CN115238826B - Model training method and device, storage medium and electronic equipment - Google Patents
Model training method and device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN115238826B CN115238826B CN202211121861.6A CN202211121861A CN115238826B CN 115238826 B CN115238826 B CN 115238826B CN 202211121861 A CN202211121861 A CN 202211121861A CN 115238826 B CN115238826 B CN 115238826B
- Authority
- CN
- China
- Prior art keywords
- data
- prediction result
- model
- prediction
- service
- 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.)
- Active
Links
- 238000012549 training Methods 0.000 title claims abstract description 78
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000003860 storage Methods 0.000 title claims abstract description 20
- 238000005457 optimization Methods 0.000 claims abstract description 21
- 238000004590 computer program Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 description 22
- 230000008569 process Effects 0.000 description 16
- 238000010586 diagram Methods 0.000 description 15
- 238000012545 processing Methods 0.000 description 10
- 230000006872 improvement Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 229920001296 polysiloxane Polymers 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000003068 static effect 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
- 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 Engineering & Computer Science (AREA)
- Bioethics (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Computer Security & Cryptography (AREA)
- Computer Hardware Design (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The specification discloses a model training method, a device, a storage medium and electronic equipment, wherein sample business data and privacy data are obtained firstly, the obtained sample business data and privacy data are input into a reference model to be trained to obtain a first prediction result, the sample business data are input into the prediction model to be trained to obtain a second prediction result, and the reference model and the prediction model are jointly trained by taking the minimum deviation between the first prediction result and the second prediction result and the minimum deviation between the first prediction result and label data corresponding to the sample business data as optimization targets, wherein the trained prediction model is applied to business execution.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a method and an apparatus for model training, a storage medium, and an electronic device.
Background
With the continuous development of machine learning, machine learning models are used in various business fields to execute various businesses.
However, in the process of using the model, the private data of the user is usually input into the model, and this process may possibly leak the private data of the user, thereby bringing a certain hidden danger to the information security of the user.
Disclosure of Invention
The present specification provides a method, an apparatus, a storage medium, and an electronic device for model training, which partially solve the above problems in the prior art.
The technical scheme adopted by the specification is as follows:
the present specification provides a method of model training, comprising:
acquiring sample service data and privacy data;
inputting the sample business data and the privacy data into a reference model to be trained to obtain a first prediction result, and inputting the sample business data into the prediction model to be trained to obtain a second prediction result;
and performing joint training on the reference model and the prediction model by taking the minimized deviation between the first prediction result and the second prediction result and the minimized deviation between the first prediction result and the label data corresponding to the sample business data as optimization targets, wherein the trained prediction model is applied to business execution.
Optionally, with the objective of minimizing a deviation between the first prediction result and the second prediction result and minimizing a deviation between the first prediction result and the label data corresponding to the sample service data as optimization objectives, performing joint training on the reference model and the prediction model specifically includes:
and performing joint training on the reference model and the prediction model by taking the minimization of the deviation between the first prediction result and the second prediction result, the minimization of the deviation between the first prediction result and the label data corresponding to the sample business data and the minimization of the deviation between the second prediction result and the label data as optimization targets.
Optionally, the method further comprises:
acquiring service data of a user;
inputting the service data into the prediction model to obtain a prediction result output by the prediction model, wherein the prediction model is obtained by training according to the model training method;
and executing the service according to the prediction result.
Optionally, the acquiring the service data of the user specifically includes:
acquiring initial data required by a user to execute a service;
identifying private data of the user from the initial data;
and eliminating the privacy data from the initial data to obtain service data required by service execution.
The present specification provides an apparatus for model training, comprising:
the acquisition module is used for acquiring sample business data and privacy data;
the input module is used for inputting the sample business data and the privacy data into a reference model to be trained to obtain a first prediction result, and inputting the sample business data into the prediction model to be trained to obtain a second prediction result;
and the training module is used for performing joint training on the reference model and the prediction model by taking the minimized deviation between the first prediction result and the second prediction result and the minimized deviation between the first prediction result and the label data corresponding to the sample business data as optimization targets, wherein the trained prediction model is applied to business execution.
