CN116415978A - Text and travel consumption data analysis method and device based on federal learning and multiparty calculation - Google Patents
Text and travel consumption data analysis method and device based on federal learning and multiparty calculation Download PDFInfo
- Publication number
- CN116415978A CN116415978A CN202310401248.8A CN202310401248A CN116415978A CN 116415978 A CN116415978 A CN 116415978A CN 202310401248 A CN202310401248 A CN 202310401248A CN 116415978 A CN116415978 A CN 116415978A
- Authority
- CN
- China
- Prior art keywords
- model
- aggregation
- secret sharing
- consumption
- passenger
- 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
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000004364 calculation method Methods 0.000 title claims description 27
- 238000007405 data analysis Methods 0.000 title claims description 20
- 230000002776 aggregation Effects 0.000 claims abstract description 150
- 238000004220 aggregation Methods 0.000 claims abstract description 150
- 238000004458 analytical method Methods 0.000 claims abstract description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 abstract description 7
- 238000004891 communication Methods 0.000 description 6
- 238000003062 neural network model Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 239000000654 additive Substances 0.000 description 2
- 230000000996 additive effect Effects 0.000 description 2
- 238000013210 evaluation model Methods 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Theoretical Computer Science (AREA)
- Finance (AREA)
- General Health & Medical Sciences (AREA)
- Bioethics (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present disclosure provides a method and a device for analyzing consumption data of a travel based on federal learning and multiparty computation, which belong to the technical field of travel management, and the method comprises: acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant; obtaining the model weight of the passenger consumption assessment model according to the secret sharing value; obtaining an aggregation model according to the passenger consumption assessment model and the model weight; and sending the aggregation model to the participant for analysis of the travel consumption data. The method solves the problems of data island and data privacy existing in the process of joint prediction of a plurality of participants of the large data consumption of the travel based on federal learning, thereby maximizing the value of releasing the large data consumption of the travel.
Description
Technical Field
The disclosure relates to the technical field of travel management, in particular to a travel consumption data analysis method and device based on federal learning and multiparty calculation.
Background
In recent years, the business of the business travel gradually begins to attach importance to new technologies such as big data, and many enterprises use the big data to establish user figures, discover and predict new wind directions and new demands of business travel, so as to expand diversified, personalized and customized business travel products and services, promote business model innovation of business travel enterprises, and improve competitiveness of business travel enterprises. Meanwhile, a plurality of tourist attraction sites also utilize big data to analyze the resource allocation of the dispatch attraction such as the source place, the consumption habit, the consumption heat period and the like of the passengers, thereby achieving the purposes of matching the supply capacity with the requirements of the passengers, controlling the operation cost and increasing the satisfaction of the passengers.
Traditional big data analysis collects all data to be processed to the same server for processing, however, in reality, due to competition in industry and use approval of data among industries, it is very difficult to integrate data of all parties together. As well as in the travel industry, the relevant data of passengers exist in the form of islands, which are difficult to exert their inherent great value.
Disclosure of Invention
The present disclosure provides a method and apparatus for analyzing travel consumption data based on federal learning and multiparty computation, so as to solve the technical problem that relevant data of passengers recognized by the inventor exist in island form, and the inherent huge value of the data is difficult to develop.
The present disclosure provides a business trip consumption data analysis method based on federal learning and multiparty computation, which is applicable to an aggregation server, and includes:
acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and sending the aggregation model to the participant for analysis of the travel consumption data.
In any of the above solutions, further, the secret sharing value is determined based on a passenger consumption assessment model locally trained by the participant, including:
and the secret sharing value is determined after processing parameters of the passenger consumption assessment model locally trained by the participators based on a secret sharing generation algorithm.
In any of the above solutions, further, the obtaining the model weight of the passenger consumption assessment model according to the secret sharing value includes:
executing an aggregation weight calculation protocol interactively with at least one aggregation server;
and inputting the secret sharing value into the aggregation weight calculation protocol to obtain the model weight of the passenger consumption assessment model.
In any of the above solutions, further, the inputting the secret sharing value into the aggregate weight computing protocol to obtain the model weight of the passenger consumption assessment model includes:
inputting the secret sharing value into the aggregation weight calculation protocol;
calculating gradient direction difference between the passenger consumption assessment model and a preset gradient reference model based on the aggregation weight calculation protocol;
and obtaining the model weight of the passenger consumption assessment model according to the gradient direction difference.
