CN110472745A - Information transferring method and device in a kind of federal study - Google Patents
Information transferring method and device in a kind of federal study Download PDFInfo
- Publication number
- CN110472745A CN110472745A CN201910720601.2A CN201910720601A CN110472745A CN 110472745 A CN110472745 A CN 110472745A CN 201910720601 A CN201910720601 A CN 201910720601A CN 110472745 A CN110472745 A CN 110472745A
- Authority
- CN
- China
- Prior art keywords
- result
- participant
- coordinator
- coding
- rule
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L2209/00—Additional information or applications relating to cryptographic mechanisms or cryptographic arrangements for secret or secure communication H04L9/00
- H04L2209/46—Secure multiparty computation, e.g. millionaire problem
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Economics (AREA)
- Finance (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses the information transferring methods and device in a kind of federal study, wherein method is suitable for the federal study including first participant and second participant, it wherein, include identical sample object and different sample characteristics between the first participant and the training dataset of the second participant；The described method includes: coordinator obtains the second result of the first result and the second participant from the first participant；The coordinator carries out operation according to default operation rule, to first result and second result, the third result after obtaining operation；The third result is sent to the first participant and the second participant by the coordinator.When the above method is applied to financial technology (Fintech), on the basis of ensuring communication safety, it is possible to reduce communications duration significantly improves communication efficiency.
Description
Technical field
The present invention relates to the financial technology field (Fintech) and computer software fields more particularly to a kind of federal study
In information transferring method and device.
Background technique
With the development of computer technology, more and more technologies (big data, distribution, block chain (Blockchain),
Artificial intelligence etc.) it applies in financial field, traditional financial industry gradually changes to financial technology (Fintech).Currently, financial
The adjustment of many monetary devices all relies on the data biography carried out in federal study to a large amount of data of financial transaction in sciemtifec and technical sphere
It is defeated as a result, the adjustment of corresponding monetary device is likely to impact the profit and loss of financial institution.Therefore, to a financial institution
For, the efficiency of the data transmission in federation's study is most important.
Two participant A and B exchange intermediate results are needed in data transmission in federation's study.However, federal at present learn
In the scene of data transmission in habit, may there is no cooperative relationship or mutual trust before participant A and B, on the one hand, participate in
Connection communication may not be established directly between person A and B, it is more long temporarily to create new connection communication time-consuming, on the other hand, even if
Connection communication is had been set up, participant A and B need to negotiate code key to be confirmed each other the information of other side, this can be related to complexity
Encryption-decryption algorithm, communication efficiency is lower.
Summary of the invention
The embodiment of the present application provides the information transferring method and device in a kind of federal study, solves and joins in the prior art
The lower problem of communication efficiency between person.
In a first aspect, the embodiment of the present application provides the information transferring method in a kind of federal study, the method is suitable for
Federal study including first participant and second participant, wherein the instruction of the first participant and the second participant
Practicing between data set includes identical sample object and different sample characteristics；It is come from the described method includes: coordinator obtains
The first result of the first participant and the second result of the second participant；Wherein, first result is described the
The knot that one participant concentrates the corresponding data of identical sample object to be calculated using the training data of the first participant
Fruit；Second result is training data concentration identical sample object of the second participant using the second participant
The result that corresponding data are calculated；The coordinator is according to default operation rule, to first result and described second
As a result operation is carried out, the third result after obtaining operation；The third result is sent to described first and participated in by the coordinator
Person and the second participant so that the first participant according to the default operation rule solve described second as a result,
And the second participant solves first result according to the default operation rule.
In the above method, coordinator obtains fortune according to the first result and second is got as a result, according to default operation rule
Third after calculation is as a result, be sent to first participant and second participant for third result again, to avoid in transmission process
To the Encryption Algorithm of the first result and the second result complexity, reduce transmission duration, so that first participant and second participates in
Person is by third result and first result or described second as a result, as a result, in addition also passing through required for getting respectively
As long as third result conceals the first result or second as a result, according to default operation rule, and owned first result or
The first participant or second participant of two results, required for capable of just solving as a result, on the basis to ensure communication safety
On, it is possible to reduce communications duration significantly improves communication efficiency.
In a kind of optional embodiment, the default operation rule includes but is not limited to binary bits transformation rule and different
Or bit arithmetic；The coordinator carries out operation according to default operation rule, to first result and second result, obtains
Third result after operation, comprising: the coordinator is according to the binary bits transformation rule, by first result and institute
State the second result respectively correspond be converted into the first coding and second coding；The coordinator is to first coding and described second
Coding carries out the bit exclusive-OR operation, using the result of the bit exclusive-OR operation as the third result.
In the above method, coordinator is right respectively by the first result and the second result first according to binary bits transformation rule
The first coding and the second coding for being easy to computer identification should be converted into；And first coding and second coding are carried out
The bit exclusive-OR operation, since bit exclusive-OR operation is the basic operation of computer, operation efficiency is higher, obtains to improve
Take the efficiency of third result.
In a kind of optional embodiment, the third result is sent to the first participant and described by the coordinator
Second participant, comprising: the third result is sent to described first by broadcast and/or multicast by the coordinator
Participant and the second participant.
In the above method, coordinator can choose broadcast and/or multicast, can simultaneously send the third result
To first participant and second participant, the flexibility for sending third result is improved.