Optionally, the training module is specifically configured to perform joint training on the reference model and the prediction model with optimization objectives of minimizing a deviation between the first prediction result and the second prediction result, minimizing a deviation between the first prediction result and the label data corresponding to the sample business data, and minimizing a deviation between the second prediction result and the label data.
Optionally, the apparatus further comprises:
the application module is used for acquiring the service data of the user; inputting the service data into the prediction model to obtain a prediction result output by the prediction model, wherein the prediction model is obtained by training according to the model training method; and executing the service according to the prediction result.
Optionally, the application module is specifically configured to obtain initial data required by the user to execute the service; identifying private data of the user from the initial data; and eliminating the privacy data from the initial data to obtain service data required by service execution.
The present specification provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the above-described method of model training.
The present specification 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 above-described method of model training when executing the program.
The technical scheme adopted by the specification can achieve the following beneficial effects:
the model training method provided by the specification comprises the steps of firstly obtaining sample business data and privacy data, inputting the obtained sample business data and the obtained privacy data into a reference model to be trained to obtain a first prediction result, inputting the sample business data into the prediction model to be trained to obtain a second prediction result, and performing combined training on the reference model and the prediction model by taking the deviation between the first prediction result and the second prediction result and the deviation between the first prediction result and label data corresponding to the sample business data as optimization targets, wherein the trained prediction model is applied to business execution.
As can be seen from the above method, the prediction model can be trained according to the prediction result output by the reference model by the above model training method. Therefore, when the trained prediction model is used for executing the service, the privacy data of the user does not need to be acquired, and the corresponding service can be accurately executed under the condition that the privacy data of the user is not input, so that the possibility of privacy disclosure of the user is avoided on the premise that the service can be accurately executed, and the information safety of the user is ensured.
Drawings
The accompanying drawings, which are included to provide a further understanding of the specification and are incorporated in and constitute a part of this specification, illustrate embodiments of the specification and together with the description serve to explain the specification and not to limit the specification in a non-limiting sense. In the drawings:
FIG. 1 is a schematic flow chart diagram of a method for model training provided herein;
FIG. 2 is a schematic diagram of model training provided herein;
FIG. 3 is a schematic diagram of an apparatus for model training provided herein;
fig. 4 is a schematic diagram of an electronic device corresponding to fig. 1 provided in the present specification.
Detailed Description
In order to make the objects, technical solutions and advantages of the present disclosure more clear, the technical solutions of the present disclosure will be clearly and completely described below with reference to the specific embodiments of the present disclosure and the accompanying drawings. It is to be understood that the embodiments described are only a few embodiments of the present disclosure, and not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present specification without any creative effort belong to the protection scope of the present specification.
The technical solutions provided by the embodiments of the present description are described in detail below with reference to the accompanying drawings.
Fig. 1 is a schematic flow chart of a method for model training provided in this specification, including the following steps:
s101: and acquiring sample business data and privacy data.
In the present specification, the execution subject of the method for implementing model training may be a device such as a server installed on a business platform, or may also be a terminal device such as a desktop computer or a notebook computer.
At present, in order to improve the performance of a trained model, the private data of a user is often input into the model. Thus, there is a possibility that private data of the user is leaked, and thus the information security of the user cannot be guaranteed.
In order to solve the above mentioned problems, the present specification provides a method for model training, first, a server may obtain sample service data and privacy data, where the sample service data may be data that is generated when a user performs a service and does not relate to user privacy, for example, for a payment service performed by the user, data such as a payment amount of the user, a user account number, information for characterizing a payment method of the user, information of a payment account, and the like may be regarded as service data of the payment service performed by the user. The data such as the age, name, telephone number, and device information of the user can be regarded as the private data of the user.
For another example, for an information recommendation service, data such as a search term input by a user in a search term input box displayed by a client, an information preference tag set by the user (each information preference tag corresponds to one type of information), and a historical browsing record of the user may be regarded as service data of the user, and information such as age, gender, income, and geographic location of the user may be regarded as privacy data of the user.