In any of the above solutions, further, the obtaining an aggregate model according to the passenger consumption assessment model and the model weight includes:
executing a model aggregation protocol interactively with at least one aggregation server;
and inputting the passenger consumption assessment model and the model weight into the model aggregation protocol to obtain an aggregation model.
The present disclosure also provides a method for analyzing consumption data of a travel based on federal learning and multiparty computation, which is applicable to a participant, and includes:
determining a secret sharing value according to a locally trained passenger consumption assessment model;
uploading the secret sharing value to an aggregation server;
acquiring an aggregation model returned by the aggregation server to analyze the consumption data of the travel;
and the aggregation model obtains the model weight of the passenger consumption assessment model according to the secret sharing value by the aggregation server, and determines according to the passenger consumption assessment model and the model weight.
In any of the above technical solutions, further comprising:
calculating a hash value corresponding to the aggregation model based on a hash algorithm;
and broadcasting and sending the hash value corresponding to the aggregation model to other participants to verify the uniqueness of the aggregation model.
The present disclosure also provides a business trip consumption data analysis device based on federal learning and multiparty computation, comprising:
the acquisition module is used for acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
the weight determining module is used for obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
the model determining module is used for obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and the sending module is used for sending the aggregation model to the participant so as to analyze the consumption data of the travel.
The present disclosure also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the federal learning and multiparty computing-based method of analysis of travel consumption data when executing the program.
The present disclosure also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the federal learning and multiparty computing-based travel consumption data analysis method.
The beneficial effects of the present disclosure mainly lie in:
1. based on federal learning, the problems of data island and data privacy existing in the process of joint prediction of a plurality of participants of the large consumption data of the travel are solved, so that the value of releasing the large consumption data of the travel is maximized;
2. by determining the model weight of the passenger consumption assessment model, the accuracy of the aggregation model in assessing the customer consumption level based on the travel consumption big data is ensured, and meanwhile, the efficiency of an algorithm is ensured on the premise of protecting the privacy of the customer consumption data.
It is to be understood that both the foregoing general description and the following detailed description are for purposes of example and explanation and are not necessarily limiting of the disclosure. The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate the subject matter of the present disclosure. Meanwhile, the description and drawings are used to explain the principles of the present disclosure.
Drawings
In order to more clearly illustrate the embodiments of the present disclosure or the prior art, the drawings that are required in the detailed description or the prior art will be briefly described, it will be apparent that the drawings in the following description are some embodiments of the present disclosure, and other drawings may be obtained according to the drawings without inventive effort for a person of ordinary skill in the art.
FIG. 1 is a flow diagram of one of the methods for analyzing travel consumption data based on federal learning and multiparty computing provided by the present disclosure;
fig. 2 is a schematic flow chart of step S120 in fig. 1 provided in the present disclosure;
fig. 3 is a schematic flow chart of step S220 in fig. 2 provided in the present disclosure;
fig. 4 is a schematic flow chart of step S130 in fig. 1 provided in the present disclosure;
FIG. 5 is a second flow chart of the method for analyzing travel consumption data based on federal learning and multiparty computing provided by the present disclosure;
FIG. 6 is a third flow chart of the method for analyzing travel consumption data based on federal learning and multiparty computing provided by the present disclosure;
FIG. 7 is a functional block diagram of a business trip consumption data analysis method based on federal learning and multiparty computing provided by the present disclosure;
Detailed Description
The following description of the embodiments of the present disclosure will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all, of the embodiments of the present disclosure.
Based on the embodiments in this disclosure, all other embodiments that a person of ordinary skill in the art would obtain without making any inventive effort are within the scope of protection of this disclosure.
In the description of the present disclosure, it should be noted that the directions or positional relationships indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings, are merely for convenience of describing the present disclosure and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present disclosure. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present disclosure, it should be noted that, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the terms in this disclosure will be understood by those of ordinary skill in the art in the specific context.
Fig. 1 is a flow chart of a method for analyzing travel consumption data based on federal learning and multiparty computation provided by the present disclosure, as shown in fig. 1, the present disclosure provides a method for analyzing travel consumption data based on federal learning and multiparty computation, which is suitable for an aggregation server, and includes:
s110, acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
s120, obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
s130, obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and S140, the aggregation model is sent to the participant to analyze the consumption data of the travel.