Second aspect, the embodiment of the present application provide the information transferring method in a kind of federal study, and this method is suitable for packet
Include the federal study of first participant and second participant, wherein the training of the first participant and the second participant
It include identical sample object and different sample characteristics between data set；The described method includes: first participant uses institute
The corresponding data of the identical sample object of training data concentration for stating first participant carry out that the first result is calculated；Described
First result is sent to coordinator by one participant；The first participant receives the third knot from the coordinator
Fruit；The third is the result is that the coordinator obtains according to default operation rule according to first result and the second result operation
The result taken；Second result is training data concentration identical sample of the second participant using the second participant
The result that the corresponding data of this object are calculated；The first participant is according to the default operation rule, according to described
One result and the third are as a result, solving second result and carrying out machine learning model instruction using second result
Practice.
Under aforesaid way, first participant concentrates identical sample object pair using the training data of the first participant
The data answered obtain first as a result, receiving the third result from the coordinator again after being trained；And third is the result is that institute
Coordinator is stated according to default operation rule, according to first result and the second result operation obtain as a result, to reduce
Duration is transmitted, so that first participant can be by third result and described first as a result, getting second as a result, being transmitted across
The Encryption Algorithm to the second result complexity is avoided in journey, conceals second as a result, only according to pre- additionally by third result
If operation rule, and owned first as a result, second could be solved according to third result as a result, to ensuring communication safety
On the basis of, it is possible to reduce communications duration significantly improves communication efficiency.
In a kind of optional embodiment, the default operation rule includes but is not limited to binary bits transformation rule and different
Or bit arithmetic；The first participant is according to the default operation rule, according to first result and the third as a result, asking
Solve second result, comprising: the first participant is according to the binary bits transformation rule, by first result
It is respectively corresponded with the third result and is converted into the first coding and third coding；The first participant to it is described first coding and
The third coding carries out the bit exclusive-OR operation, and using the result of the bit exclusive-OR operation as second result.
Under aforesaid way, first participant is first according to binary bits transformation rule, by the first result and third result point
The first coding and third coding for being easy to computer identification Dui Ying be converted into；And first coding and the third are encoded
The bit exclusive-OR operation is carried out, since bit exclusive-OR operation is the basic operation of computer, operation efficiency is higher, to be promoted
Obtain the efficiency of second result.
The third aspect, the application provide the data transmission device in a kind of federal study, which is suitable for including first
The federal study of participant and second participant, wherein the training dataset of the first participant and the second participant
Between include identical sample object and different sample characteristics；Described device includes: acquisition module, comes from institute for obtaining
State the first result of first participant and the second result of the second participant；Wherein, first result is described first
The result that participant concentrates the corresponding data of identical sample object to be calculated using the training data of the first participant；
Second result is training data concentration identical sample object pair of the second participant using the second participant
The result that the data answered are calculated；Processing module is used for according to default operation rule, to first result and described second
As a result operation is carried out, the third result after obtaining operation；And for the third result to be sent to the first participant
With the second participant so that the first participant solves second result and institute according to the default operation rule
It states second participant and solves first result according to the default operation rule.
In a kind of optional embodiment, the default operation rule includes but is not limited to binary bits transformation rule and different
Or bit arithmetic；The processing module is specifically used for: according to the binary bits transformation rule, by first result and described
Second result, which respectively corresponds, is converted into the first coding and the second coding；To described in first coding and the second coding progress
Bit exclusive-OR operation, using the result of the bit exclusive-OR operation as the third result.
In a kind of optional embodiment, the processing module is specifically used for:, will be described by broadcast and/or multicast
Third result is sent to the first participant and the second participant.
The beneficial effect of the above-mentioned third aspect and each embodiment of the third aspect can refer to above-mentioned first aspect and first
The beneficial effect of each embodiment of aspect, which is not described herein again.
Fourth aspect, the application propose that the data transmission device in a kind of federal study, the device are suitable for including first
The federal study of participant and second participant, wherein the training dataset of the first participant and the second participant
Between include identical sample object and different sample characteristics；Described device includes: training module, for using described the
The corresponding data of identical sample object obtain the first result after being trained in the data set of one participant；Data transmit mould
Block, for first result to be sent to coordinator；And for receiving the third result from the coordinator；Described
Three the result is that the result that the coordinator according to default operation rule, obtains according to first result and the second result operation；
Second result is training data concentration identical sample object pair of the second participant using the second participant
The result that the data answered are calculated；Processing module is used for according to the default operation rule, according to first result and institute
Third is stated as a result, solving second result and being trained using second result.
In a kind of optional embodiment, the default operation rule includes but is not limited to binary bits transformation rule and different
Or bit arithmetic；The processing module is specifically used for: according to the binary bits transformation rule, by first result and described
Third result, which respectively corresponds, is converted into the first coding and third coding；To described in first coding and third coding progress
Bit exclusive-OR operation, and using the result of the bit exclusive-OR operation as second result.
The beneficial effect of above-mentioned fourth aspect and each embodiment of fourth aspect can refer to above-mentioned second aspect and second
The beneficial effect of each embodiment of aspect, which is not described herein again.
5th aspect, the embodiment of the present application provide a kind of computer equipment, including program or instruction, when described program or refer to
Order is performed, each to execute above-mentioned first aspect and each embodiment of first aspect or above-mentioned second aspect and second aspect
The method of a embodiment.