In this specification, the server may acquire service data generated when the user executes a historical service, and use the acquired service data as sample service data to train the reference model and the prediction model in a subsequent process. The private data corresponding to the sample service data may refer to the private data of the user authorized to obtain the private data of the user.
For a training sample, the sample service data and the privacy data contained in the training sample may belong to the same object, for example, for a service scenario, the service data of a user and the privacy data of the user themselves belong to the user, and then the service data of the user may be used as the sample service data and construct a training sample together with the privacy data of the user themselves.
The sample service data may refer to service data generated by a service that is historically performed once by a user, for example, service data generated by a payment service that is performed once by a user. The sample service data may also refer to service data generated by services performed by the user in a historical time period, or may refer to service data generated by all services performed by the user in a historical time period.
It should be noted that the above-mentioned sample service data and the privacy data of the user may also be artificially and empirically constructed, that is, the sample service data and the privacy data of the user may also not be the service data and the privacy data obtained from the real data source.
S102: and inputting the sample business data and the privacy data into a reference model to be trained to obtain a first prediction result, and inputting the sample business data into the prediction model to be trained to obtain a second prediction result.
In this specification, the reference model and the prediction model may be trained simultaneously in a joint training manner, where the purpose of training the reference model is mainly as follows: it is desirable that the reference model can transfer "knowledge" learned from the sample business data and the private data to the prediction model, so that the trained prediction model can obtain an accurate result even without using the private data of the user. Therefore, the reference model is specifically applied in the model training phase, and the prediction model is applied in the actual use.
Therefore, the respective inputs of the parametric model and the predictive model are different in the training phase. Specifically, after the server acquires the sample service data and the privacy data, the sample service data and the privacy data need to be input into a reference model to be trained, so as to obtain a first prediction result output by the reference model.
The server can directly input the sample business data and the privacy data in one training sample into the reference model to be trained, the reference model to be trained can perform data processing such as coding and feature extraction on the sample business data and the privacy data, and a first prediction result is obtained according to the obtained processed data.
Of course, in this specification, the server may convert the sample service data and the privacy data into a feature vector before inputting them into the reference model, and then input the feature vector obtained by converting the sample service data and the privacy data into the reference model to obtain the first prediction result.
For example, in a scenario of business wind control, the server may input sample business data such as a payment amount of the user, a user account, and the like, and privacy data such as a user age, a user income, a user gender, and the like into the reference model, and the reference model may process the data to obtain a first prediction result, which may be a risk probability of the user performing the payment business, where the greater the risk probability is, the higher the possibility of the user risking performing the payment business this time is, and the smaller the risk probability is, the lower the possibility of the user risking performing the payment business this time is. For another example, for the information recommendation service, the server may input, to the reference model, sample service data such as search terms input by the user, historical browsing records of the user, and information to be recommended, and privacy data such as the age, sex, monthly payment amount, and the like of the user, and the first prediction result obtained by performing data processing on the data by the reference model may be the probability that the user clicks each information to be recommended, predicted by the reference model.
Meanwhile, the server also needs to input the sample service data into the prediction model to be trained to obtain a second prediction result output by the prediction model to be trained. The server may directly input the sample service data into the prediction model, or may convert the sample service data into a feature vector and then input the feature vector into the prediction model to obtain a second prediction result. The second prediction result mentioned here is the same as the first prediction result, and may have different specific forms for different services, and is not described herein again. It should be noted that, in the subsequent service execution process, only the trained prediction model needs to be used, so that the input of the prediction model does not contain the private data of the user in the model training stage or the actual use stage of the model.
Since the reference model and the prediction model differ in input, the reference model and the prediction model will also differ in model structure and model parameters. The model structure of the reference model is generally more complex than that of the prediction model, and the network layer included in the prediction model may be derived from the network layer included in the reference model. It will be appreciated that the function of the predictive model may be derived from the reference model.
S103: and performing joint training on the reference model and the prediction model by taking the minimized deviation between the first prediction result and the second prediction result and the minimized deviation between the first prediction result and the label data corresponding to the sample business data as optimization targets, wherein the trained prediction model is applied to business execution.
