CN112632611A - Method, apparatus, electronic device, and storage medium for data aggregation - Google Patents
Method, apparatus, electronic device, and storage medium for data aggregation Download PDFInfo
- Publication number
- CN112632611A CN112632611A CN202011578538.2A CN202011578538A CN112632611A CN 112632611 A CN112632611 A CN 112632611A CN 202011578538 A CN202011578538 A CN 202011578538A CN 112632611 A CN112632611 A CN 112632611A
- Authority
- CN
- China
- Prior art keywords
- calculation
- item
- data
- mask
- encryption
- 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
- 230000002776 aggregation Effects 0.000 title claims abstract description 100
- 238000004220 aggregation Methods 0.000 title claims abstract description 100
- 238000000034 method Methods 0.000 title claims abstract description 55
- 238000004364 calculation method Methods 0.000 claims abstract description 340
- 239000003999 initiator Substances 0.000 claims abstract description 54
- 238000004590 computer program Methods 0.000 claims description 12
- 230000000295 complement effect Effects 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 abstract description 22
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 19
- 230000007246 mechanism Effects 0.000 description 19
- 238000010586 diagram Methods 0.000 description 7
- 230000005540 biological transmission Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000008520 organization Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2455—Query execution
- G06F16/24553—Query execution of query operations
- G06F16/24554—Unary operations; Data partitioning operations
- G06F16/24556—Aggregation; Duplicate elimination
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Bioethics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Storage Device Security (AREA)
Abstract
The application relates to a method, equipment, an electronic device and a storage medium for data aggregation, wherein the method for data aggregation comprises the following steps: acquiring an aggregation rule of an initiator in federated calculation, and splitting the aggregation rule into a sum of a plurality of target items according to Taylor expansion, wherein a data item of each target item corresponds to sample data of one participant only; encrypting the data item of each participant according to the first mask, and calculating a first calculation item of the target item according to the encrypted data item; acquiring a second calculation item of the target item according to the target item and the first calculation item, and acquiring an encryption calculation result of each participant on the second calculation item; and acquiring a data aggregation result of the participant according to the encryption calculation results of the first calculation item and the second calculation item. By the method and the device, the problems that the efficiency of a general algorithm for carrying out federal calculation in the related technology is low and the consumed time is long are solved, and the data privacy of the participants is protected while the data aggregation efficiency is improved.
Description
Technical Field
The present application relates to the field of federal computing, and more particularly to a method, apparatus, electronic device, and storage medium for data aggregation.
Background
With the development of artificial intelligence, people provide a concept of 'federal learning' for solving the problem of data islanding, so that both federal parties can train a model to obtain model parameters without providing own data, and the problem of data privacy disclosure can be avoided. In the development process of enterprise digital transformation, in order to improve the data use quality, data cooperation between organizations is increasingly frequent, such as cooperation between hospitals and pharmaceutical companies. Federal learning is a feasible solution which can meet privacy protection and data security, and private data of all parties cannot be locally obtained through homomorphic encryption, secret sharing and other modes, so that joint calculation and modeling are realized.
In the related art, the special algorithm for performing federal calculation has limited applicable scenes, and the general algorithm has low efficiency and long time consumption.
At present, an effective solution is not provided aiming at the problems of low efficiency and long time consumption of a universal algorithm for carrying out federal calculation in the related technology.
Disclosure of Invention
The embodiment of the application provides a data aggregation method, data aggregation equipment, an electronic device and a storage medium, and aims to at least solve the problems that the efficiency of a universal algorithm for carrying out federal calculation in the related art is low and the time consumption is long.
In a first aspect, an embodiment of the present application provides a method for data aggregation, including:
acquiring an aggregation rule of an initiator in federated calculation, and splitting the aggregation rule into the sum of a plurality of target items according to Taylor expansion, wherein each target item is composed of a plurality of data items, and each data item corresponds to sample data of only one participant;
acquiring a first mask, encrypting the data item of each participant according to the first mask, and calculating a first calculation item of the target item according to the encrypted data item;
acquiring a second calculation item of the target item according to the target item and the first calculation item, and acquiring an encryption calculation result of each participant on the second calculation item;
and acquiring a data aggregation result of the participant according to the encryption calculation results of the first calculation item and the second calculation item, wherein the second calculation item is complementary to the first calculation item, so that the first mask is eliminated.
In one embodiment, obtaining a first mask, encrypting the data item of each participant according to the first mask, and calculating a first calculation item of the target item according to the encrypted data item includes:
acquiring a second mask;
controlling the first participant to perform first encryption calculation on the encrypted data item according to the second mask;
controlling a second party to perform second encryption calculation on the encrypted data item according to the second mask, wherein the first encryption calculation corresponds to the second encryption calculation;
and acquiring a first calculation item of the target item according to the data item after the first encryption calculation and the data item after the second encryption calculation.
In one embodiment, after obtaining the first calculation item of the target item according to the first encryption-calculated data item and the second encryption-calculated data item, the method includes:
acquiring a plurality of first masks, sequentially calculating data items in the target items according to the first masks, and acquiring the first calculation items in all the target items, wherein the first masks are in one-to-one correspondence with the first calculation items.
In one embodiment, obtaining a second computation item of the target item according to the target item and the first computation item, and obtaining an encryption computation result of each participant on the second computation item includes:
acquiring a third mask;
controlling the first participant to carry out third encryption calculation on sample data related to the first participant in the second calculation item according to the third mask;
controlling a second participant to perform fourth encryption calculation on sample data related to the second participant in the second calculation item according to the third mask, wherein the third encryption calculation corresponds to the fourth encryption calculation;
and acquiring the encryption calculation result of the second calculation item according to the results of the third encryption calculation and the fourth encryption calculation.
In one embodiment, obtaining the first mask includes:
acquiring a first reference mask determined by a first party in the federal calculation and a second reference mask determined by a second party in the federal calculation;
calculating the first mask according to the first reference mask and the second reference mask.