6th aspect, the embodiment of the present application provides a kind of storage medium, including program or instruction, when described program or instruction
It is performed, it is each to execute above-mentioned first aspect and each embodiment of first aspect or above-mentioned second aspect and second aspect
The method of embodiment.
Detailed description of the invention
Fig. 1 be information transferring method in a kind of federal study provided by the embodiments of the present application can application architecture signal
Figure；
Fig. 2 is the step flow diagram of the information transferring method in a kind of federal study provided by the embodiments of the present application；
Fig. 3 is the step flow diagram of the information transferring method in a kind of federal study provided by the embodiments of the present application；
Fig. 4 is the structural schematic diagram of the data transmission device in a kind of federal study provided by the embodiments of the present application；
Fig. 5 is the structural schematic diagram of the data transmission device in a kind of federal study provided by the embodiments of the present application.
Specific embodiment
In order to better understand the above technical scheme, below in conjunction with Figure of description and specific embodiment to above-mentioned
Technical solution is described in detail, it should be understood that the specific features in the embodiment of the present application and embodiment are to the application skill
The detailed description of art scheme, rather than the restriction to technical scheme, in the absence of conflict, the embodiment of the present application
And the technical characteristic in embodiment can be combined with each other.
In financial institution's (banking institution, insurance institution or security organization), in the business of progress, (loan transaction of such as bank is deposited
Money business etc.) in operation process, the adjustment of many monetary devices is all relied on to a large amount of financial transaction numbers in financial technology field
According to the data transmission carried out in federal study as a result, the adjustment of corresponding monetary device is likely to cause the profit and loss of financial institution
It influences.Federation's study refers to the method for carrying out machine learning by combining different participants.In a wheel parameter of federation's study
Update link and be divided into two steps: (a) each participant only uses one's own data and carrys out training machine learning model, and to one
Central coordinator transmission pattern parameter updates；(b) coordinator by the model modification from different participants received into
Row fusion (for example, being averaged), and the update of fused model parameter is distributed to each participant again.In federation's study,
Participant does not need the data to stick one's chin out to other participants or coordinator, thus federal study can be very good to protect
Privacy of user and guarantee data safety.
For a financial institution, the efficiency of the data transmission in federation's study is most important.In the prior art, join
May there is no cooperative relationship or mutual trust before with person A and B, on the one hand, may not build directly between participant A and B
Vertical connection communication, it is more long temporarily to create new connection communication time-consuming, on the other hand, even if having been set up connection communication, participates in
Person A and B need to negotiate code key to be confirmed each other the information of other side, this can be related to complicated encryption-decryption algorithm, also result in
Communication efficiency is lower.Such case does not meet the demand of bank and other financial mechanism, not can guarantee the height of financial institution's items business
Effect operating.
For this purpose, the embodiment of the present application provides the information transferring method in a kind of federal study, this method is suitable for including the
The federal study of one participant and second participant, wherein the training data of the first participant and the second participant
It include identical sample object and different sample characteristics between collection.As shown in Figure 1, for the information transmission in federation study
Method can application structure schematic diagram.For example, there are two participants for belonging to the same area: first participant and second
Participant, wherein first participant is a bank, and second participant is an electric business platform.First participant and second participates in
Person possesses more identical user in areal, but first participant is different from the business of second participant, the use of record
User data feature is different.Particularly, the user data feature of first participant and second participant's record may be complementary
's.Under such scene, framework shown in fig. 1 can be used to help first participant and second participant to construct combination machines
Learning model.In order to help first participant and second participant to combine modeling, coordinator is needed to participate in.Generally assume that the first ginseng
Coordinator is all trusted with person and second participant, but not necessarily has mutual trust to close between first participant and second participant
Direct communication connection cannot be established between system or even first participant and second participant.It is provided using the embodiment of the present application
Method can to avoid federation study first participant and second participant between need directly to exchange intermediate result, be suitable for
The field that code key is distrusted or cannot negotiated between the scene or participant of communication connection cannot be directly established between participant
Scape.
Below with reference to Fig. 2 and Fig. 3, the information transmission side in a kind of federal study provided by the embodiments of the present application is described in detail
Method.Wherein, Fig. 2 shows step flow chart be, coordinator's side data transmission step flow chart；Step process shown in Fig. 3
Figure is that the step flow chart of participant's side data transmission (only illustrates by taking first participant as an example, step performed by second participant
Suddenly it can refer to step performed by first participant, details are not described herein).
Step 201: coordinator obtains second of the first result and the second participant from the first participant
As a result.
Wherein, first result is that the first participant is identical using the training data concentration of the first participant
The corresponding data of sample object calculated after result；Second result is that the second participant uses described second
The training data of participant concentrates the corresponding data of identical sample object to carry out the result after this calculating.It should be noted that
In a kind of possible situation, the first result is that first participant using the training data of the first participant concentrates identical sample
Result of the corresponding data of this object in local computing.
Step 202: the coordinator transports first result and second result according to default operation rule
It calculates, the third result after obtaining operation.
Step 203: the third result is sent to the first participant and the second participant by the coordinator.
Step 203 makes the first participant solve second result and described according to the default operation rule
Second participant solves first result according to the default operation rule.
Step 301: first participant concentrates identical sample object corresponding using the training data of the first participant
Data carry out that the first result is calculated.
Step 302: first result is sent to coordinator by the first participant；
Step 303: the first participant receives the third result from the coordinator.