Since the reference model and the prediction model are trained simultaneously in a joint training manner in this specification, the server needs to ensure that the deviation between the first prediction result output by the reference model and the tag data corresponding to the sample business data is minimized, and the server aims to ensure that the reference model can better learn the corresponding "knowledge" through the sample business data and the privacy data, and the learning of the corresponding "knowledge" here can be understood as the ability of the reference model to accurately obtain the output result according to the input sample business data and the privacy data by minimizing the deviation between the first prediction result and the tag data corresponding to the sample business data. The other is to ensure that the deviation between the first prediction result output by the reference model and the second prediction result output by the prediction model is minimized, and the aim is to transfer the learned knowledge to the prediction model, that is, the capability of the reference model for accurately obtaining the output result according to the input data and the capability of learning how to obtain the result output by the reference model only by using the sample service data can be learned by minimizing the deviation between the first prediction result and the second prediction result through the prediction model.
Based on this, in this specification, the server may determine the label data corresponding to the sample business data, and train the prediction model with the objective of minimizing the deviation between the first prediction result and the second prediction result, and minimizing the deviation between the second prediction result and the label data corresponding to the sample business data as an optimization objective. The mode of determining the label data corresponding to the sample service data may be manual labeling, or may be automatically labeled by using a server. The training is to automatically adjust model parameters in the model by setting a reasonable optimization target, so that the model can obtain a reasonable and accurate output result according to input data.
Of course, in order to further improve the training effect of the model, in the present specification, the prediction model and the reference model may be jointly trained by introducing an optimization target for minimizing a deviation between the second prediction result output by the prediction model and the above-described label data. That is, the server may perform the joint training on the reference model and the prediction model with the optimization objectives of minimizing a deviation between the first prediction result and the second prediction result, minimizing a deviation between the first prediction result and the label data corresponding to the sample traffic data, and minimizing a deviation between the second prediction result and the label data. The joint training means that model parameters of the reference model and the prediction model are adjusted simultaneously through a set reasonable optimization target, so that the effect of training the reference model and the prediction model together is achieved.
As can be seen from the above method, the prediction model can be trained according to the prediction result output by the reference model by the above model training method. Therefore, when the trained prediction model is used for executing the service, the privacy data of the user does not need to be acquired, and the corresponding service can be accurately executed under the condition that the privacy data of the user is not input, so that the possibility of privacy disclosure of the user is avoided on the premise that the service can be accurately executed, and the information safety of the user is ensured.
In addition, in this specification, the reference model may also be a model trained in advance, the training mode may be that a training sample for training the reference model is first constructed through sample business data and privacy data, and the server trains the reference model according to the constructed training sample by using a conventional training mode such as supervised training.
In this case, one of the main functions of the reference model is to output label data for training the prediction model. Therefore, the server can train the prediction model by minimizing the deviation between the first prediction result output by the reference model and the second prediction result output by the prediction model, and minimizing the deviation between the second prediction result and the label data corresponding to the sample business data.
Specifically, the prediction model to be trained may be trained with reference to the following loss function.
Wherein,may be a loss function between the second prediction and the label data corresponding to the sample traffic data,may be a loss function between the first prediction and the second prediction,as a function of lossThe weight of (a) is determined,as a function of lossThe weight of (c). The loss function between the second prediction result and the label data corresponding to the sample service data, and the loss function between the first prediction result and the second prediction result may be a conventional loss function, such as a cross entropy loss function, and the like, and the adopted loss function is not specifically limited in this specification.
After the server obtains the service data of the user, the service data can be input into the prediction model to obtain the prediction result output by the prediction model, and the service can be executed according to the prediction result.
In order to avoid leakage of the private data of the user, the server may first obtain initial data required by the user to execute the service, and identify the private data of the user from the initial data, and then the server may remove the private data from the initial data, so as to obtain service data required by executing the service, and then execute the corresponding service through the obtained service data.
It is worth emphasizing that in practical applications, the server may execute the business using only the trained resulting predictive model, and not using the reference model.