In a second aspect, an embodiment of the present application provides a data aggregation device, where the device includes an obtaining module, a first computing module, a second computing module, and an aggregation module:
the obtaining module is used for obtaining an aggregation rule of an initiator in federated calculation, and splitting the aggregation rule into the sum of a plurality of target items according to Taylor expansion, wherein each target item is composed of a plurality of data items, and each data item corresponds to sample data of only one participant;
the first calculation module is configured to obtain a first mask, encrypt the data item of each participant according to the first mask, and calculate a first calculation item of the target item according to the encrypted data item;
the second calculation module is used for acquiring a second calculation item of the target item according to the target item and the first calculation item and acquiring an encryption calculation result of each participant on the second calculation item;
the aggregation module is configured to obtain a data aggregation result of the participant according to an encryption calculation result of the first calculation item and the second calculation item, where the second calculation item is complementary to the first calculation item, so that the first mask is eliminated.
In one embodiment, the first calculation module is further configured to:
acquiring a second mask;
controlling the first participant to perform first encryption calculation on the encrypted data item according to the second mask;
controlling a second party to perform second encryption calculation on the encrypted data item according to the second mask, wherein the first encryption calculation corresponds to the second encryption calculation;
and acquiring a first calculation item of the target item according to the data item after the first encryption calculation and the data item after the second encryption calculation.
In one embodiment, the second calculation module is further configured to:
acquiring a third mask;
controlling the first participant to carry out third encryption calculation on sample data related to the first participant in the second calculation item according to the third mask;
controlling a second participant to perform fourth encryption calculation on sample data related to the second participant in the second calculation item according to the third mask, wherein the third encryption calculation corresponds to the fourth encryption calculation;
and acquiring the encryption calculation result of the second calculation item according to the results of the third encryption calculation and the fourth encryption calculation.
In a third aspect, an embodiment of the present application provides an electronic device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor, when executing the computer program, implements the method for data aggregation according to the first aspect.
In a fourth aspect, the present application provides a storage medium, on which a computer program is stored, where the program is executed by a processor to implement the method for data aggregation as described in the first aspect.
Compared with the related art, the method for data aggregation provided by the embodiment of the application divides the aggregation rule into the sum of a plurality of target items according to Taylor expansion by obtaining the aggregation rule of an initiator in federal calculation, wherein each target item is composed of a plurality of data items, and each data item only corresponds to sample data of one participant; acquiring a first mask, encrypting the data item of each participant according to the first mask, and calculating a first calculation item of a target item according to the encrypted data item; acquiring a second calculation item of the target item according to the target item and the first calculation item, and acquiring an encryption calculation result of each participant on the second calculation item; the data aggregation result of the participator is obtained according to the encryption calculation result of the first calculation item and the second calculation item, wherein the second calculation item is complementary with the first calculation item, so that the first mask is eliminated, the problems of low efficiency and long time consumption of a universal algorithm for carrying out federal calculation in the related art are solved, and the data privacy of the participator is protected while the data aggregation efficiency is improved.
The details of one or more embodiments of the application are set forth in the accompanying drawings and the description below to provide a more thorough understanding of the application.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
fig. 1 is a flowchart of a method of data aggregation in the related art;
FIG. 2 is a flow chart of a method of data aggregation according to an embodiment of the present application;
FIG. 3 is a flow chart of a method of computing a first computation term according to an embodiment of the present application;
FIG. 4 is a flow chart of a method of computing a second computation term according to an embodiment of the present application;
fig. 5 is a block diagram of a hardware structure of a terminal of a data aggregation method according to an embodiment of the present application;
fig. 6 is a block diagram of a data aggregation device according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments provided in the present application without any inventive step are within the scope of protection of the present application. Moreover, it should be appreciated that in the development of any such actual implementation, as in any engineering or design project, numerous implementation-specific decisions must be made to achieve the developers' specific goals, such as compliance with system-related and business-related constraints, which may vary from one implementation to another.
Reference in the specification to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the specification. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of ordinary skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments without conflict.
Unless defined otherwise, technical or scientific terms referred to herein shall have the ordinary meaning as understood by those of ordinary skill in the art to which this application belongs. Reference to "a," "an," "the," and similar words throughout this application are not to be construed as limiting in number, and may refer to the singular or the plural. The present application is directed to the use of the terms "including," "comprising," "having," and any variations thereof, which are intended to cover non-exclusive inclusions; for example, a process, method, system, article, or apparatus that comprises a list of steps or modules (elements) is not limited to the listed steps or elements, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus. Reference to "connected," "coupled," and the like in this application is not intended to be limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. Reference herein to "a plurality" means greater than or equal to two. "and/or" describes an association relationship of associated objects, meaning that three relationships may exist, for example, "A and/or B" may mean: a exists alone, A and B exist simultaneously, and B exists alone. Reference herein to the terms "first," "second," "third," and the like, are merely to distinguish similar objects and do not denote a particular ordering for the objects.
In most industries, data often exists in an isolated island form due to problems of industry competition, privacy security, complex administrative procedures, and the like. The cost of integrating data scattered around a site or at various facilities is enormous. Aiming at the problems of data islanding and data privacy, researchers provide federal calculation so as to obtain correct data calculation results under the condition that data are not exported.
Typically, federal calculations include both proprietary and general purpose algorithms. The Private algorithm, for example, a Private Set Intersection (PSI) can ensure that both parties can safely find the Intersection and do not know data other than the Intersection of each other, a Practical Secure Aggregation (PSA) can find the result of addition, subtraction, multiplication and division of both parties 'data without exposing own data to each other, and a privacy comparison magnitude algorithm can find whose data is larger when both parties compare both parties' data without exposing own data to each other, but the Private algorithm is only applicable to specific scenes, such as Intersection finding, calculation of addition, subtraction, multiplication and division or comparison of magnitude, and has low scene adaptability; general algorithms, such as Garbled Circuits (GC for short) and Fully Homomorphic Encryption (FHE for short), can be applied to any scene, and can obtain any required calculation result without exposing own data.