The third is the result is that the coordinator transports according to default operation rule according to first result and the second result
Calculate the result obtained；Second result is that the second participant is identical using the training data concentration of the second participant
The corresponding data of sample object be trained after result；
Step 304: the first participant is according to the default operation rule, according to first result and the third
As a result, solving second result and carrying out machine learning model training using second result.
Before step 201, first participant executes step 301, obtains the first result.First participant executes step again
302, the first result is sent to coordinator.It should be noted that second participant can be also according to the step of first participant
It is rapid to execute, obtain the second result.
To which following coordinator executes step 201, coordinator receives the first result and the second result.
Step 202, in a kind of optional embodiment, the default operation rule includes but is not limited to binary bits conversion
Rule and bit exclusive-OR operation；Step 202 is specifically as follows:
The coordinator distinguishes first result and second result according to the binary bits transformation rule
It is corresponding to be converted into the first coding and the second coding；The coordinator carries out first coding and second coding described different
Or bit arithmetic, using the result of the bit exclusive-OR operation as the third result.
First example, first the second coding of coding is binary bit stream.First is encoded to 0010, and second is encoded to
0101, then according to bit exclusive-OR operation, third is encoded to 0111.
It should be noted that the default operation rule in step 202 is not limited to binary bits transformation rule and exclusive or
Bit arithmetic can be illustrated by taking addition and subtraction and same or bit arithmetic as an example there are many selection:
Second example, when default operation rule is addition and subtraction, for example, the first result is X1, and the second result is
X2, third result are Y=X1+X2.
Third example, when default operation rule is same or bit arithmetic, for example, first is encoded to 0010, second
It is encoded to 0101, then according to bit exclusive-OR operation, third is encoded to 1000.
A kind of optional embodiment of step 203 is as follows:
Coordinator is sent to the first participant and described by broadcast and/or multicast, by the third result
Second participant.So as to reduce communications number, communication efficiency is significantly improved.
After step 203, first participant executes step 303, to receive third result.Correspondingly, it second participates in
Person can also receive third result.
Next, first participant executes step 304.A kind of optional embodiment of step 304 are as follows: the default operation
Rule includes but is not limited to binary bits transformation rule and bit exclusive-OR operation；The first participant is according to the binary system ratio
First result and the third result are respectively corresponded and are converted into the first coding and third coding by special transformation rule；It is described
First participant carries out the bit exclusive-OR operation to first coding and third coding, and by the bit exclusive-OR operation
As a result it is used as second result.
Illustrate the optional embodiment of step 304 with the correspondence example of three specific examples in step 202 below.
For first example, first is encoded to 0010, and third is encoded to 0111, then is compiled according to bit exclusive-OR operation, second
Code is 0101.
For second example, first is encoded to X1, and third is encoded to Y, then is encoded to Y-X1=according to addition and subtraction, second
X2。
For third example, first is encoded to 0010, and third is encoded to 1000, then is compiled according to same or bit arithmetic, second
Code is 0101.
Illustrate in the information transferring method in a kind of federal study provided by the embodiments of the present application the below with an example
One result and the second result.First participant is participant A；Second participant is participant B.
Participant A and B one linear regression model (LRM) of joint mapping together.Participant A possesses dataWherein, DA
Indicate the training dataset of participant A.Participant B possesses data and labelWherein, DBIndicate the instruction of participant B
Practice data set, yiIndicate i-th group of dataCorresponding label, i.e.,And function f () exactly joins
The function of combination learning is needed with person A and B.Under normal circumstances,WithIt is all multi-C vector, and yiIt is scalar (for example, taking
The scalar that value is 0 or 1, expression are or no).
In linear regression model (LRM), loss function (also referred to as cost function) can be defined asWherein, wAAnd wBIt is to correspond respectively toWithMachine learning model parameter, one
As in the case of be also all multi-C vector.The process of machine learning is exactly to find optimal parameter wAAnd wBSo that loss function L reaches
Small as far as possible (minimizing loss function).
DefinitionWithWork as yiWhen being scalar, loss function L can be further broken into L=LA
+LB+LAB, wherein
In machine learning model training process, decline (gradient descent) method usually using gradient, or
It is the deformation based on gradient descent algorithm, to solve parameter wAAnd wB.Need exist for using loss function L to parameter wAAnd wB's
Gradient.Loss function L is to parameter wAAnd wBGradient can be decomposed into LA, LBAnd LABTo parameter wAAnd wBGradient sum.By with
Upper description is it is found that LAOnly with parameter wAIt is related, and LBOnly with parameter wBIt is related.
Further definitionUse GAIndicate loss function L for parameter wAGradient, thenSimilar uses GBIndicate loss function L for parameter wBGradient, then
Participant A is needed using gradient GATo calculate and undated parameter wA, and participant B is needed using gradient GBTo calculate
With undated parameter wB.However, diWith dataWithAnd parameter wAAnd wBIt is all related.Therefore, participant A can not be calculated independently
Gradient GA, correspondingly, participant B also can not independently calculate gradient GB.For linear regression model (LRM), in a wheel model of federation's study
In shape parameter renewal process, participant A needs to calculate the first resultAnd participant B will be sent to by coordinator；Participant
B needs to calculate the second resultAnd participant A will be sent by coordinator.
