portable electronic device can be made to execute embodiment illustrated in fig. 1, and be specifically used for executing following behaviour
Make:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control
Model.
Fig. 4 is the structural schematic diagram of the updating device of the unsupervised risk prevention system model of one embodiment of the application.It please join
Fig. 4 is examined, in a kind of Software Implementation, the updating device of unsupervised risk prevention system model can include:
Sampling module 410, the business datum based on prefixed time interval to on-line system in target service sample；
Resampling module 420, business datum based on sampling carry out resampling to obtain target service data, described heavy to adopt
Sample includes at least the credible filter operation of data；
Training module 430, when having neither part nor lot in trained target service data and reaching predetermined quantity, not based on predetermined quantity
The target service data for participating in training, the current risk prevention and control model to the on-line system in the target service carry out increment
Training, to obtain target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model；
Update module 440 replaces in the current risk prevention and control model of the target service on-line system described
Target risk prevention and control model.
The method that the device can also carry out Fig. 1, and realize the function of embodiment illustrated in fig. 1, the embodiment of the present application is herein not
It repeats again.
Fig. 5 is the structural schematic diagram of one embodiment electronic equipment of the application.Referring to FIG. 5, in hardware view, the electricity
Sub- equipment includes processor, optionally further comprising internal bus, network interface, memory.Wherein, memory may be comprising interior
It deposits, such as high-speed random access memory (Random-Access Memory, RAM), it is also possible to further include non-volatile memories
Device (non-volatile memory), for example, at least 1 magnetic disk storage etc..Certainly, which is also possible that other
Hardware required for business.
Processor, network interface and memory can be connected with each other by internal bus, which can be ISA
(Industry Standard Architecture, industry standard architecture) bus, PCI (Peripheral
Component Interconnect, Peripheral Component Interconnect standard) bus or EISA (Extended Industry Standard
Architecture, expanding the industrial standard structure) bus etc..The bus can be divided into address bus, data/address bus, control always
Line etc..Only to be indicated with a four-headed arrow in Fig. 5, it is not intended that an only bus or a type of convenient for indicating
Bus.
Memory, for storing program.Specifically, program may include program code, and said program code includes calculating
Machine operational order.Memory may include memory and nonvolatile memory, and provide instruction and data to processor.
Processor is from the then operation into memory of corresponding computer program is read in nonvolatile memory, in logical layer
Model training apparatus is formed on face.Processor executes the program that memory is stored, and is specifically used for executing following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
The method that the updating device of unsupervised risk prevention system model disclosed in the above-mentioned embodiment illustrated in fig. 2 such as the application executes
It can be applied in processor, or realized by processor.Processor may be a kind of IC chip, the place with signal
Reason ability.During realization, each step of the above method can by the integrated logic circuit of the hardware in processor or
The instruction of software form is completed.Above-mentioned processor can be general processor, including central processing unit (Central
Processing Unit, CPU), network processing unit (Network Processor, NP) etc.；It can also be Digital Signal Processing
Device (Digital Signal Processor, DSP), specific integrated circuit (Application Specific Integrated
Circuit, ASIC), field programmable gate array (Field-Programmable Gate Array, FPGA) or other can
Programmed logic device, discrete gate or transistor logic, discrete hardware components.It may be implemented or execute the application implementation
Disclosed each method, step and logic diagram in example.General processor can be microprocessor or the processor can also be with
It is any conventional processor etc..The step of method in conjunction with disclosed in the embodiment of the present application, can be embodied directly in hardware decoding
Processor executes completion, or in decoding processor hardware and software module combination execute completion.Software module can position
In random access memory, flash memory, read-only memory, programmable read only memory or electrically erasable programmable memory, register
In the storage medium of equal this fields maturation.The storage medium is located at memory, and processor reads the information in memory, in conjunction with it
Hardware completes the step of above method.