In the related art, a simple two-party PSA is applied in a scenario where if there is a mechanism B that owns data B and a mechanism C that owns data C, the mechanism a finds the results of B + C, B-C, B C and/or B/C without acquiring specific values of B and C. For example, in a corporate credit investigation scenario, a bank may query the customer's credit score sum in all other banks according to the PSA algorithm, but need not know the customer's specific score in other banks.
In the case that the organization a needs to obtain the result of b + c because of business needs, fig. 1 is a flow chart of a method of data aggregation in the related art, as shown in fig. 1, the method includes the following steps:
step S110, negotiating a mask r between a mechanism B and a mechanism C;
step S120, the mechanism B sends data B-r to the mechanism A, and the mechanism C sends data C + r to the mechanism A;
in step S130, the means a calculates b-r + c + r as b + c.
Through the steps S110 to S130, it is ensured that the original data of the mechanism B and the mechanism C are not exported, the data sent to the party a is a result after being confused, and the party a cannot know the specific numerical value of the mask, so that the values of B and C cannot be restored, but the result after aggregation by the party a is correct.
Further, if the organization a needs to obtain the result of the data b × c, the following calculation may be performed: 1) mechanism B negotiates a mask r with mechanism C; 2) the mechanism B sends data bxr to the mechanism A, and the mechanism C sends data C/r to the mechanism A; 3) the mechanism a calculates b × c ═ b × r × c/r.
The application provides a data aggregation calculation method suitable for complex scenes based on a PSA algorithm.
The embodiment provides a data aggregation method. Fig. 2 is a flowchart of a method of data aggregation according to an embodiment of the present application, as shown in fig. 2, the method including the steps of:
step S210, obtaining an aggregation rule of an initiator in federated calculation, and splitting the aggregation rule into the sum of a plurality of target items according to Taylor expansion, wherein each target item is composed of a plurality of data items, and each data item corresponds to sample data of only one participant.
In this embodiment, the initiator is an initiator for federal learning, the participants are participants who accept invitation or actively enter federal learning, the aggregation rule is formulated by the initiator, specifically, the aggregation rule is a calculation rule that requires sample data of the participants, except for addition, subtraction and multiplication, and under the condition that the initiator is a and the participants include B and C, the aggregation rule in this embodiment may also be more complex calculation, for example, as shown in formula 1:
f(b,c)＝bc-b3+ lnc formula 1
Wherein f (B, C) is the aggregation rule formulated by the initiator a, B is the data in the participant B, and C is the data in the participant C.
According to taylor formula, formula 1 can be decomposed into a sum of a plurality of function products, wherein the function product is a target term, the functions in the function product are data terms, each function is only related to data of one participant, and the decomposed aggregation rule is as shown in formula 2:
f(b,c)＝g1(b)h1(c)+g2(b)h2(c)+…+gn(b)hn(c) equation 2
In equation 2, g1(b) To gn(b) Are all functions related only to b, h1(c) To hn(c) Are all functions related only to c, g1(b)h1(c)、g2(b)h2(c)……gn(b)hn(c) Are target items of the aggregation rule, g1(b)、h1(c)、g2(b)、h2(c) And the like are data items in the aggregation rule. gi(b) Or hi(c) May be in the form of wi(x)＝x2+1-logx, wherein i is 1 to n.
The proof procedure of equation 2 is as follows:
from the binary taylor expansion, the following equation 3 can be obtained:
in formula 3, R (b, c) is a remainder, n is the number of terms of the function f (b, c), and p is 0 to n, and is negligible when n is sufficiently large, so that formula 2 can be obtained.
For example, for functions
A binary taylor expansion is performed at the (0,0) point, as shown in equation 4:
wherein D is a Jacobian matrix which is a first-order partial derivative matrix, H is a Hessian matrix which is a second-order partial derivative matrix,
is the vertical quantity formed by data b and data c, [ b, c]For the horizontal vector formed by data b and data c, R (b, c) is a quadratic residue, which can be directly truncated.
The sample data in this embodiment is data unique to each participant. For example, where the participant is a bank, the sample data may be an account run of the bank user, and where the participant is a hospital, the sample data may be sample data of a patient.
Step S220, acquiring a first mask, encrypting the data item of each participant according to the first mask, and calculating a first calculation item of the target item according to the encrypted data item.
The first mask in this embodiment is a parameter, such as a number, required for the parties B and C to perform cryptographic calculation on their own data. Each participant calculates only the data items belonging to him, for example, in the case of i 1 to n, participant B only calculates gi(b) Performing encryption calculation, and the participator C only performs the encryption calculation on hi(c) Carry out an encryption meterAnd (4) calculating. And each participant sends the result after the encryption calculation to the initiator, and the initiator calculates the result after the encryption calculation to obtain a first data item, wherein the first data item is not equal to the target item because the first data item comprises the first mask. The calculation of the encrypted data of each participant by the initiator can be at least one of addition, subtraction, multiplication and division.
Further, in a case that the aggregation rule includes a plurality of target items, each participant may have a plurality of data items, at this time, a plurality of first masks may be obtained, the first masks correspond to the data items one to one, the participant B and the participant C respectively perform calculation on the data items corresponding to the first masks according to each first mask, and a sum of all encrypted data items constitutes a first calculation item.
And step S230, acquiring a second calculation item of the target item according to the target item and the first calculation item, and acquiring an encryption calculation result of each participant on the second calculation item.
After the first calculation item is obtained, a second calculation item can be obtained according to the difference value between the target item and the first calculation item, for the second calculation item, each participant performs encryption calculation again on data only related to the participant, and sends the calculated encryption results to the initiator respectively.
The second calculation term calculated from the target term and the first calculation term may be one term or multiple terms. For example, when the second calculation item is the sum of the encrypted data items, there is only one second calculation item that needs to be calculated by the participating parties B and C, and at this time, the participating parties B and C perform one operation to complete the calculation of the second calculation item. Under the condition that the second calculation item still needs to calculate the product of the encrypted data items, the second calculation item comprises a plurality of items, and at the moment, the participator B and the participator C need to perform a plurality of operations to complete the second calculation item.