For general machine learning model, useIt indicates in the kth wheel model parameter renewal process of federation's study, participates in
Person A needs the first result sent to participant B.Correspondingly, withIndicate updated in the kth wheel model parameter of federation's study
Cheng Zhong, participant B need the second result sent to participant A.The intermediate resultWithIt can not only include parameter
Information and gradient information, it is also possible to further comprise loss function information.Because may be closed without cooperation before participant A and B
System or mutual trust cannot may also directly establish communication connection and transmission intermediate result between A and B.Even if between A and B
It can establish connection, can also be related to establishing between A and B and trust and the additional communication overheads such as code key negotiation.It can be seen that the application
The information transferring method in a kind of federal study that embodiment provides, suitable for cannot directly establish communication connection between participant
Scene or participant between distrust or cannot negotiate the scene of code key.
Subsequent step is referred to step 201~step 203 and the description of step 301~step 304.
It should be noted that in view of data safety, the communication of participant and coordinator and participant and participant it
Between communication all may be encryption, for example, using homomorphic cryptography.
Fig. 4 is the structural schematic diagram of the data transmission device in a kind of federal study provided by the embodiments of the present application.
The application provides the data transmission device in a kind of federal study, the device be suitable for including first participant and
The federal study of two participants, wherein include between the first participant and the training dataset of the second participant
Identical sample object and different sample characteristics；Described device includes: to obtain module 401, comes from described first for obtaining
The first result of participant and the second result of the second participant；Wherein, first result is the first participant
The result for concentrating the corresponding data of identical sample object to be calculated using the training data of the first participant；Described
Two results are training data concentration identical sample object corresponding number of the second participant using the second participant
According to the result calculated；Processing module 402 is used for according to default operation rule, to first result and second knot
Fruit carries out operation, the third result after obtaining operation；And for by the third result be sent to the first participant and
The second participant, so that the first participant solves second result and described according to the default operation rule
Second participant solves first result according to the default operation rule.
In a kind of optional embodiment, the default operation rule includes but is not limited to binary bits transformation rule and different
Or bit arithmetic；The processing module 402 is specifically used for: according to the binary bits transformation rule, by first result and
Second result, which respectively corresponds, is converted into the first coding and the second coding；First coding and second coding are carried out
The bit exclusive-OR operation, using the result of the bit exclusive-OR operation as the third result.
In a kind of optional embodiment, the processing module 402 is specifically used for: by broadcast and/or multicast, by institute
It states third result and is sent to the first participant and the second participant.
Fig. 5 is the structural schematic diagram of the data transmission device in a kind of federal study provided by the embodiments of the present application.
The application proposes the data transmission device in a kind of federal study, the device be suitable for including first participant and
The federal study of two participants, wherein include between the first participant and the training dataset of the second participant
Identical sample object and different sample characteristics；Described device includes: training module 501, for participating in using described first
The corresponding data of identical sample object obtain the first result after being trained in the data set of person；Data transmission module 502 is used
In first result is sent to coordinator；And for receiving the third result from the coordinator；The third knot
Fruit is the coordinator according to default operation rule, the result obtained according to first result and the second result operation；It is described
Second result is that the second participant is corresponding using the identical sample object of training data concentration of the second participant
The result that data are calculated；Processing module 503 is used for according to the default operation rule, according to first result and institute
Third is stated as a result, solving second result and being trained using second result.
In a kind of optional embodiment, the default operation rule includes but is not limited to binary bits transformation rule and different
Or bit arithmetic；The processing module 503 is specifically used for: according to the binary bits transformation rule, by first result and
The third result, which respectively corresponds, is converted into the first coding and third coding；First coding and third coding are carried out
The bit exclusive-OR operation, and using the result of the bit exclusive-OR operation as second result.
The embodiment of the present application provides a kind of computer equipment, including program or instruction, when described program or instruction are performed
When, to execute information transferring method and any optional method in a kind of federal study provided by the embodiments of the present application.
The embodiment of the present application provides a kind of storage medium, including program or instruction, when described program or instruction be performed,
To execute information transferring method and any optional method in a kind of federal study provided by the embodiments of the present application.
Finally, it should be noted that it should be understood by those skilled in the art that, embodiments herein can provide as method, be
System or computer program product.Therefore, the application can be used complete hardware embodiment, complete software embodiment or combine software
With the form of the embodiment of hardware aspect.Moreover, it wherein includes that computer can use journey that the application, which can be used in one or more,
The computer implemented in the computer-usable storage medium (including but not limited to magnetic disk storage, optical memory etc.) of sequence code
The form of program product.
The application be referring to according to the present processes, equipment (system) and computer program product flow chart and/or
Block diagram describes.It should be understood that each process that can be realized by computer program instructions in flowchart and/or the block diagram and/or
The combination of process and/or box in box and flowchart and/or the block diagram.It can provide these computer program instructions to arrive
General purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices processor to generate one
Machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for realizing flowing
The device for the function of being specified in journey figure one process or multiple processes and/or block diagrams one box or multiple boxes.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
Obviously, those skilled in the art can carry out various modification and variations without departing from the model of the application to the application
It encloses.In this way, if these modifications and variations of the application belong within the scope of the claim of this application and its equivalent technologies, then
The application is also intended to include these modifications and variations.