The participants in step S110 include a plurality of organization participants such as scenic spots, hotels, and malls radiated in the travel area, each participant obtains desensitized consumption data of the traveler, and models according to the desensitized consumption data of the traveler to obtain a traveler consumption assessment model, which is used for analyzing consumption habits and consumption levels of the traveler, and providing images of the traveler users and consumption pattern wind directions for the plurality of organization participants such as scenic spots, hotels, and malls radiated in the travel area.
The federal learning technology aims at completing joint modeling under the condition that data are not shared, in short, each participant firstly utilizes own data to train to obtain a preliminary model locally, and then all participants jointly establish a general model in a parameter aggregation mode under the multiparty computing technology. Compared with a single model, the aggregation model obtained by using the federal learning technology has stronger generalization capability, and can simultaneously solve two major problems of data island and data privacy leakage.
Specifically, the method and the device are based on federal learning and safe multiparty computing technology, can protect consumption data, model parameters and analysis results of a single model from leakage in a data analysis modeling process, and support model aggregation strategies of various nonlinear and linear operations.
Specifically, in step S110, the passenger consumption assessment model is obtained by training the neural network model, the secret sharing value uploaded by the participant is the secret sharing value corresponding to the weight value of the trained neural network model, the type of the weight value is determined based on the type of the neural network model, and the participant only uploads the secret sharing value corresponding to the weight value, so that the passenger consumption assessment model can be ensured not to be maliciously stolen by the aggregation server, and the transmitted traffic is greatly reduced.
In step S120, according to the secret sharing value, the data distribution of each participant is determined, and the influence degree of the passenger consumption assessment model trained by different participants on the precision of the aggregation model is calculated, so that different aggregation weights are given to the passenger consumption assessment model, the precision of the aggregation model in evaluating the customer consumption level based on the travel consumption big data is ensured, and meanwhile, the efficiency of the algorithm is ensured on the premise of protecting the privacy of the customer consumption data.
In step S130, the aggregation server inputs the passenger consumption assessment model and the model weights into a model aggregation protocol, so as to obtain an aggregation model.
In step S140, the aggregation server sends the secret sharing value corresponding to the aggregation model to the participant, so as to analyze consumption data of the travel, so that the aggregation server can be ensured not to steal the information of the aggregation model.
It can be appreciated that the present disclosure addresses data islanding and data privacy issues that exist in the process of joint prediction of multiple participants for travel consumption big data based on federal learning, thereby maximizing the value of releasing travel consumption big data; by determining the model weight of the passenger consumption assessment model, the accuracy of the aggregation model in assessing the customer consumption level based on the travel consumption big data is ensured, and meanwhile, the efficiency of an algorithm is ensured on the premise of protecting the privacy of the customer consumption data.
In any of the above solutions, further, the secret sharing value is determined based on a passenger consumption assessment model locally trained by the participant, including:
and the secret sharing value is determined after processing parameters of the passenger consumption assessment model locally trained by the participators based on a secret sharing generation algorithm.
Specifically, the passenger consumption assessment model is obtained by training a neural network model, the secret sharing value uploaded by the participant is the secret sharing value corresponding to the weight value of the trained neural network model, the type of the weight value is determined based on the type of the neural network model, and the participant only uploads the secret sharing value corresponding to the weight value.
Secret sharing generation algorithm, namely secret sharing (secret sharing), the splits of data are respectively owned by a group of participants, the participants generally have peer-to-peer roles, and the sharing, splitting and restoring of the participants do not involve encryption and decryption. The advantage of secret sharing is that the cost of data splitting and restoring operations is generally very low, such as additive secret sharing (Additive Secret Sharing), and the splitting and restoring operations are only addition and subtraction, in addition, the selection of the splitting points needs to perform a random number generation, and the operation cost is far lower than that of common encryption and decryption operations.
It can be appreciated that the present disclosure can ensure that the passenger consumption assessment model is not maliciously stolen by the aggregation server, and the traffic of transmission is greatly reduced.