The method that the electronic equipment can also carry out Fig. 2, and realize the function of embodiment illustrated in fig. 2, the embodiment of the present application exists
This is repeated no more.
Certainly, other than software realization mode, other implementations are not precluded in the electronic equipment of the application, for example patrol
Collect device or the mode of software and hardware combining etc., that is to say, that the executing subject of following process flow is not limited to each patrol
Unit is collected, hardware or logical device are also possible to.
The embodiment of the present application also proposed a kind of computer readable storage medium, the computer-readable recording medium storage one
A or multiple programs, the one or more program include instruction, and the instruction is when by the portable electronic including multiple application programs
When equipment executes, the method that the portable electronic device can be made to execute embodiment illustrated in fig. 2, and be specifically used for executing following behaviour
Make:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations are at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained mesh is had neither part nor lot in based on predetermined quantity
Business datum is marked, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain
Target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
Fig. 6 is the structural schematic diagram of the updating device of the unsupervised risk prevention system model of one embodiment of the application.It please join
Fig. 6 is examined, in a kind of Software Implementation, the updating device of unsupervised risk prevention system model can include:
Sampling module 610, the business datum based on prefixed time interval to on-line system in target service sample；
Resampling module 620, business datum based on sampling carry out resampling to obtain target service data, described heavy to adopt
Sample includes at least the credible filter operation of data；
Training module 630, when having neither part nor lot in trained target service data and reaching predetermined quantity, not based on predetermined quantity
The target service data for participating in training, the current risk prevention and control model to the on-line system in the target service carry out increment
Training, to obtain target risk prevention and control model, wherein the current risk prevention and control model is isolated forest model.
The method that the device can also carry out Fig. 2, and realize the function of embodiment illustrated in fig. 2, the embodiment of the present application is herein not
It repeats again.
In short, being not intended to limit the protection scope of the application the foregoing is merely the preferred embodiment of the application.
Within the spirit and principles of this application, any modification, equivalent replacement, improvement and so on should be included in the application's
Within protection scope.
System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity,
Or it is realized by the product with certain function.It is a kind of typically to realize that equipment is computer.Specifically, computer for example may be used
Think personal computer, laptop computer, cellular phone, camera phone, smart phone, personal digital assistant, media play
It is any in device, navigation equipment, electronic mail equipment, game console, tablet computer, wearable device or these equipment
The combination of equipment.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including described want
There is also other identical elements in the process, method of element, commodity or equipment.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for system reality
For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
Part explanation.
Claims (17)
1. a kind of update method of unsupervised risk prevention system model, comprising:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out re-sampling operations to obtain target service data, and the re-sampling operations include at least
Filter operation that data are credible；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control model.
2. the method as described in claim 1,
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control model,
Include:
AB test is carried out in the target service to the on-line system based on the target risk prevention and control model；
When the first evaluation index of AB test meets the first preset condition, by the on-line system in the target service
Current risk prevention and control model replace with the target risk prevention and control model.
3. method according to claim 2, first preset condition includes: target risk prevention and control described in the AB test
First evaluation index of model is greater than the first evaluation index of the current risk prevention and control model.
4. method as claimed in claim 2 or claim 3, based on the target risk prevention and control model to the on-line system in the mesh
Mark business carries out AB test, comprising:
It is anti-based on the target risk when the second evaluation index of the target risk prevention and control model meets the second preset condition
It controls model and AB test is carried out in the target service to the on-line system.
5. method as claimed in claim 4,
Second preset condition includes:
Second evaluation index of the target risk prevention and control model is greater than preset threshold；And/or
Second evaluation index of the target risk prevention and control model is greater than the second evaluation index of the current risk prevention and control model
Preset ratio.
6. method as claimed in claim 5,
Second evaluation index includes following at least one:
Trade company's remittance rate after the transaction of the target service；
More businessman visitor group degrees of overlapping after the transaction of the target service.
7. the method as described in claim 1,
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control model,
Include:
When the second evaluation index of the target risk prevention and control model meets the second preset condition, by the on-line system in institute
The current risk prevention and control model for stating target service replaces with the target risk prevention and control model.
8. the method as described in claim 1, the re-sampling operations further include: filter feature in the business datum of the sampling
Data of the dimension other than preset range.
9. the method as described in claim 1, the maximum height that the current risk prevention and control model allows is to be pre-configured.
10. the method as described in claim 1, the average path length of the current risk prevention and control model is based on described current
Leaf node number weighting determination in risk prevention system model.
11. a kind of training method of unsupervised risk prevention system model, comprising:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out resampling to obtain target service data, and it is credible that the resampling includes at least data
Filter operation；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model.
12. a kind of updating device of unsupervised risk prevention system model, comprising:
Sampling module, the business datum based on prefixed time interval to on-line system in target service sample；
Resampling module, the business datum based on sampling carry out resampling to obtain target service data, and the resampling is at least
Including the credible filter operation of data；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control model.
13. a kind of training device of unsupervised risk prevention system model, comprising:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out resampling to obtain target service data, and it is credible that the resampling includes at least data
Filter operation；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model.
14. a kind of electronic equipment, comprising:
Processor；And
It is arranged to the memory of storage computer executable instructions, the executable instruction makes the processor when executed
Execute following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out resampling to obtain target service data, and it is credible that the resampling includes at least data
Filter operation；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control model.
15. a kind of computer readable storage medium, the computer-readable recording medium storage one or more program, described one
A or multiple programs are when the electronic equipment for being included multiple application programs executes, so that the electronic equipment executes following behaviour
Make:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out resampling to obtain target service data, and it is credible that the resampling includes at least data
Filter operation；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model；
Current risk prevention and control model by the on-line system in the target service replaces with the target risk prevention and control model.
16. a kind of electronic equipment, comprising:
Processor；And
It is arranged to the memory of storage computer executable instructions, the executable instruction makes the processor when executed
Execute following operation:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out resampling to obtain target service data, and it is credible that the resampling includes at least data
Filter operation；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model.
17. a kind of computer readable storage medium, the computer-readable recording medium storage one or more program, described one
A or multiple programs are when the electronic equipment for being included multiple application programs executes, so that the electronic equipment executes following behaviour
Make:
Business datum based on prefixed time interval to on-line system in target service samples；
Business datum based on sampling carries out resampling to obtain target service data, and it is credible that the resampling includes at least data
Filter operation；
When having neither part nor lot in trained target service data and reaching predetermined quantity, trained target industry is had neither part nor lot in based on predetermined quantity
Business data, the current risk prevention and control model to the on-line system in the target service carries out incremental training, to obtain target
Risk prevention system model, wherein the current risk prevention and control model is isolated forest model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910058892.3A CN110009359A (en) | 2019-01-22 | 2019-01-22 | Training method, update method and the device of unsupervised risk prevention system model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910058892.3A CN110009359A (en) | 2019-01-22 | 2019-01-22 | Training method, update method and the device of unsupervised risk prevention system model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110009359A true CN110009359A (en) | 2019-07-12 |
Family
ID=67165472
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910058892.3A Pending CN110009359A (en) | 2019-01-22 | 2019-01-22 | Training method, update method and the device of unsupervised risk prevention system model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110009359A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111275547A (en) * | 2020-03-19 | 2020-06-12 | 重庆富民银行股份有限公司 | Wind control system and method based on isolated forest |
| CN111275071A (en) * | 2020-01-06 | 2020-06-12 | 支付宝(杭州)信息技术有限公司 | Prediction model training method, prediction device and electronic equipment |
| CN111488170A (en) * | 2020-04-07 | 2020-08-04 | 支付宝(杭州)信息技术有限公司 | Method, device and equipment for updating business processing model |
| CN111815177A (en) * | 2020-07-10 | 2020-10-23 | 杭州海康消防科技有限公司 | Fire safety assessment method, server, system and storage medium |
| CN112269794A (en) * | 2020-09-16 | 2021-01-26 | 连尚（新昌）网络科技有限公司 | Method and equipment for violation prediction based on block chain |
| CN112597209A (en) * | 2020-12-15 | 2021-04-02 | 深圳前海微众银行股份有限公司 | Data verification method, device and system and computer readable storage medium |
-
2019
- 2019-01-22 CN CN201910058892.3A patent/CN110009359A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111275071A (en) * | 2020-01-06 | 2020-06-12 | 支付宝(杭州)信息技术有限公司 | Prediction model training method, prediction device and electronic equipment |
| CN111275547A (en) * | 2020-03-19 | 2020-06-12 | 重庆富民银行股份有限公司 | Wind control system and method based on isolated forest |
| CN111275547B (en) * | 2020-03-19 | 2023-07-18 | 重庆富民银行股份有限公司 | Wind control system and method based on isolated forest |
| CN111488170A (en) * | 2020-04-07 | 2020-08-04 | 支付宝(杭州)信息技术有限公司 | Method, device and equipment for updating business processing model |
| CN111815177A (en) * | 2020-07-10 | 2020-10-23 | 杭州海康消防科技有限公司 | Fire safety assessment method, server, system and storage medium |
| CN112269794A (en) * | 2020-09-16 | 2021-01-26 | 连尚（新昌）网络科技有限公司 | Method and equipment for violation prediction based on block chain |
| CN112597209A (en) * | 2020-12-15 | 2021-04-02 | 深圳前海微众银行股份有限公司 | Data verification method, device and system and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110009359A (en) | Training method, update method and the device of unsupervised risk prevention system model | |
| Pennekamp et al. | The intrinsic predictability of ecological time series and its potential to guide forecasting | |
| US11240125B2 (en) | System and method for predicting and reducing subscriber churn | |
| CN111539811B (en) | Risk account identification method and device | |
| CN106897930A (en) | A kind of method and device of credit evaluation | |
| CN104598632A (en) | Hot event detection method and device | |
| CN108550046A (en) | A kind of resource and market recommendation method, apparatus and electronic equipment | |
| CN110060053A (en) | A kind of recognition methods, equipment and computer-readable medium | |
| CN110119860A (en) | A kind of rubbish account detection method, device and equipment | |
| CN110020866A (en) | A kind of training method of identification model, device and electronic equipment | |
| CN113362158A (en) | Credit evaluation method, device and computer readable storage medium | |
| CN111695938B (en) | Product pushing method and system | |
| CN110490635B (en) | Commercial tenant dish transaction prediction and meal preparation method and device | |
| CN115171771A (en) | Solid state disk testing method, device, equipment and storage medium | |
| CN110930078A (en) | Business object identification method, device and equipment | |
| CN110544052A (en) | method and device for displaying relationship network diagram | |
| CN109586990B (en) | Method and device for identifying cheating flow | |
| Cortes et al. | Giga-Mining. | |
| CN110929285B (en) | Method and device for processing private data | |
| CN109271453A (en) | A kind of determination method and apparatus of database volume | |
| CN108733707A (en) | A kind of determining function of search stability and device | |
| CN110852602A (en) | Data monitoring method and device based on machine learning | |
| CN110490595A (en) | A kind of risk control method and device | |
| Riley | Can mutual fund stars still pick stocks?: A replication and extension of Kosowski, Timmermann, Wermers, and White (2006) | |
| CN116362823A (en) | Recommendation model training method, recommendation method and recommendation device for behavior sparse scene |
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: 20200922 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman, British Islands Applicant after: Innovative advanced technology Co.,Ltd. Address before: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman, British Islands Applicant before: Advanced innovation technology Co.,Ltd. Effective date of registration: 20200922 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman, British Islands 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. |