Claims (12)
1. the information transferring method in a kind of federal study, which is characterized in that be suitable for participating in including first participant and second
The federal study of person, wherein include identical between the first participant and the training dataset of the second participant
Sample object and different sample characteristics；The described method includes:
Second result of first result and the second participant of coordinator's acquisition from the first participant；Wherein, institute
Stating the first result is that the first participant is corresponding using the identical sample object of training data concentration of the first participant
The result that is calculated of data；Second result is the training data that the second participant uses the second participant
The result for concentrating the corresponding data of identical sample object to be calculated；
The coordinator carries out operation according to default operation rule, to first result and second result, obtains operation
Third result afterwards；The third result is sent to the first participant and the second participant by the coordinator, with
Make the first participant according to the default operation rule solve described second as a result, and the second participant according to institute
It states default operation rule and solves first result.
2. the method as described in claim 1, which is characterized in that the default operation rule includes but is not limited to binary bits
Transformation rule and bit exclusive-OR operation；The coordinator is according to default operation rule, to first result and second result
Operation is carried out, the third result after obtaining operation, comprising:
The coordinator respectively corresponds first result and second result according to the binary bits transformation rule
It is converted into the first coding and the second coding；
The coordinator carries out the bit exclusive-OR operation to first coding and second coding, by the bit exclusive-OR operation
Result as the third result.
3. method according to claim 1 or 2, which is characterized in that the third result is sent to described by the coordinator
First participant and the second participant, comprising:
The coordinator is sent to the first participant and described by broadcast and/or multicast, by the third result
Second participant.
4. the information transferring method in a kind of federal study, which is characterized in that be suitable for participating in including first participant and second
The federal study of person, wherein include identical between the first participant and the training dataset of the second participant
Sample object and different sample characteristics；The described method includes:
First participant is counted using the corresponding data of the identical sample object of training data concentration of the first participant
Calculation obtains the first result；
First result is sent to coordinator by the first participant；
The first participant receives the third result from the coordinator；The third the result is that the coordinator according to pre-
If operation rule, according to the result of first result and the second result operation acquisition；Second result is second ginseng
The result for concentrating the corresponding data of identical sample object to be calculated using the training data of the second participant with person；
The first participant is according to the default operation rule, according to first result and the third as a result, solving
Second result is simultaneously carrying out machine learning model training using second result.
5. method as claimed in claim 4, which is characterized in that the default operation rule includes but is not limited to binary bits
Transformation rule and bit exclusive-OR operation；The first participant is according to the default operation rule, according to first result and institute
Third is stated as a result, solving second result, comprising:
The first participant distinguishes first result and the third result according to the binary bits transformation rule
It is corresponding to be converted into the first coding and third coding；
The first participant carries out the bit exclusive-OR operation to first coding and third coding, and by the exclusive or
The result of bit arithmetic is as second result.
6. the data transmission device in a kind of federal study, which is characterized in that be suitable for participating in including first participant and second
The federal study of person, wherein include identical between the first participant and the training dataset of the second participant
Sample object and different sample characteristics；Described device includes:
Module is obtained, for obtaining the first result from the first participant and the second result of the second participant；
Wherein, first result is training data concentration identical sample pair of the first participant using the first participant
The result calculated as corresponding data；Second result is the instruction that the second participant uses the second participant
Practice the result that the corresponding data of identical sample object are calculated in data set；
Processing module, for carrying out operation to first result and second result, obtaining fortune according to default operation rule
Third result after calculation；And for the third result to be sent to the first participant and the second participant, with
The first participant is set to solve second result and the second participant according to institute according to the default operation rule
It states default operation rule and solves first result.
7. device as claimed in claim 6, which is characterized in that the default operation rule includes but is not limited to binary bits
Transformation rule and bit exclusive-OR operation；The processing module is specifically used for:
According to the binary bits transformation rule, first result and second result are respectively corresponded and be converted into first
Coding and the second coding；
The bit exclusive-OR operation is carried out to first coding and second coding, using the result of the bit exclusive-OR operation as
The third result.
8. device as claimed in claims 6 or 7, which is characterized in that the processing module is specifically used for:
By broadcast and/or multicast, the third result is sent to the first participant and the second participant.
9. the data transmission device in a kind of federal study, which is characterized in that be suitable for participating in including first participant and second
The federal study of person, wherein include identical between the first participant and the training dataset of the second participant
Sample object and different sample characteristics；Described device includes:
Training module is trained for the corresponding data of sample object identical in the data set using the first participant
After obtain the first result；
Data transmission module, for first result to be sent to coordinator；And for receiving from the coordinator's
Third result；The third the result is that the coordinator according to default operation rule, according to first result and the second result
The result that operation obtains；Second result is training data concentration phase of the second participant using the second participant
The result that the same corresponding data of sample object are calculated；
Processing module is used for according to the default operation rule, according to first result and the third as a result, solving institute
It states the second result and is trained using second result.
10. device as claimed in claim 9, which is characterized in that the default operation rule includes but is not limited to binary system ratio
Special transformation rule and bit exclusive-OR operation；The processing module is specifically used for:
According to the binary bits transformation rule, first result and the third result are respectively corresponded and be converted into first
Coding and third coding；
The bit exclusive-OR operation is carried out to first coding and third coding, and the result of the bit exclusive-OR operation is made
For second result.
11. a kind of computer equipment, which is characterized in that including program or instruction, when described program or instruction are performed, as weighed
Benefit require any one of 1 to 3 or 4 to 5 described in method be performed.