Fig. 2 is a schematic flowchart of step S120 in fig. 1 provided by the present disclosure, as shown in fig. 2, in any one of the foregoing technical solutions, further, according to the secret sharing value, the obtaining a model weight of the passenger consumption assessment model includes:
s210, executing an aggregation weight calculation protocol interactively with at least one aggregation server;
s220, inputting the secret sharing value into the aggregation weight calculation protocol to obtain the model weight of the passenger consumption assessment model.
Fig. 3 is a schematic flowchart of step S220 in fig. 2 provided by the present disclosure, as shown in fig. 3, in any one of the above technical solutions, further, the inputting the secret sharing value into the aggregate weight calculation protocol to obtain a model weight of the passenger consumption assessment model includes:
s310, inputting the secret sharing value into the aggregation weight calculation protocol;
s320, calculating gradient direction difference between the passenger consumption assessment model and a preset gradient reference model based on the aggregation weight calculation protocol;
s330, obtaining the model weight of the passenger consumption assessment model according to the gradient direction difference.
In step S320, the aggregation weight calculation protocol calculates the aggregation weight for the local model of each participant according to the gradient direction difference between the local model of each participant (passenger consumption assessment model) and the preset gradient reference model, that is, the smaller the gradient direction difference between the local model and the aggregation model is, the greater the influence of the local model on the prediction accuracy of the aggregation model is, and thus the greater the obtained weight is. The preset gradient reference model is obtained by adding all local models.
It can be appreciated that the method and the system ensure the accuracy of the aggregate model in evaluating the customer consumption level based on the travel consumption big data by determining the model weight of the passenger consumption evaluation model, and simultaneously ensure the efficiency of the algorithm on the premise of protecting the privacy of the customer consumption data.
Fig. 4 is a schematic flow chart of step S130 in fig. 1 provided by the present disclosure, as shown in fig. 4, in any one of the foregoing technical solutions, further, according to the passenger consumption assessment model and the model weight, an aggregation model is obtained, including:
s410, executing a model aggregation protocol interactively with at least one aggregation server;
s420, inputting the passenger consumption assessment model and the model weight into the model aggregation protocol to obtain an aggregation model.
In step S420, the two aggregation servers performing interaction multiply the weight value of the passenger consumption assessment model, the secret sharing value corresponding to the weight value and the model weight secret sharing value corresponding to the passenger consumption assessment model, and then perform secure multiparty computation on the two parties, thereby obtaining the final aggregation model.
To ensure the accuracy of the aggregate model, the aggregate weights of the local models cannot be exposed to other participants and the aggregate server in plain text. And therefore, the aggregation weight calculation protocol outputs the secret sharing value of each aggregation weight to the two parties respectively.
It can be understood that the invention obtains the aggregation model through the model weight of the passenger consumption evaluation model, ensures the accuracy of the aggregation model in evaluating the customer consumption level based on the big data of the travel consumption, and simultaneously ensures the efficiency of the algorithm on the premise of protecting the privacy of the customer consumption data.
Fig. 5 is a second flow chart of a method for analyzing consumption data of a travel based on federal learning and multiparty computing provided in the present disclosure, as shown in fig. 5, the present disclosure further provides a method for analyzing consumption data of a travel based on federal learning and multiparty computing, which is suitable for participants, and includes:
s510, determining a secret sharing value according to a locally trained passenger consumption assessment model;
s520, uploading the secret sharing value to an aggregation server;
s530, acquiring an aggregation model returned by the aggregation server to analyze the consumption data of the travel;
and the aggregation model obtains the model weight of the passenger consumption assessment model according to the secret sharing value by the aggregation server, and determines according to the passenger consumption assessment model and the model weight.
It can be appreciated that the present disclosure addresses data islanding and data privacy issues that exist in the process of joint prediction of multiple participants for travel consumption big data based on federal learning, thereby maximizing the value of releasing travel consumption big data; by determining the model weight of the passenger consumption assessment model, the accuracy of the aggregation model in assessing the customer consumption level based on the travel consumption big data is ensured, and meanwhile, the efficiency of an algorithm is ensured on the premise of protecting the privacy of the customer consumption data.
In any of the above technical solutions, further comprising:
calculating a hash value corresponding to the aggregation model based on a hash algorithm;
and broadcasting and sending the hash value corresponding to the aggregation model to other participants to verify the uniqueness of the aggregation model.
In order to ensure that the aggregation model sent by the aggregation server to each participant is unique, each participant calculates the hash value of the restored aggregation model by using a hash algorithm. And the hash values of the aggregation model are broadcast and sent among all the participants through the authentication channel, and if all the hash values are the same, the uniqueness of the aggregation model is authenticated. Otherwise, indicating that the aggregation model received by a certain participant is tampered, and stopping the federal learning process at the moment.
It can be appreciated that the method and the device verify the uniqueness of the aggregation model through the hash algorithm, and can timely judge whether the aggregation model is tampered.
FIG. 6 is a third flow chart of the method for analyzing travel consumption data based on federal learning and multiparty computing provided by the present disclosure; as shown in fig. 6, the present disclosure is illustrated with one embodiment.
S1: predictive passenger consumption assessment model for local training of each participant
Each participant, such as a plurality of enterprise participants of scenic spots, hotels, shops, electronic commerce and the like, locally trains a passenger consumption assessment model x by utilizing the safely collected passenger consumption data i 。
S2: generating secret sharing values for a local consumption assessment model
In order to ensure the privacy of each participant model, the participant in this step generates an algorithm ss.share (x i ) To generate secret sharing value { [ x ] i ] 0 ,[x i ] 1 And respectively send to two aggregation servers S 0 And S is 1 。
S3: the aggregation server calculates an aggregation weight
Aggregation server S 0 And S is 1 After receiving the secret sharing values of the local models of the participants, respectively, the two parties interactively execute an aggregate weight calculation protocol which can be according to the local model of each participantGradient direction difference between earth model and aggregation model, calculating aggregation weight mu for each participant's local model i That is, the smaller the gradient direction difference between the local model and the aggregation model, the larger the influence of the model on the prediction accuracy of the aggregation model is, and the larger the obtained weight is.
S4: generating secret sharing values for aggregated weights
To ensure the accuracy of the aggregate model, the aggregate weights of the local models cannot be exposed to other participants and the aggregate server in plain text. Thus, the aggregate weight calculation protocol outputs each aggregate weight mu to both parties respectively i Secret sharing value { [ mu ] i ] 0 ,[μ i ] 1 }。
S5: computing to generate an aggregate model
Aggregation server S 0 And S is 1 Interactive execution model aggregation protocol, both parties respectively input local models x of all parties i And the aggregate weights mu calculated in the previous step i And calculating the aggregation model a.
S6: generating secret sharing values for an aggregated model
In order to ensure that the aggregation model is not maliciously stolen by the aggregation servers, the model aggregation protocol outputs secret sharing values { [ a ] of the aggregation model to the two aggregation servers respectively] 0 ,[a] 1 And the aggregation server can not know any information of the aggregation model, so that the privacy security of the data is protected.
S7: sending secret sharing values for an aggregated model
Aggregation server S 0 And S is 1 Secret sharing value { [ a ] of aggregation model calculated in the previous step] 0 ,[a] 1 And transmitted to the participants via authenticated channels.
S8: restoring aggregation model for each participant
After each participant receives the secret sharing values of the aggregation models from the two aggregation servers, the secret sharing recovery algorithm SS.recon ([ a ]] 0 ,[a] 1 ) Restoring the aggregation modelA plaintext value a of (a).
S9: verifying uniqueness of an aggregation model
In this step, in order to ensure that the aggregation model sent by the aggregation server to each participant is unique, each participant calculates the hash value of the restored aggregation model using a hash algorithm. And the hash values of the aggregation model are broadcast and sent among all the participants through the authentication channel, and if all the hash values are the same, the uniqueness of the aggregation model is authenticated. Otherwise, indicating that the aggregation model received by a certain participant is tampered, and stopping the federal learning process at the moment.
The description of the business trip consumption data analysis device based on federal learning and multiparty calculation provided by the invention is provided below, and the business trip consumption data analysis device based on federal learning and multiparty calculation described below and the business trip consumption data analysis method based on federal learning and multiparty calculation described above can be referred to correspondingly.
Fig. 7 is a schematic block diagram of a method for analyzing travel consumption data based on federal learning and multiparty computation provided in the present disclosure, and as shown in fig. 7, the present disclosure further provides a device for analyzing travel consumption data based on federal learning and multiparty computation, including:
an obtaining module 710, configured to obtain a secret sharing value uploaded by a participant, where the secret sharing value is determined based on a passenger consumption assessment model that is locally trained by the participant;
the weight determining module 720 is configured to obtain a model weight of the passenger consumption assessment model according to the secret sharing value;
the model determining module 730 is configured to obtain an aggregate model according to the passenger consumption assessment model and the model weight;
and a sending module 740, configured to send the aggregation model to the participant for performing travel consumption data analysis.
In any of the above solutions, further, the secret sharing value is determined after processing parameters of a passenger consumption assessment model locally trained by the participant based on a secret sharing generation algorithm.
In any of the above solutions, further, the weight determining module 720 is further configured to:
executing an aggregation weight calculation protocol interactively with at least one aggregation server;
and inputting the secret sharing value into the aggregation weight calculation protocol to obtain the model weight of the passenger consumption assessment model.
In any of the above solutions, further, the weight determining module 720 is further configured to:
inputting the secret sharing value into the aggregation weight calculation protocol;
calculating gradient direction difference between the passenger consumption assessment model and a preset gradient reference model based on the aggregation weight calculation protocol;
and obtaining the model weight of the passenger consumption assessment model according to the gradient direction difference.
In any of the above embodiments, further, the model determining module 730 is configured to:
executing a model aggregation protocol interactively with at least one aggregation server;
and inputting the passenger consumption assessment model and the model weight into the model aggregation protocol to obtain an aggregation model.
The present disclosure also provides an electronic device, which may include: a processor (processor), a communication interface (Communications Interface), a memory (memory) and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus. The processor may invoke logic instructions in the memory to perform a business trip consumption data analysis method based on federal learning and multiparty computing, the method comprising:
acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and sending the aggregation model to the participant for analysis of the travel consumption data.
Further, the logic instructions in the memory described above may be implemented in the form of software functional units and stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of performing the method of analysis of travel consumption data based on federal learning and multiparty computing provided by the methods described above, the method comprising:
acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and sending the aggregation model to the participant for analysis of the travel consumption data.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method of business trip consumption data analysis based on federal learning and multiparty computing provided by the methods described above, the method comprising:
acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and sending the aggregation model to the participant for analysis of the travel consumption data.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present disclosure, and not for limiting the same; although the present disclosure has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present disclosure.
Claims (10)
1. A business trip consumption data analysis method based on federal learning and multiparty computation, which is suitable for an aggregation server, comprising:
acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and sending the aggregation model to the participant for analysis of the travel consumption data.
2. The method of claim 1, wherein the secret sharing value is determined based on a locally trained passenger consumption assessment model of the participant, comprising:
and the secret sharing value is determined after processing parameters of the passenger consumption assessment model locally trained by the participators based on a secret sharing generation algorithm.
3. The method for analyzing consumption data of a travel itinerary based on federal learning and multiparty computing according to claim 1, wherein said obtaining model weights of the passenger consumption assessment model according to the secret sharing values comprises:
executing an aggregation weight calculation protocol interactively with at least one aggregation server;
and inputting the secret sharing value into the aggregation weight calculation protocol to obtain the model weight of the passenger consumption assessment model.
4. The method for analyzing travel consumption data based on federal learning and multiparty computing according to claim 3, wherein said inputting the secret sharing value into the aggregate weight computing protocol, obtaining the model weight of the passenger consumption assessment model, comprises:
inputting the secret sharing value into the aggregation weight calculation protocol;
calculating gradient direction difference between the passenger consumption assessment model and a preset gradient reference model based on the aggregation weight calculation protocol;
and obtaining the model weight of the passenger consumption assessment model according to the gradient direction difference.
5. The method of claim 1, wherein the deriving an aggregate model from the passenger consumption assessment model and the model weights comprises:
executing a model aggregation protocol interactively with at least one aggregation server;
and inputting the passenger consumption assessment model and the model weight into the model aggregation protocol to obtain an aggregation model.
6. A business trip consumption data analysis method based on federal learning and multiparty computing, which is suitable for participants, comprising:
determining a secret sharing value according to a locally trained passenger consumption assessment model;
uploading the secret sharing value to an aggregation server;
acquiring an aggregation model returned by the aggregation server to analyze the consumption data of the travel;
and the aggregation model obtains the model weight of the passenger consumption assessment model according to the secret sharing value by the aggregation server, and determines according to the passenger consumption assessment model and the model weight.
7. The method for analysis of travel consumption data based on federal learning and multiparty computing of claim 6, further comprising:
calculating a hash value corresponding to the aggregation model based on a hash algorithm;
and broadcasting and sending the hash value corresponding to the aggregation model to other participants to verify the uniqueness of the aggregation model.
8. A business trip consumption data analysis device based on federal learning and multiparty computing, comprising:
the acquisition module is used for acquiring a secret sharing value uploaded by a participant, wherein the secret sharing value is determined based on a passenger consumption assessment model trained locally by the participant;
the weight determining module is used for obtaining the model weight of the passenger consumption assessment model according to the secret sharing value;
the model determining module is used for obtaining an aggregation model according to the passenger consumption assessment model and the model weight;
and the sending module is used for sending the aggregation model to the participant so as to analyze the consumption data of the travel.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the federal learning and multiparty computing-based travel consumption data analysis method according to any one of claims 1 to 7 when the program is executed by the processor.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the federal learning and multiparty computing-based travel consumption data analysis method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310401248.8A CN116415978B (en) | 2023-04-15 | 2023-04-15 | Text and travel consumption data analysis method and device based on federal learning and multiparty calculation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310401248.8A CN116415978B (en) | 2023-04-15 | 2023-04-15 | Text and travel consumption data analysis method and device based on federal learning and multiparty calculation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116415978A true CN116415978A (en) | 2023-07-11 |
| CN116415978B CN116415978B (en) | 2024-03-22 |
Family
ID=87049336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310401248.8A Active CN116415978B (en) | 2023-04-15 | 2023-04-15 | Text and travel consumption data analysis method and device based on federal learning and multiparty calculation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116415978B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340453A (en) * | 2020-02-28 | 2020-06-26 | 深圳前海微众银行股份有限公司 | Federal learning development method, device, equipment and storage medium |
| CN111600707A (en) * | 2020-05-15 | 2020-08-28 | 华南师范大学 | Decentralized federal machine learning method under privacy protection |
| CN113642737A (en) * | 2021-08-12 | 2021-11-12 | 广域铭岛数字科技有限公司 | Federal learning method and system based on automobile user data |
| CN114785481A (en) * | 2022-04-13 | 2022-07-22 | 上海阵方科技有限公司 | System for interconnecting safe multi-party calculation and federal learning data |
| CN114819190A (en) * | 2022-06-21 | 2022-07-29 | 平安科技（深圳）有限公司 | Model training method, device, system and storage medium based on federal learning |
| US20220374544A1 (en) * | 2021-05-05 | 2022-11-24 | Jpmorgan Chase Bank, N.A. | Secure aggregation of information using federated learning |
| US20220374762A1 (en) * | 2021-05-18 | 2022-11-24 | International Business Machines Corporation | Trusted and decentralized aggregation for federated learning |
| CN115758127A (en) * | 2021-09-01 | 2023-03-07 | 全球能源互联网研究院有限公司 | Model training system and method based on federal learning |
| CN115775010A (en) * | 2022-11-23 | 2023-03-10 | 国网江苏省电力有限公司信息通信分公司 | Electric power data sharing method based on horizontal federal learning |
| CN115883076A (en) * | 2022-11-16 | 2023-03-31 | 西安邮电大学 | Method, system, medium, device and terminal for protecting privacy of federated learning |
| CN115935136A (en) * | 2022-12-19 | 2023-04-07 | 广州芳禾数据有限公司 | Multi-source data fusion oriented statistical method and device for information of cultural passenger user |
| CN115952539A (en) * | 2022-12-27 | 2023-04-11 | 西安电子科技大学 | Majority-of-offence robust privacy federated learning method, system, device, and medium |
-
2023
- 2023-04-15 CN CN202310401248.8A patent/CN116415978B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340453A (en) * | 2020-02-28 | 2020-06-26 | 深圳前海微众银行股份有限公司 | Federal learning development method, device, equipment and storage medium |