Step S240, obtaining a data aggregation result of the participating party according to the encryption calculation results of the first calculation item and the second calculation item, wherein the second calculation item is complementary to the first calculation item, so that the first mask is eliminated.
After obtaining the first computation item and the second computation item, the initiator may sum the first computation item and the second computation item to obtain a target item, and then obtain a computation result of the aggregation rule based on the target item. Since the first computation term is complementary to the second computation term, the result obtained from the first computation term and the second computation term does not include the first mask.
Through the steps S210 to S240, the present application splits the aggregation rule of the initiator into the sum of a plurality of function products based on taylor' S theorem, each participant performs encryption calculation on data items only related to itself according to the first mask, and the initiator obtains the data aggregation result of the participant according to the result after the encryption calculation.
Further, the preset aggregation rule can be a function of any complex situation, so that the problem that the applicable scene of the special algorithm for carrying out federal calculation in the related technology is limited is solved, and the scene applicability of the PSA algorithm is improved.
In some embodiments, fig. 3 is a flowchart of a method for calculating a first calculation item according to an embodiment of the present application, and as shown in fig. 3, the method includes the following steps:
in step S310, a second mask is obtained.
In order to improve privacy protection of sample data of the participants, when the first calculation item is calculated, each participant can obtain the second mask to calculate sample data of the participant.
And step S320, controlling the first participant to perform first encryption calculation on the encrypted data item of the first participant according to the second mask, and controlling the second participant to perform second encryption calculation on the encrypted data item of the second participant according to the second mask.
In this embodiment, a case where there are only two parties is discussed, where the first party performs encryption calculation again on the encrypted data item by itself according to the second mask, and similarly, the second party also performs encryption again on the encrypted data item according to the second mask.
It should be noted that the first encryption calculation in the present embodiment corresponds to the second encryption calculation, for example, in the case where the first encryption calculation is to add the second mask, the second encryption calculation is to subtract the second mask, and in the case where the first encryption calculation is to multiply the second mask, the second encryption calculation is to divide the second mask, so that the last first calculation item does not include the second mask.
Step S330, a first calculation item of the target item is obtained according to the data item after the first encryption calculation and the data item after the second encryption calculation.
In this embodiment, the first participant and the second participant may send the encryption result calculated according to the second mask to the initiator, and then the initiator obtains the first calculation item according to the result, specifically, the initiator may perform multiplication or summation on the encryption results of the participants to implement the first calculation item.
Through the steps S310 to S330, in this embodiment, the first party and the second party encrypt respective data items again according to the second mask, and send the data items encrypted again to the initiator, so that privacy protection of their own data is optimized.
Further, in some embodiments, since the preset aggregation rule may be split into a plurality of target items according to taylor expansion, for each target item, a corresponding first mask may be obtained, for example, for different gi(b)hi(c) A corresponding first mask k may be obtainediSo that the first and second parties are respectively paired with gi(b) And hi(c) And (4) calculating, wherein i is 1-n. Therefore, in this embodiment, a plurality of first masks need to be obtained, where the first masks correspond to the first calculation items one to one, and each participant calculates the data items in the plurality of target items in sequence according to each first mask, so as to obtain the first calculation items in all the target items, thereby further improving privacy of sample data in the participantsAnd (4) protecting.
After the plurality of first calculation items are obtained, second calculation items are obtained according to all the target items and the plurality of first calculation items, then encryption calculation results of each participant on the second calculation items are obtained, and finally data aggregation results of the participants are obtained according to the encryption calculation results of the plurality of first calculation items and the second calculation items.
In some embodiments, fig. 4 is a flowchart of a method for calculating a second calculation item according to an embodiment of the present application, and as shown in fig. 4, the method includes:
in step S410, a third mask is obtained.
Similarly, in order to strengthen the protection of the own data, when the second calculation item is obtained, the third mask is obtained to calculate the own data.
And step S420, controlling the first party to perform third encryption calculation on the sample data related to the first party in the second calculation item according to the third mask, and controlling the second party to perform fourth encryption calculation on the sample data related to the second party in the second calculation item according to the third mask.
The third encryption calculation corresponds to the fourth encryption calculation, for example, in the case where the third encryption calculation is to add the third mask, the fourth encryption calculation is to subtract the third mask, and in the case where the third encryption calculation is to multiply the third mask, the fourth encryption calculation is to divide the third mask, so that the final second calculation item does not include the third mask.
Step S430, obtaining the encryption calculation result of the second calculation item according to the results of the third encryption calculation and the fourth encryption calculation.
In this embodiment, the first participant and the second participant may send the encryption result calculated according to the third mask to the initiator, and then the initiator obtains the second calculation item according to the result, specifically, the initiator may perform multiplication or summation on the encryption results of the participants to implement the second calculation item.
Through the above steps S410 to S430, in this embodiment, the first party and the second party encrypt their respective data items again according to the third mask, and then send the encrypted data items to the initiator, so as to further optimize privacy protection on their own data.
In some embodiments, the manner of obtaining the first mask may be various, and an initial mask may be randomly generated by one of the participants, and in the case of confirmation of all the participants, the initial mask may be used as the first mask, and may also be, for federal calculation of only two participants, obtaining the first participant to determine the first reference mask, obtaining the second participant to determine the second reference mask, and then calculating the first mask according to the first reference mask and the second reference mask, where the calculation may be in a form of weighted summation of the first reference mask and the second reference mask, or in a form of product, difference, quotient, remainder, or the like, and may also perform a logical operation, such as an exclusive or operation, on the first reference mask and the second reference mask. Further, the second mask and the third mask in the present application may be generated by any one of the above manners. It should be noted that the first mask, the second mask, and the third mask are not known by the initiator, so as to protect data privacy of the participant. In the embodiment, a plurality of methods for generating masks are provided so as to adapt to different scene requirements.
The embodiments of the present application are described and illustrated below by means of preferred embodiments.
The aggregation rule of the initiator is shown in formula 5, and this embodiment includes an initiator a and two participants, specifically, a first participant B and a second participant C:
equation 5 where f (b, c) ═ b + c + bc
Wherein B is data of the first participant B, and C is data of the participant C, the data aggregation steps are as follows:
s510, the first party B and the second party C jointly determine a first mask k;
s520, the first party B and the second party C share the result of (B-k) (C-k) to the organization A in a PSA mode, wherein (B-k) (C-k) is a target item, B and C are data items respectively, and the specific mode is as follows:
s521, the first party B and the second party C jointly determine a second mask r;
s522, the first party B sends (B-k) xr to the initiator A, and the second party C sends (C-k)/r to the initiator A, wherein (B-k) xr is an encryption result obtained after the first encryption calculation, and (C-k)/r is an encryption result obtained after the second encryption calculation;
s523, the initiator a calculates (b-k) × r × (c-k)/r ═ b-k (c-k), thereby obtaining a first calculation term;
s530, the first party B and the second party C share (k +1) (B + C) -k in a PSA mode2To initiator A, where (k +1) (c + c) -k2The specific way of calculating the second calculation term is that the second calculation term is obtained according to the target term and the first calculation term:
s531, the first participant B and the second participant C jointly determine a third mask t;
s532, the first party B sends (k +1) B-k2+ t to the initiator A, the second participant C sends (k +1) C-t to the initiator A, where (k +1) b-k2+ t is the encryption result after the third encryption calculation, and (k +1) c-t is the encryption result after the fourth encryption calculation;
s533, the initiator A calculates (k +1) b-k2+t+(k+1)c-t＝(k+1)(b+c)-k2Thereby obtaining a calculation result of the second calculation item;
s540, the initiator A calculates the sum of the first calculation item and the second calculation item, namely (b-k) (c-k) + (k +1) (b + c) -k2＝bc-k(b+c)+k2+(k+1)(b+c)-k2B + c + bc, so as to finally obtain a data clustering result corresponding to the aggregation rule.
In the calculation process of step S510 to step S540, the results obtained by the initiator a include the following formulas 6 and 7:
eqn1 ═ b-k (c-k) formula 6
eqn2＝(k+1)(b+c)-k2Equation 7
In formula 6 and formula 7, eqn1 represents the calculation result of the first calculation item, eqn2 represents the calculation result of the second calculation item, and obviously, the initiator a obtains two equations, but three unknowns, B, C, and k, are to be solved, so the initiator a cannot obtain the results of B and C, and thus the data privacy security of the first participant B and the second participant C can be protected.
In this embodiment, the aggregation rule is split into function products, the function products are calculated through a plurality of masks, and finally, an aggregation result is obtained while the masks are eliminated based on the first encryption calculation, the second encryption calculation, the third encryption calculation and the fourth encryption calculation which correspond to each other, so that the efficiency of an aggregation algorithm is improved, and the privacy of the sample data of the participants is also improved.
Further, in some embodiments, it is assumed that the aggregation rule is disassembled to obtain a plurality of function products, as shown in formula 2, at this time, the method of data aggregation specifically includes:
s610, the first participant B and the second participant C jointly determine a first mask k1；
S620, the first party B shares with the second party C through PSA (g)1(b)-k1)(h1(c)-k1) Results of (1) to mechanism A, wherein g1(b)h1(c)、g2(b)h2(c)、……、gn(b)hn(c) Are all target items, and therefore there are multiple target items in this embodiment, each of which includes gi(b) And hi(c)，gi(b) And hi(c) Are all functions, and gi(b) As a function of sample data relating to the first party B only, hi(c) As a function only related to the sample data of the second participant C, the specific way is:
s621, the first party B and the second party C jointly determine a second mask r1；
S622, the first party B sends (g)1(b)-k1)×r1To the initiator A, the second participant C sends (h)1(c)-k1)/r1To initiator A, where g1(b)-k1Pairing data items g according to a first mask for a first participant1(b) The result of the cryptographic calculation of h1(c)-k1Pairing data items h for the second participant according to the first mask1(c) The result of the cryptographic calculation of，(g1(b)-k1)×r1For the encryption result obtained after the first encryption calculation, (h)1(c)-k1)/r1Is the result obtained after the second encryption calculation;
s623, the initiator A acquires the calculation result of the first calculation item, (g)1(b)-k1)×r1×(h1(c)-k1)/r1＝(g1(b)-k1)(h1(c)-k1)；
S630, obtaining a plurality of first masks kiAnd a second mask riRepeating the above steps S621 to S623 until all (g) values are calculatedi(b)-ki)(hi(c)-ki) I is 1 to n, i.e., there are a plurality of first calculation terms in the present embodiment;
s640, sharing by PSA between the first party B and the second party C
Wherein the content of the first and second substances,
is a second calculation item, and there is only one second calculation item in this embodiment, and the specific way to obtain the second calculation item is:
s641, the first participant B and the second participant C jointly determine a third mask t;
s642, first Party B sends
Sending to initiator A, second participant C
To the originator a, wherein,
for addition after third cryptographic calculationAs a result of the encryption, the user can,
the encryption result after the fourth encryption calculation;
s643, the initiator a obtains a calculation result of the second calculation item, specifically as follows:
s650, the initiator A calculates the sum of the first calculation item and the second calculation item, so as to obtain a data clustering result corresponding to the aggregation rule, and the method specifically comprises the following steps:
in the calculation process of step S610 to step S650, the result obtained by the initiator a includes the following formula group 8 and formula 9:
eqn1i＝(gi(b)-ki)(hi(c)-ki) Formula set 8
In equation set 8, eqn1iThe calculation result of the first calculation item obtained in each step S620 is indicated, and eqn2 the calculation result of the second calculation item is indicated.
Step S620 is always executed n times, so that formula 8 and formula 9 have n +1 equations, and b, c, k are included1,k2,…,knAnd n +2 total unknown numbers, so that the initiator A cannot acquire the results of B and C, and the data privacy security of the first participant B and the second participant C can be protected.
In the embodiment, an algorithm which can be suitable for a complex scene is obtained based on a simple PSA algorithm, the complex operation problem of two-party two-data aggregation is achieved through a special algorithm, and compared with a general algorithm, the algorithm in the embodiment does not need a complex framework and a large amount of data calculation, so that the calculation efficiency is greatly improved. Further, compared with the method of performing aggregation calculation by using a unit-divided dedicated algorithm, in this embodiment, since a plurality of masks are used, the calculation process is safer and more reliable, and there is no risk of privacy disclosure.
It should be noted that the steps illustrated in the above-described flow diagrams or in the flow diagrams of the figures may be performed in a computer system, such as a set of computer-executable instructions, and that, although a logical order is illustrated in the flow diagrams, in some cases, the steps illustrated or described may be performed in an order different than here. For example, the order of computation of the first computation term and the second computation term may be exchanged.
The method embodiments provided in the present application may be executed in a terminal, a computer or a similar computing device. Taking the operation on the terminal as an example, fig. 5 is a hardware structure block diagram of the terminal of the data aggregation method according to the embodiment of the present application. As shown in fig. 5, the terminal 50 may include one or more processors 502 (only one is shown in fig. 5) (the processor 502 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 504 for storing data, and optionally may also include a transmission device 506 for communication functions and an input-output device 508. It will be understood by those skilled in the art that the structure shown in fig. 5 is only an illustration and is not intended to limit the structure of the terminal. For example, terminal 50 may also include more or fewer components than shown in FIG. 5, or have a different configuration than shown in FIG. 5.
The memory 504 can be used for storing control programs, for example, software programs and modules of application software, such as the control program corresponding to the method for data aggregation in the embodiment of the present application, and the processor 502 executes various functional applications and data processing by running the control programs stored in the memory 504, that is, implementing the method described above. The memory 504 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 504 may further include memory located remotely from processor 502, which may be connected to terminal 50 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The transmission device 506 is used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the terminal 50. In one example, the transmission device 506 includes a Network adapter (NIC) that can be connected to other Network devices through a base station to communicate with the internet. In one example, the transmission device 506 may be a Radio Frequency (RF) module, which is used for communicating with the internet in a wireless manner.
The present embodiment further provides a data aggregation device, which is used to implement the foregoing embodiments and preferred embodiments, and the description of the data aggregation device is omitted here. As used hereinafter, the terms "module," "unit," "subunit," and the like may implement a combination of software and/or hardware for a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
Fig. 6 is a block diagram of a data aggregation device according to an embodiment of the present application, and as shown in fig. 6, the device includes an obtaining module 61, a first calculating module 62, a second calculating module 63, and an aggregation module 64:
the obtaining module 61 is configured to obtain an aggregation rule of an initiator in federated calculation, and split the aggregation rule into a sum of a plurality of target items according to taylor expansion, where each target item is composed of a plurality of data items, and each data item corresponds to sample data of only one participant;
a first calculation module 62, configured to obtain a first mask, encrypt the data item of each participant according to the first mask, and calculate a first calculation item of the target item according to the encrypted data item;
the second calculation module 63 is configured to obtain a second calculation item of the target item according to the target item and the first calculation item, and obtain an encryption calculation result of each participant on the second calculation item;
and an aggregation module 64, configured to obtain a data aggregation result of the participant according to an encryption calculation result of the first calculation item and a second calculation item, where the second calculation item is complementary to the first calculation item, so that the first mask is eliminated.
The obtaining module 61 splits the aggregation rule of the initiator into the sum of a plurality of function products based on taylor's theorem, the first calculating module 62 and the second calculating module 63 perform encryption calculation on data items only related to the corresponding participants according to the first mask for each participant, and the aggregation module 64 obtains the data aggregation result of the participants according to the result after the encryption calculation.
Further, the first calculation module 62 is further configured to: acquiring a second mask; controlling the first participant to perform first encryption calculation on the encrypted data item according to the second mask; controlling a second party to perform second encryption calculation on the encrypted data item according to a second mask, wherein the first encryption calculation corresponds to the second encryption calculation; and acquiring a first calculation item of the target item according to the data item after the first encryption calculation and the data item after the second encryption calculation. In this embodiment, the first party and the second party encrypt respective data items again according to the second mask, and send the data items encrypted again to the initiator, so that privacy protection of their own data is optimized.
Further, the second calculating module 63 is further configured to: acquiring a third mask; controlling the first participant to perform third encryption calculation on sample data related to the first participant in the second calculation item according to the third mask; controlling a second participant to perform fourth encryption calculation on sample data related to the second participant in the second calculation item according to a third mask, wherein the third encryption calculation corresponds to the fourth encryption calculation; and acquiring the encryption calculation result of the second calculation item according to the results of the third encryption calculation and the fourth encryption calculation. In this embodiment, the first party and the second party encrypt their respective data items again according to the third mask, and then send the encrypted data items to the initiator, thereby further optimizing privacy protection of their own data.
The above modules may be functional modules or program modules, and may be implemented by software or hardware. For a module implemented by hardware, the modules may be located in the same processor; or the modules can be respectively positioned in different processors in any combination.
The present embodiment also provides an electronic device comprising a memory having a computer program stored therein and a processor configured to execute the computer program to perform the steps of any of the above method embodiments.
Optionally, the electronic apparatus may further include a transmission device and an input/output device, wherein the transmission device is connected to the processor, and the input/output device is connected to the processor.
Optionally, in this embodiment, the processor may be configured to execute the following steps by a computer program:
s1, acquiring an aggregation rule of an initiator in federal calculation, and splitting the aggregation rule into the sum of a plurality of target items according to Taylor expansion, wherein each target item is composed of a plurality of data items, and each data item corresponds to sample data of only one participant.
S2, obtaining a first mask, encrypting the data item of each participant according to the first mask, and calculating the first calculation item of the target item according to the encrypted data item.
And S3, acquiring a second calculation item of the target item according to the target item and the first calculation item, and acquiring the encryption calculation result of each participant on the second calculation item.
S4, obtaining a data aggregation result of the participants according to the encryption calculation results of the first calculation item and the second calculation item, wherein the second calculation item is complementary to the first calculation item, so that the first mask is eliminated.
It should be noted that, for specific examples in this embodiment, reference may be made to examples described in the foregoing embodiments and optional implementations, and details of this embodiment are not described herein again.
In addition, in combination with the method for data aggregation in the foregoing embodiments, the present application embodiment may provide a storage medium to implement. The storage medium having stored thereon a computer program; the computer program, when executed by a processor, implements the method of data aggregation of any of the above embodiments.
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method of data aggregation, comprising:
acquiring an aggregation rule of an initiator in federated calculation, and splitting the aggregation rule into the sum of a plurality of target items according to Taylor expansion, wherein each target item is composed of a plurality of data items, and each data item corresponds to sample data of only one participant;
acquiring a first mask, encrypting the data item of each participant according to the first mask, and calculating a first calculation item of the target item according to the encrypted data item;
acquiring a second calculation item of the target item according to the target item and the first calculation item, and acquiring an encryption calculation result of each participant on the second calculation item;
and acquiring a data aggregation result of the participant according to the encryption calculation results of the first calculation item and the second calculation item, wherein the second calculation item is complementary to the first calculation item, so that the first mask is eliminated.
2. The method of claim 1, wherein obtaining a first mask, encrypting the data item of each participant according to the first mask, and wherein calculating a first calculation item for the target item according to the encrypted data item comprises:
acquiring a second mask;
controlling the first participant to perform first encryption calculation on the encrypted data item according to the second mask;
controlling a second party to perform second encryption calculation on the encrypted data item according to the second mask, wherein the first encryption calculation corresponds to the second encryption calculation;
and acquiring a first calculation item of the target item according to the data item after the first encryption calculation and the data item after the second encryption calculation.
3. The method according to claim 2, wherein after obtaining the first calculation item of the target item from the first encryption-calculated data item and the second encryption-calculated data item, the method comprises:
acquiring a plurality of first masks, sequentially calculating data items in the target items according to the first masks, and acquiring the first calculation items in all the target items, wherein the first masks are in one-to-one correspondence with the first calculation items.
4. The method of data aggregation according to claim 1, wherein obtaining a second computation item of the target item from the target item and the first computation item, and obtaining an encryption computation result of each participant on the second computation item comprises:
acquiring a third mask;
controlling the first participant to carry out third encryption calculation on sample data related to the first participant in the second calculation item according to the third mask;
controlling a second participant to perform fourth encryption calculation on sample data related to the second participant in the second calculation item according to the third mask, wherein the third encryption calculation corresponds to the fourth encryption calculation;
and acquiring the encryption calculation result of the second calculation item according to the results of the third encryption calculation and the fourth encryption calculation.
5. The method of data aggregation according to claim 1, wherein obtaining the first mask comprises:
acquiring a first reference mask determined by a first party in the federal calculation and a second reference mask determined by a second party in the federal calculation;
calculating the first mask according to the first reference mask and the second reference mask.
6. The equipment for data aggregation is characterized by comprising an acquisition module, a first calculation module, a second calculation module and an aggregation module:
the obtaining module is used for obtaining an aggregation rule of an initiator in federated calculation, and splitting the aggregation rule into the sum of a plurality of target items according to Taylor expansion, wherein each target item is composed of a plurality of data items, and each data item corresponds to sample data of only one participant;
the first calculation module is configured to obtain a first mask, encrypt the data item of each participant according to the first mask, and calculate a first calculation item of the target item according to the encrypted data item;
the second calculation module is used for acquiring a second calculation item of the target item according to the target item and the first calculation item and acquiring an encryption calculation result of each participant on the second calculation item;
the aggregation module is configured to obtain a data aggregation result of the participant according to an encryption calculation result of the first calculation item and the second calculation item, where the second calculation item is complementary to the first calculation item, so that the first mask is eliminated.
7. The data aggregation device of claim 6, wherein the first computing module is further configured to:
acquiring a second mask;
controlling the first participant to perform first encryption calculation on the encrypted data item according to the second mask;
controlling a second party to perform second encryption calculation on the encrypted data item according to the second mask, wherein the first encryption calculation corresponds to the second encryption calculation;
and acquiring a first calculation item of the target item according to the data item after the first encryption calculation and the data item after the second encryption calculation.
8. The data aggregation device of claim 6, wherein the second computing module is further configured to:
acquiring a third mask;
controlling the first participant to carry out third encryption calculation on sample data related to the first participant in the second calculation item according to the third mask;
controlling a second participant to perform fourth encryption calculation on sample data related to the second participant in the second calculation item according to the third mask, wherein the third encryption calculation corresponds to the fourth encryption calculation;
and acquiring the encryption calculation result of the second calculation item according to the results of the third encryption calculation and the fourth encryption calculation.
9. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, and the processor is configured to execute the computer program to perform the method of data aggregation of any of claims 1 to 5.
10. A storage medium, in which a computer program is stored, wherein the computer program is arranged to perform the method of data aggregation of any one of claims 1 to 5 when executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011578538.2A CN112632611A (en) | 2020-12-28 | 2020-12-28 | Method, apparatus, electronic device, and storage medium for data aggregation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011578538.2A CN112632611A (en) | 2020-12-28 | 2020-12-28 | Method, apparatus, electronic device, and storage medium for data aggregation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112632611A true CN112632611A (en) | 2021-04-09 |
Family
ID=75326108
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011578538.2A Pending CN112632611A (en) | 2020-12-28 | 2020-12-28 | Method, apparatus, electronic device, and storage medium for data aggregation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112632611A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113987594A (en) * | 2021-10-26 | 2022-01-28 | 深圳前海微众银行股份有限公司 | Block chain signature management method and device |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150324607A1 (en) * | 2014-05-07 | 2015-11-12 | Hush Hush | Methods and systems for obfuscating sensitive information in computer systems |
| CN110262819A (en) * | 2019-06-04 | 2019-09-20 | 深圳前海微众银行股份有限公司 | A kind of the model parameter update method and device of federal study |
| CN110826096A (en) * | 2019-10-29 | 2020-02-21 | 杭州趣链科技有限公司 | Data source sharing method based on block chain |
| CN111125735A (en) * | 2019-12-20 | 2020-05-08 | 支付宝(杭州)信息技术有限公司 | Method and system for model training based on private data |
| CN111178547A (en) * | 2020-04-10 | 2020-05-19 | 支付宝(杭州)信息技术有限公司 | Method and system for model training based on private data |
| CN111401483A (en) * | 2020-05-15 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Sample data processing method and device and multi-party model training system |
| CN111898769A (en) * | 2020-08-17 | 2020-11-06 | 中国银行股份有限公司 | Method and system for establishing user behavior period model based on horizontal federal learning |
| US20200358599A1 (en) * | 2019-05-07 | 2020-11-12 | International Business Machines Corporation | Private and federated learning |
| CN112101579A (en) * | 2020-11-18 | 2020-12-18 | 杭州趣链科技有限公司 | Federal learning-based machine learning method, electronic device, and storage medium |
-
2020
- 2020-12-28 CN CN202011578538.2A patent/CN112632611A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150324607A1 (en) * | 2014-05-07 | 2015-11-12 | Hush Hush | Methods and systems for obfuscating sensitive information in computer systems |
| US20200358599A1 (en) * | 2019-05-07 | 2020-11-12 | International Business Machines Corporation | Private and federated learning |
| CN110262819A (en) * | 2019-06-04 | 2019-09-20 | 深圳前海微众银行股份有限公司 | A kind of the model parameter update method and device of federal study |
| CN110826096A (en) * | 2019-10-29 | 2020-02-21 | 杭州趣链科技有限公司 | Data source sharing method based on block chain |
| CN111125735A (en) * | 2019-12-20 | 2020-05-08 | 支付宝(杭州)信息技术有限公司 | Method and system for model training based on private data |
| CN111178547A (en) * | 2020-04-10 | 2020-05-19 | 支付宝(杭州)信息技术有限公司 | Method and system for model training based on private data |
| CN111401483A (en) * | 2020-05-15 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Sample data processing method and device and multi-party model training system |
| CN111898769A (en) * | 2020-08-17 | 2020-11-06 | 中国银行股份有限公司 | Method and system for establishing user behavior period model based on horizontal federal learning |
| CN112101579A (en) * | 2020-11-18 | 2020-12-18 | 杭州趣链科技有限公司 | Federal learning-based machine learning method, electronic device, and storage medium |
Non-Patent Citations (4)
| Title |
|---|
| CHEN, JUQIN等: "A privacy protection scheme based on certificateless aggregate signcryption and masking random number in smart grid", PROCEEDINGS OF THE 2016 4TH INTERNATIONAL CONFERENCE ON MECHANICAL MATERIALS AND MANUFACTURING ENGINEERING * |
| 周俊;方国英;吴楠;: "联邦学习安全与隐私保护研究综述", 西华大学学报(自然科学版), vol. 39, no. 04, 10 July 2020 (2020-07-10) * |
| 袁颖等: "基于数据持有性证明的完整性验证技术综述", 计算机工程与应用, vol. 55, no. 18 * |
| 陈兵等: "联邦学习安全与隐私保护综述", 南京航空航天大学学报, vol. 52, no. 5, 10 November 2020 (2020-11-10) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113987594A (en) * | 2021-10-26 | 2022-01-28 | 深圳前海微众银行股份有限公司 | Block chain signature management method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111512589B (en) | Method for fast secure multiparty inner product with SPDZ | |
| Xing et al. | Mutual privacy preserving $ k $-means clustering in social participatory sensing | |
| US11201734B2 (en) | Method and system for fault tolerant and secure multiparty computation with SPDZ | |
| EP4235550A1 (en) | Blockchain-implemented method and system | |
| CN111723404B (en) | Method and device for jointly training business model | |
| CN110084063B (en) | Gradient descent calculation method for protecting private data | |
| CN109886029B (en) | Polynomial expression based privacy protection set intersection calculation method and system | |
| CN114595835B (en) | Model training method and device based on federal learning, equipment and storage medium | |
| CN113127916A (en) | Data set processing method, data processing device and storage medium | |
| CN111104968B (en) | Safety SVM training method based on block chain | |
| CN112632611A (en) | Method, apparatus, electronic device, and storage medium for data aggregation | |
| Saddam et al. | A lightweight image encryption and blowfish decryption for the secure internet of things | |
| CN114448598A (en) | Ciphertext compression method, ciphertext decompression method, device, equipment and storage medium | |
| CN115913537A (en) | Data intersection method and system based on privacy protection and related equipment | |
| Shin et al. | Securing a local training dataset size in federated learning | |
| CN110598427B (en) | Data processing method, system and storage medium | |
| JP7297131B2 (en) | Distributed Machine Learning Model Training Method, Apparatus, Equipment and Medium | |
| Zhang et al. | Outsourcing hierarchical threshold secret sharing scheme based on reputation | |
| CN115361196A (en) | Service interaction method based on block chain network | |
| Salem et al. | A secure telemedicine electronic platform based on lightweight cryptographic approach | |
| CN114124345A (en) | Data homomorphic encryption reasoning method, device, equipment, system and storage medium | |
| Liu et al. | Secure Judgment of Point and Line Relationship Against Malicious Adversaries and Its Applications | |
| CN113362168A (en) | Risk prediction method and device, storage medium and electronic equipment | |
| Hu et al. | An e-commerce agreement based on the points system of the blockchain and the secure multi-party platform | |
| CN115345307B (en) | Secure convolution neural network reasoning method and system on ciphertext image |
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 |