12. a kind of storage medium, which is characterized in that including program or instruction, when described program or instruction are performed, such as right
It is required that method described in any one of 1 to 3 or 4 to 5 is performed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910720601.2A CN110472745B (en) | 2019-08-06 | 2019-08-06 | Information transmission method and device in federated learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910720601.2A CN110472745B (en) | 2019-08-06 | 2019-08-06 | Information transmission method and device in federated learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110472745A true CN110472745A (en) | 2019-11-19 |
| CN110472745B CN110472745B (en) | 2021-04-27 |
Family
ID=68511327
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910720601.2A Active CN110472745B (en) | 2019-08-06 | 2019-08-06 | Information transmission method and device in federated learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110472745B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062044A (en) * | 2019-12-09 | 2020-04-24 | 支付宝(杭州)信息技术有限公司 | Model joint training method and device based on block chain |
| CN111212110A (en) * | 2019-12-13 | 2020-05-29 | 清华大学深圳国际研究生院 | Block chain-based federal learning system and method |
| CN111242316A (en) * | 2020-01-09 | 2020-06-05 | 深圳前海微众银行股份有限公司 | Longitudinal federated learning model training optimization method, device, equipment and medium |
| CN112217706A (en) * | 2020-12-02 | 2021-01-12 | 腾讯科技（深圳）有限公司 | Data processing method, device, equipment and storage medium |
| WO2021154652A1 (en) | 2020-01-27 | 2021-08-05 | Hewlett Packard Enterprise Development Lp | Secure parameter merging using homomorphic encryption for swarm learning |
| CN113240127A (en) * | 2021-04-07 | 2021-08-10 | 睿蜂群(北京)科技有限公司 | Federal learning-based training method and device, electronic equipment and storage medium |
| CN113592097A (en) * | 2021-07-23 | 2021-11-02 | 京东科技控股股份有限公司 | Federal model training method and device and electronic equipment |
| US11748835B2 (en) | 2020-01-27 | 2023-09-05 | Hewlett Packard Enterprise Development Lp | Systems and methods for monetizing data in decentralized model building for machine learning using a blockchain |
| US11775667B2 (en) | 2020-11-04 | 2023-10-03 | Hewlett Packard Enterprise Development Lp | Virtualizing secure storage of a baseboard management controller to a host computing device |
| US11966818B2 (en) | 2019-02-21 | 2024-04-23 | Hewlett Packard Enterprise Development Lp | System and method for self-healing in decentralized model building for machine learning using blockchain |
| CN111242316B (en) * | 2020-01-09 | 2024-05-28 | 深圳前海微众银行股份有限公司 | Longitudinal federal learning model training optimization method, device, equipment and medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180089587A1 (en) * | 2016-09-26 | 2018-03-29 | Google Inc. | Systems and Methods for Communication Efficient Distributed Mean Estimation |
| CN109165515A (en) * | 2018-08-10 | 2019-01-08 | 深圳前海微众银行股份有限公司 | Model parameter acquisition methods, system and readable storage medium storing program for executing based on federation's study |
| CN109167695A (en) * | 2018-10-26 | 2019-01-08 | 深圳前海微众银行股份有限公司 | Alliance Network construction method, equipment and readable storage medium storing program for executing based on federation's study |
| CN109492420A (en) * | 2018-12-28 | 2019-03-19 | 深圳前海微众银行股份有限公司 | Model parameter training method, terminal, system and medium based on federation's study |
| CN109886417A (en) * | 2019-03-01 | 2019-06-14 | 深圳前海微众银行股份有限公司 | Model parameter training method, device, equipment and medium based on federation's study |
| CN110008696A (en) * | 2019-03-29 | 2019-07-12 | 武汉大学 | A kind of user data Rebuilding Attack method towards the study of depth federation |
-
2019
- 2019-08-06 CN CN201910720601.2A patent/CN110472745B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180089587A1 (en) * | 2016-09-26 | 2018-03-29 | Google Inc. | Systems and Methods for Communication Efficient Distributed Mean Estimation |
| CN109165515A (en) * | 2018-08-10 | 2019-01-08 | 深圳前海微众银行股份有限公司 | Model parameter acquisition methods, system and readable storage medium storing program for executing based on federation's study |
| CN109167695A (en) * | 2018-10-26 | 2019-01-08 | 深圳前海微众银行股份有限公司 | Alliance Network construction method, equipment and readable storage medium storing program for executing based on federation's study |
| CN109492420A (en) * | 2018-12-28 | 2019-03-19 | 深圳前海微众银行股份有限公司 | Model parameter training method, terminal, system and medium based on federation's study |
| CN109886417A (en) * | 2019-03-01 | 2019-06-14 | 深圳前海微众银行股份有限公司 | Model parameter training method, device, equipment and medium based on federation's study |
| CN110008696A (en) * | 2019-03-29 | 2019-07-12 | 武汉大学 | A kind of user data Rebuilding Attack method towards the study of depth federation |
Non-Patent Citations (2)
| Title |
|---|
| KEWEI CHENG 等: "SecureBoost: A Lossless Federated Learning Framework", 《ARXIV》 * |
| YUE ZHAO 等: "Federated Learning with Non-IID Data", 《ARXIV》 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11966818B2 (en) | 2019-02-21 | 2024-04-23 | Hewlett Packard Enterprise Development Lp | System and method for self-healing in decentralized model building for machine learning using blockchain |