| CN111600707A (en) * | 2020-05-15 | 2020-08-28 | 华南师范大学 | Decentralized federal machine learning method under privacy protection |
| US20220374544A1 (en) * | 2021-05-05 | 2022-11-24 | Jpmorgan Chase Bank, N.A. | Secure aggregation of information using federated learning |
| US20220374762A1 (en) * | 2021-05-18 | 2022-11-24 | International Business Machines Corporation | Trusted and decentralized aggregation for federated learning |
| CN113642737A (en) * | 2021-08-12 | 2021-11-12 | 广域铭岛数字科技有限公司 | Federal learning method and system based on automobile user data |
| CN115758127A (en) * | 2021-09-01 | 2023-03-07 | 全球能源互联网研究院有限公司 | Model training system and method based on federal learning |
| CN114785481A (en) * | 2022-04-13 | 2022-07-22 | 上海阵方科技有限公司 | System for interconnecting safe multi-party calculation and federal learning data |
| CN114819190A (en) * | 2022-06-21 | 2022-07-29 | 平安科技（深圳）有限公司 | Model training method, device, system and storage medium based on federal learning |
| CN115883076A (en) * | 2022-11-16 | 2023-03-31 | 西安邮电大学 | Method, system, medium, device and terminal for protecting privacy of federated learning |
| CN115775010A (en) * | 2022-11-23 | 2023-03-10 | 国网江苏省电力有限公司信息通信分公司 | Electric power data sharing method based on horizontal federal learning |
| CN115935136A (en) * | 2022-12-19 | 2023-04-07 | 广州芳禾数据有限公司 | Multi-source data fusion oriented statistical method and device for information of cultural passenger user |
| CN115952539A (en) * | 2022-12-27 | 2023-04-11 | 西安电子科技大学 | Majority-of-offence robust privacy federated learning method, system, device, and medium |
Non-Patent Citations (1)
| Title |
|---|
| 董业 等: "基于秘密分享和梯度选择的高效安全联邦学习", 计算机研究与发展, no. 10, pages 2241 - 2250 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116415978B (en) | 2024-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113127916B (en) | Data set processing method, data processing method, device and storage medium | |
| CN110084377A (en) | Method and apparatus for constructing decision tree | |
| CN101547095B (en) | Application service management system and management method based on digital certificate | |
| CN108964926A (en) | User trust negotiation establishing method based on two-layer block chain in heterogeneous alliance system | |
| CN110298421A (en) | A kind of online generation, offline generation and the verification method and device of two dimensional code | |
| CN110399735A (en) | Encryption data size relation method of proof, device, equipment and storage medium | |
| CN108650269A (en) | A kind of graded encryption method and system based on intensified learning | |
| CN114971796B (en) | Bidding system based on cloud service platform | |
| CN114138731A (en) | Smart city data sharing method and system based on big data | |
| CN113569263A (en) | Secure processing method and device for cross-private-domain data and electronic equipment | |
| Guo et al. | Slim-FCP: Lightweight-feature-based cooperative perception for connected automated vehicles | |
| Feng et al. | Vulnerability-aware task scheduling for edge intelligence empowered trajectory analysis in intelligent transportation systems | |
| CN102694818B (en) | The online distribution method of online private key and system | |
| Yang et al. | Accountable and verifiable secure aggregation for federated learning in IoT networks | |
| CN114168295A (en) | Hybrid architecture system and task scheduling method based on historical task effect | |
| CN116415978B (en) | Text and travel consumption data analysis method and device based on federal learning and multiparty calculation | |
| Wang et al. | Blockchain-Enabled Lightweight Fine-Grained Searchable Knowledge Sharing for Intelligent IoT | |
| CN116208340A (en) | Trusted data flow platform system method based on privacy calculation and blockchain | |
| CN109977919A (en) | Data processing method, medium, equipment and device based on recognition of face | |
| CN113014540B (en) | Data processing method, device, equipment and storage medium | |
| CN115361196A (en) | Service interaction method based on block chain network | |
| Ahmed-Nacer et al. | Strong authentication for mobile cloud computing | |
| Abbas et al. | IEEE Access Special Section: Security Analytics and Intelligence for Cyber Physical Systems | |
| CN113762970A (en) | Data processing method and device, computer readable storage medium and computer equipment | |
| CN114866268A (en) | Method and device for realizing account intercommunication and electronic equipment |
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 |