The following description is made with reference to fig. 2 for a schematic diagram of model training provided in this specification:
the server can input the sample business data and the privacy data into the reference model, obtain a first prediction result output by the reference model, input the sample business data into the prediction model to be trained, obtain a second prediction result, and then adjust the reference model and the prediction model according to the loss between the second prediction result and the first prediction result and the loss between the first prediction result and the label data corresponding to the sample business data, so as to achieve the purpose of simultaneously carrying out combined training on the reference model and the prediction model.
For convenience of explanation, the above model training process will be described in detail with a specific example.
For the wind control service, the server can acquire service data generated by the user performing the payment service historically as sample service data and acquire privacy data of the user.
Then, the server can input the sample business data and the privacy data into a reference model to be trained, and the reference model can obtain a first prediction result by performing data processing on the data, wherein the first prediction result is used for representing the probability of the historical payment business execution risk of the user predicted by the reference model.
Meanwhile, the server can input the sample service data into the prediction model to be trained, and the prediction model can obtain a second prediction result by performing data processing on the sample service data, wherein the second prediction result is used for representing the probability of the user, predicted by the prediction model, of risk in executing the historical payment service.
The server can perform combined training on the reference model and the prediction model by taking the optimization target that the probability of the user performing the historical payment service with risk predicted by the reference model is as close as possible to the label data used for indicating whether the user actually has risk when performing the historical payment service, and the probability of the user performing the historical payment service with risk predicted by the prediction model is as close as possible to the probability of the user performing the historical payment service with risk predicted by the prediction model.
After the training of the prediction model is completed, the prediction model can be applied to the actual wind control business, and the reference model does not need to be used in the actual wind control business. Specifically, when the user executes the payment service, the server may input service data generated in the process of executing the payment service by the user into the trained prediction model, the prediction model may obtain a prediction result according to the service data, the prediction result is used to indicate a probability that the user has a risk when executing the payment service at this time, and the server may perform service wind control on the user according to the prediction result. If the probability exceeds the preset probability threshold value, it is determined that the payment service executed by the user at this time has risks, and then the completion of the final payment stage can be stopped, and prompt information for prompting the user that the payment service at this time has risks is sent to the user.
In addition, in the training process, which kind of privacy data is to be used may be determined manually through practical experience, or may be screened by the server through sample service data of different users and practical service results of different users. The data of gender, age, income and academic degree can be regarded as different types of privacy data.
For a specific screening process, the server may first obtain users that are highly similar in service data but differ significantly in service labels. For example, for a wind-control service, the server may screen out users who are highly similar in payment amount, payment method, and payment time, but have different actual facing risk conditions (i.e., service labels) (i.e., some users do not have risks during payment, and some users do have risks during payment).
After screening out the users, the server may select a part of types from all data types related to the private data to obtain a type set. Then, for each screened user, the server may input the sample service data of the user and the privacy data of the user under the type included in the type set into a pre-trained reference model to obtain a prediction result corresponding to the user.
After the prediction result corresponding to each user is obtained, the server may determine, according to a deviation between the prediction result corresponding to each user and an actual service tag of each user, an accuracy rate of the server in predicting based on the private data of the type included in the type set, and further determine, according to the accuracy rate, which types of private data should be used specifically.
Specifically, if it is determined that the accuracy exceeds the preset threshold, it may be determined that the types included in the type set are reasonable, and then, on this basis, some types may be selected and added to the type set, and then prediction is performed according to the privacy data corresponding to the type set to which the new type is added, and the type set is adjusted again according to the obtained accuracy until the preset condition is satisfied. The preset stopping condition mentioned here may be various, for example, the number of types included in the screened type set exceeds a threshold, or the accuracy of the prediction performed by the server based on the privacy data corresponding to the type set exceeds a set accuracy, and the like, which is not described in detail herein.
Of course, if it is determined that the accuracy does not exceed the preset threshold, at least part of the types in the type set may be removed, some new types may be added, prediction may be performed according to the privacy data corresponding to the type set to which the new types are added, and the type set may be adjusted according to the obtained accuracy until a preset stop condition is satisfied, so as to determine which types of privacy data should be used.
Fig. 3 is a schematic diagram of a model training apparatus provided in the present specification, including:
an obtaining module 301, configured to obtain sample service data and privacy data;
an input module 302, configured to input the sample service data and the privacy data into a reference model to be trained to obtain a first prediction result, and input the sample service data into a prediction model to be trained to obtain a second prediction result;
a training module 303, configured to perform joint training on the reference model and the prediction model with the objective of minimizing a deviation between the first prediction result and the second prediction result and minimizing a deviation between the first prediction result and the label data corresponding to the sample business data as optimization objectives, where the trained prediction model is applied to business execution.
Optionally, the training module 303 is specifically configured to perform joint training on the reference model and the prediction model with optimization objectives of minimizing a deviation between the first prediction result and the second prediction result, minimizing a deviation between the first prediction result and the label data corresponding to the sample service data, and minimizing a deviation between the second prediction result and the label data.
Optionally, the apparatus further comprises:
an application module 304, configured to obtain service data of a user; inputting the service data into the prediction model to obtain a prediction result output by the prediction model, wherein the prediction model is obtained by training according to the model training method; and executing the service according to the prediction result.
Optionally, the application module 304 is specifically configured to obtain initial data required by the user to execute the service; identifying private data of the user from the initial data; and eliminating the privacy data from the initial data to obtain service data required by service execution.
The present specification also provides a computer readable storage medium having stored thereon a computer program operable to perform a method of model training as provided above with respect to fig. 1.
The present specification also provides a schematic block diagram of an electronic device corresponding to fig. 1 shown in fig. 4. As shown in fig. 4, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include hardware required for other services. The processor reads a corresponding computer program from the non-volatile memory into the memory and then runs the computer program to implement the model training method described in fig. 1. Of course, besides the software implementation, the present specification does not exclude other implementations, such as logic devices or a combination of software and hardware, and the like, that is, the execution subject of the following processing flow is not limited to each logic unit, and may be hardware or logic devices.
In the 90 s of the 20 th century, improvements in a technology could clearly distinguish between improvements in hardware (e.g., improvements in circuit structures such as diodes, transistors, switches, etc.) and improvements in software (improvements in process flow). However, as technology advances, many of today's process flow improvements have been seen as direct improvements in hardware circuit architecture. Designers almost always obtain the corresponding hardware circuit structure by programming an improved method flow into the hardware circuit. Thus, it cannot be said that an improvement in the process flow cannot be realized by hardware physical modules. For example, a Programmable Logic Device (PLD) (e.g., a Field Programmable Gate Array (FPGA)) is an integrated circuit whose Logic functions are determined by a user programming the Device. A digital system is "integrated" on a PLD by the designer's own programming without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing an Integrated Circuit chip, such Programming is often implemented by "logic compiler" software, which is similar to a software compiler used in program development and writing, but the original code before compiling is also written by a specific Programming Language, which is called Hardware Description Language (HDL), and HDL is not only one but many, such as ABEL (Advanced Boolean Expression Language), AHDL (alternate Hardware Description Language), traffic, CUPL (core universal Programming Language), HDCal, jhddl (Java Hardware Description Language), lava, lola, HDL, PALASM, rhyd (Hardware Description Language), and vhigh-Language (Hardware Description Language), which is currently used in most popular applications. It will also be apparent to those skilled in the art that hardware circuitry that implements the logical method flows can be readily obtained by merely slightly programming the method flows into an integrated circuit using the hardware description languages described above.
The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, atmel AT91SAM, microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic for the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may thus be considered a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.
The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
For convenience of description, the above devices are described as being divided into various units by function, respectively. Of course, the functions of the various elements may be implemented in the same one or more software and/or hardware implementations of the present description.
As will be appreciated by one skilled in the art, embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The description has been presented with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the description. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or apparatus that comprises the element.
As will be appreciated by one skilled in the art, embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
This description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the system embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and reference may be made to the partial description of the method embodiment for relevant points.
The above description is only an example of the present disclosure, and is not intended to limit the present disclosure. Various modifications and alterations to this description will become apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification should be included in the scope of the claims of the present specification.
Claims (10)
1. A method of model training, comprising:
acquiring sample service data and privacy data;
inputting the sample business data and the privacy data into a reference model to be trained to obtain a first prediction result, and inputting the sample business data into the prediction model to be trained to obtain a second prediction result;
and performing joint training on the reference model and the prediction model by taking the minimized deviation between the first prediction result and the second prediction result and the minimized deviation between the first prediction result and the label data corresponding to the sample business data as optimization targets, wherein the trained prediction model is applied to business execution.
2. The method according to claim 1, wherein the joint training of the reference model and the prediction model is performed with the optimization objectives of minimizing a deviation between the first prediction result and the second prediction result and minimizing a deviation between the first prediction result and the label data corresponding to the sample traffic data, specifically comprising:
and performing joint training on the reference model and the prediction model by taking the minimization of the deviation between the first prediction result and the second prediction result, the minimization of the deviation between the first prediction result and the label data corresponding to the sample business data and the minimization of the deviation between the second prediction result and the label data as optimization targets.
3. The method of claim 1, further comprising:
acquiring service data of a user;
inputting the service data into the prediction model, and obtaining a prediction result output by the prediction model, wherein the prediction model is obtained by training according to the method of the claim 1 or 2;
and executing the service according to the prediction result.
4. The method of claim 3, wherein the obtaining of the service data of the user specifically comprises:
acquiring initial data required by a user to execute a service;
identifying private data of the user from the initial data;
and eliminating the privacy data from the initial data to obtain service data required by service execution.
5. An apparatus for model training, comprising:
the acquisition module is used for acquiring sample service data and privacy data;
the input module is used for inputting the sample business data and the privacy data into a reference model to be trained to obtain a first prediction result, and inputting the sample business data into the prediction model to be trained to obtain a second prediction result;
and the training module is used for performing joint training on the reference model and the prediction model by taking the minimized deviation between the first prediction result and the second prediction result and the minimized deviation between the first prediction result and the label data corresponding to the sample business data as optimization targets, wherein the trained prediction model is applied to business execution.
6. The apparatus of claim 5, wherein the training module is specifically configured to jointly train the reference model and the prediction model with optimization objectives of minimizing a deviation between the first prediction result and the second prediction result, minimizing a deviation between the first prediction result and the label data corresponding to the sample traffic data, and minimizing a deviation between the second prediction result and the label data.
7. The apparatus of claim 5, further comprising:
the application module is used for acquiring the service data of the user; inputting the service data into the prediction model, and obtaining a prediction result output by the prediction model, wherein the prediction model is obtained by training according to the method of the claim 1 or 2; and executing the service according to the prediction result.
8. The apparatus according to claim 7, wherein the application module is specifically configured to obtain initial data required by a user to execute a service; identifying private data of the user from the initial data; and eliminating the privacy data from the initial data to obtain service data required by service execution.
9. A computer readable storage medium storing a computer program which when executed by a processor implements the method of any of claims 1~4 above.
10. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor when executing the program implementing the method of any of claims 1~4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211121861.6A CN115238826B (en) | 2022-09-15 | 2022-09-15 | Model training method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211121861.6A CN115238826B (en) | 2022-09-15 | 2022-09-15 | Model training method and device, storage medium and electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115238826A CN115238826A (en) | 2022-10-25 |
CN115238826B true CN115238826B (en) | 2022-12-27 |
Family
ID=83682027
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211121861.6A Active CN115238826B (en) | 2022-09-15 | 2022-09-15 | Model training method and device, storage medium and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115238826B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115545720B (en) * | 2022-11-29 | 2023-03-10 | 支付宝(杭州)信息技术有限公司 | Model training method, business wind control method and business wind control device |
CN116109008B (en) * | 2023-04-04 | 2023-07-04 | 支付宝(杭州)信息技术有限公司 | Method and device for executing service, storage medium and electronic equipment |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113221717A (en) * | 2021-05-06 | 2021-08-06 | 支付宝(杭州)信息技术有限公司 | Model construction method, device and equipment based on privacy protection |
Family Cites Families (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110868326B (en) * | 2019-11-27 | 2022-07-19 | 武汉虹旭信息技术有限责任公司 | Network service quality analysis method, edge device and central server |
US11803657B2 (en) * | 2020-04-23 | 2023-10-31 | International Business Machines Corporation | Generation of representative data to preserve membership privacy |
CN111738441B (en) * | 2020-07-31 | 2020-11-17 | 支付宝(杭州)信息技术有限公司 | Prediction model training method and device considering prediction precision and privacy protection |
CN111931216B (en) * | 2020-09-16 | 2021-03-30 | 支付宝(杭州)信息技术有限公司 | Method and system for obtaining joint training model based on privacy protection |
US11770407B2 (en) * | 2021-01-12 | 2023-09-26 | Walmart Apollo, Llc | Methods and apparatuses for defending against data poisoning attacks in recommender systems |
CN112800468B (en) * | 2021-02-18 | 2022-04-08 | 支付宝(杭州)信息技术有限公司 | Data processing method, device and equipment based on privacy protection |
CN113221747B (en) * | 2021-05-13 | 2022-04-29 | 支付宝(杭州)信息技术有限公司 | Privacy data processing method, device and equipment based on privacy protection |
CN113094758B (en) * | 2021-06-08 | 2021-08-13 | 华中科技大学 | Gradient disturbance-based federated learning data privacy protection method and system |
CN113378982A (en) * | 2021-07-02 | 2021-09-10 | 支付宝(杭州)信息技术有限公司 | Training method and system of image processing model |
CN114003949B (en) * | 2021-09-30 | 2022-08-30 | 清华大学 | Model training method and device based on private data set |
-
2022
- 2022-09-15 CN CN202211121861.6A patent/CN115238826B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113221717A (en) * | 2021-05-06 | 2021-08-06 | 支付宝(杭州)信息技术有限公司 | Model construction method, device and equipment based on privacy protection |
Non-Patent Citations (1)
Title |
---|
面向移动终端智能的自治学习系统;徐梦炜等;《软件学报》;20201014(第10期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN115238826A (en) | 2022-10-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115238826B (en) | Model training method and device, storage medium and electronic equipment | |
CN110008991B (en) | Risk event identification method, risk identification model generation method, risk event identification device, risk identification equipment and risk identification medium | |
CN113688313A (en) | Training method of prediction model, information pushing method and device | |
CN112966186A (en) | Model training and information recommendation method and device | |
CN113643119A (en) | Model training method, business wind control method and business wind control device | |
CN114997472A (en) | Model training method, business wind control method and business wind control device | |
CN111507726A (en) | Message generation method, device and equipment | |
CN112966577B (en) | Method and device for model training and information providing | |
CN112561162A (en) | Information recommendation method and device | |
CN116308738B (en) | Model training method, business wind control method and device | |
CN115545353B (en) | Business wind control method, device, storage medium and electronic equipment | |
CN115563584B (en) | Model training method and device, storage medium and electronic equipment | |
CN116822606A (en) | Training method, device, equipment and storage medium of anomaly detection model | |
CN116188023A (en) | Method, device, storage medium and equipment for training risk identification model | |
CN114996570A (en) | Information recommendation method and device | |
CN113343085B (en) | Information recommendation method and device, storage medium and electronic equipment | |
CN115758141A (en) | Method and device for model training and business wind control | |
CN115017905A (en) | Model training and information recommendation method and device | |
CN114331602A (en) | Model training method based on transfer learning, information recommendation method and device | |
CN115018866A (en) | Boundary determining method and device, storage medium and electronic equipment | |
CN110704742B (en) | Feature extraction method and device | |
CN109325127B (en) | Risk identification method and device | |
CN111539962A (en) | Target image classification method, device and medium | |
CN111160973A (en) | Advertisement pushing method and device | |
CN115545938B (en) | Method, device, storage medium and equipment for executing risk identification service |
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 |