| CN111062044A (en) * | 2019-12-09 | 2020-04-24 | 支付宝(杭州)信息技术有限公司 | Model joint training method and device based on block chain |
| CN111062044B (en) * | 2019-12-09 | 2021-03-23 | 支付宝(杭州)信息技术有限公司 | Model joint training method and device based on block chain |
| TWI732557B (en) * | 2019-12-09 | 2021-07-01 | 大陸商支付寶（杭州）信息技術有限公司 | Model joint training method and device based on blockchain |
| CN111212110A (en) * | 2019-12-13 | 2020-05-29 | 清华大学深圳国际研究生院 | Block chain-based federal learning system and method |
| CN111242316A (en) * | 2020-01-09 | 2020-06-05 | 深圳前海微众银行股份有限公司 | Longitudinal federated learning model training optimization method, device, equipment and medium |
| CN111242316B (en) * | 2020-01-09 | 2024-05-28 | 深圳前海微众银行股份有限公司 | Longitudinal federal learning model training optimization method, device, equipment and medium |
| US11876891B2 (en) | 2020-01-27 | 2024-01-16 | Hewlett Packard Enterprise Development Lp | Secure parameter merging using homomorphic encryption for swarm learning |
| CN113767386A (en) * | 2020-01-27 | 2021-12-07 | 慧与发展有限责任合伙企业 | Security parameter merging using homomorphic encryption for group learning |
| EP3935776A4 (en) * | 2020-01-27 | 2022-11-30 | Hewlett Packard Enterprise Development LP | Secure parameter merging using homomorphic encryption for swarm learning |
| CN113767386B (en) * | 2020-01-27 | 2023-06-06 | 慧与发展有限责任合伙企业 | Network node operating in a distributed network |
| US11748835B2 (en) | 2020-01-27 | 2023-09-05 | Hewlett Packard Enterprise Development Lp | Systems and methods for monetizing data in decentralized model building for machine learning using a blockchain |
| US11887204B2 (en) | 2020-01-27 | 2024-01-30 | Hewlett Packard Enterprise Development Lp | Systems and methods for monetizing data in decentralized model building for machine learning using a blockchain |
| WO2021154652A1 (en) | 2020-01-27 | 2021-08-05 | Hewlett Packard Enterprise Development Lp | Secure parameter merging using homomorphic encryption for swarm learning |
| US11775667B2 (en) | 2020-11-04 | 2023-10-03 | Hewlett Packard Enterprise Development Lp | Virtualizing secure storage of a baseboard management controller to a host computing device |
| WO2022116725A1 (en) * | 2020-12-02 | 2022-06-09 | 腾讯科技（深圳）有限公司 | Data processing method, apparatus, device, and storage medium |
| CN112217706A (en) * | 2020-12-02 | 2021-01-12 | 腾讯科技（深圳）有限公司 | Data processing method, device, equipment and storage medium |
| CN113240127A (en) * | 2021-04-07 | 2021-08-10 | 睿蜂群(北京)科技有限公司 | Federal learning-based training method and device, electronic equipment and storage medium |
| CN113592097A (en) * | 2021-07-23 | 2021-11-02 | 京东科技控股股份有限公司 | Federal model training method and device and electronic equipment |
| CN113592097B (en) * | 2021-07-23 | 2024-02-06 | 京东科技控股股份有限公司 | Training method and device of federal model and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110472745B (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110472745A (en) | Information transferring method and device in a kind of federal study | |
| CN112101579B (en) | Federal learning-based machine learning method, electronic device, and storage medium | |
| CN112183730B (en) | Neural network model training method based on shared learning | |
| WO2022206510A1 (en) | Model training method and apparatus for federated learning, and device and storage medium | |
| CN110490738A (en) | A kind of federal learning method of mixing and framework | |
| Shao et al. | Multi-objective neural evolutionary algorithm for combinatorial optimization problems | |
| CN110399742A (en) | A kind of training, prediction technique and the device of federation's transfer learning model | |
| CN112257873A (en) | Training method, device, system, equipment and storage medium of machine learning model | |
| CN112132293A (en) | Calculation method, device, equipment and medium in longitudinal federal learning | |
| Du et al. | Beyond deep reinforcement learning: A tutorial on generative diffusion models in network optimization | |
| CN111081337B (en) | Collaborative task prediction method and computer readable storage medium | |
| CN112288094B (en) | Federal network representation learning method and system | |
| Xu et al. | A trustless architecture of blockchain-enabled metaverse | |
| CN111104968A (en) | Safety SVM training method based on block chain | |
| Zuo et al. | A survey of blockchain and artificial intelligence for 6G wireless communications | |
| WO2023174036A1 (en) | Federated learning model training method, electronic device and storage medium | |
| CN114742240A (en) | Transverse federated learning method, device and storage medium | |
| Yun et al. | SlimFL: Federated learning with superposition coding over slimmable neural networks | |
| CN110610098A (en) | Data set generation method and device | |
| CN114491616A (en) | Block chain and homomorphic encryption-based federated learning method and application | |
| Jiang et al. | Deterministic joint refdemote preparation of general multi-qubit states | |
| Tanaka et al. | A faithful distributed implementation of dual decomposition and average consensus algorithms | |
| Alnajar et al. | Tactile internet of federated things: Toward fine-grained design of FL-based architecture to meet TIoT demands | |
| CN113055902A (en) | Intelligent mobile communication network system | |
| CN114723012A (en) | Computing method and device based on distributed training system |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |