WO2020000689A1

WO2020000689A1 - Transfer-learning-based robo-advisor strategy generation method and apparatus, and electronic device and storage medium

Info

Publication number: WO2020000689A1
Application number: PCT/CN2018/106260
Authority: WO
Inventors: 毕野; 黄博; 吴振宇; 王建明; 肖京
Original assignee: 平安科技（深圳）有限公司
Priority date: 2018-06-29
Filing date: 2018-09-18
Publication date: 2020-01-02
Also published as: CN109035028B; CN109035028A

Abstract

The present disclosure belongs to the technical field of artificial intelligence and relates to a transfer-learning-based robo-advisor strategy generation method and apparatus, and an electronic device and a computer-readable storage medium, wherein same belong to the technical field of data processing. The method comprises: acquiring a source financial management strategy generation model; according to sample data of a robo-advisor strategy and a result tag, adjusting parameters of the source financial management strategy generation model and the dimension of an input vector thereof to acquire a robo-advisor strategy generation model; and inputting acquired basic data of a target object into the robo-advisor strategy generation model for analysis to generate a robo-advisor strategy for the target object. The present disclosure can transfer a financial management strategy generation model to a robo-advisor strategy generation model, thereby reducing the amount of training of a machine learning model, compensating for the deficiency of robo-advisor sample data and improving the accuracy of a robo-advisor strategy.

Description

Method and device for generating intelligent investment consulting strategy based on transfer learning, electronic equipment and storage medium

Technical field

Cross-reference to related applications: This application claims the priority of the Chinese patent application CN201810700426.6, filed on June 29, 2018, with the invention name of "Intelligent Investment Advisory Strategy Generation Method and Device, Electronic Equipment, Storage Medium", which is hereby adopted This is incorporated by reference in its entirety.

The present disclosure relates to the field of artificial intelligence technology, and in particular, to a method and device for generating intelligent investment advisory strategies based on transfer learning, an electronic device, and a computer-readable storage medium.

Background technique

Data processing and big data analysis technologies are increasingly used in the financial services sector, providing theoretical investment advice to investor users. Robo-Advisor is an emerging product in the field of financial services. It refers to the application of theoretical models such as intelligent algorithms or optimal investment portfolios based on the user's basic information, risk tolerance level, return goals, and style preferences. Users provide comprehensive investment references, including strategies for asset allocation and asset dynamic balance. The inventors of the present application realized that due to the short development time of intelligent investment advisory, in the process of developing intelligent investment advisory strategies, there is generally a lack of sufficient sample data and historical experience, resulting in accurate existing intelligent investment advisory algorithm algorithm models. The rate is low, it is difficult to accurately classify users, and the final strategy is difficult to ensure ideal investment returns, which affects the quality of intelligent investment advisory products.

It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of the disclosure, and therefore may include information that does not constitute the prior art known to those of ordinary skill in the art.

technical problem

One of the objectives of the present disclosure is to provide a method and device for generating intelligent investment advisory strategies based on transfer learning, an electronic device, and a computer-readable storage medium, so as to at least to some extent overcome the problems caused by the limitations and defects of the prior art. The problem of low accuracy of the intelligent investment advisory strategy.

Other features and advantages of the disclosure will become apparent from the following detailed description, or may be learned in part through the practice of the disclosure.

Technical solutions

According to one aspect of the present disclosure, a method for generating an intelligent investment advisory strategy based on transfer learning is provided, which includes: obtaining a source financial strategy generation model; and adjusting the source financial strategy generation model according to sample data and result tags of the intelligent investment advisory strategy. Parameters and the dimensions of the input vector to obtain an intelligent investment advisory strategy generation model; input the obtained basic data of the target object into the intelligent investment advisory strategy generation model for analysis to generate the intelligent investment advisory strategy of the target object.

According to another aspect of the present disclosure, there is provided an intelligent investment and advisory strategy generating device based on transfer learning, including: a source model acquisition module for acquiring a source financial strategy generation model; and a model migration learning module for using an intelligent investment advisory strategy Sample data and classification labels, adjust the parameters of the source financial strategy generation model and the dimensions of the input vector to obtain a smart investment strategy generation model; a target object analysis module for inputting the acquired basic data of the target object into the smart An advisory strategy generation model is analyzed to generate an intelligent advisory strategy for the target object.

According to yet another aspect of the present disclosure, there is provided an electronic device including: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the executable instructions by Perform the method described in any one of the exemplary embodiments described above.

According to yet another aspect of the present disclosure, there is provided a computer-readable storage medium having stored thereon a computer program that, when executed by a processor, implements the method described in any one of the exemplary embodiments described above.

Beneficial effect

The exemplary embodiments of the present disclosure have the following beneficial effects:

In the above method and device, based on the source financial strategy generation model, the parameters of the source model and the dimensions of the input vector are adjusted based on a small amount of smart investment strategy sample data and result labels, and the smart investment strategy generation model is obtained through training. Thus, the transfer learning from the source financial strategy generation model to the intelligent investment advisory strategy generation model is completed; and then the intelligent investment advisory strategy generation model analyzes the basic data of the target object to generate an intelligent investment advisory strategy for the target object. On the one hand, this embodiment solves the cold start problem of the intelligent investment advisory strategy generation model by migrating the wealth management strategy generation model to the intelligent investment advisory strategy generation model, sharing knowledge and experience between the wealth management product and the intelligent investment advisory product, The training volume of the model is reduced, and the short board of insufficient sample data is made up, so that the generated intelligent investment advisory strategy has a higher accuracy rate. On the other hand, after obtaining the basic data of the target object in this embodiment, the intelligent investment advisory policy generation model can be used to generate the intelligent investment advisory policy for the target object through training, thereby realizing the automatic generation of the intelligent investment advisory policy, thereby reducing manpower. Cost, improving the efficiency of the intelligent investment advisory strategy generation process.

It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and should not limit the present disclosure.

BRIEF DESCRIPTION OF THE DRAWINGS

The drawings herein are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure, and together with the description serve to explain the principles of the present disclosure. Obviously, the drawings in the following description are just some embodiments of the present disclosure. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 illustrates a system architecture diagram of a method for generating an intelligent investment advisory strategy applying an exemplary embodiment of the present disclosure.

FIG. 2 shows a flowchart of a method for generating an intelligent investment advisory strategy in an exemplary embodiment of the present disclosure.

FIG. 3 illustrates a sub-flow chart of a method for generating an intelligent investment advisory strategy in an exemplary embodiment of the present disclosure.

FIG. 4 shows a schematic diagram of an intelligent investment advisory strategy generation model in an exemplary embodiment of the present disclosure.

FIG. 5 shows a structural block diagram of an intelligent investment advisory strategy generating device in an exemplary embodiment of the present disclosure.

FIG. 6 illustrates an electronic device for implementing the above method in an exemplary embodiment of the present disclosure.

FIG. 7 illustrates a computer-readable storage medium for implementing a method in an exemplary embodiment of the present disclosure.

Embodiments of the invention

Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as being limited to the examples set forth herein; rather, the embodiments are provided so that this disclosure will be more comprehensive and complete, and the concepts of the example embodiments will be fully conveyed To those skilled in the art. The described attributes, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

An exemplary embodiment of the present disclosure first provides a method for generating an intelligent investment advisory strategy based on transfer learning. Transfer learning is a method of machine learning. The intelligent investment advisory strategy refers to a specific investment portfolio or asset allocation recommendation strategy for different investor users, enterprise customers, and other objects. The method of this exemplary embodiment uses a financial strategy generation model as a source model, and transforms it into an intelligent investment advisory strategy generation model through transfer learning.

FIG. 1 shows a schematic diagram of a system architecture that can run the intelligent investment advisory strategy generation method of the exemplary embodiment. As shown in FIG. 1, the system 100 may include terminal devices 101, 102, 103, a network 104, a server 105, and a database 106. The user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to send or receive information. The server 105 can provide service support for users by reading and writing data from the database 106, for example, obtaining user information data of the terminal devices 101, 102, 103 from the database 106, generating corresponding intelligent investment advisory policies, and feeding back to the terminal device 101, 102, 103. In some cases, the database 106 may also be installed on the server 105.

Based on the above description, the method for generating an intelligent investment advisory strategy in this exemplary embodiment can be applied to the server 105.

It should be understood that the number of terminal devices, networks, servers, and databases in FIG. 1 is merely exemplary, and any number of terminal devices, networks, servers, and databases may be set according to actual needs.

It should be noted that, not limited to the system shown in FIG. 1, the method of this exemplary embodiment can also be applied to a computer that is not connected to a terminal device. Based on the acquired user information data, an intelligent investment advisory policy is generated, and the intelligent The investment advisory strategy is sent to the corresponding server, which is fed back to the user terminal through the server, or it can be fed back to the user via phone, text message, and other methods. This embodiment does not specifically limit this.

The method for generating the intelligent investment advisory strategy of this embodiment is further described below with reference to FIG. 2. Referring to FIG. 2, the method may include the following steps:

Step S21: Obtain a source financial strategy generation model.

The source financial strategy generation model can be an existing model. A financial strategy usually refers to a strategy that provides recommendations for users to purchase financial products. Compared with intelligent investment advisory strategies, financial management strategies involve less data features and lower strategic complexity, so they cannot be directly used in intelligent investment advisory products. And it can be used as an aid or guide for intelligent investment advisory strategies.

Compared to smart investment advisors, the development of wealth management services has taken longer, and companies often accumulate a large amount of historical wealth management data and experience. In an exemplary embodiment, step S21 may be implemented by obtaining the sample data and result tags of the source financial strategy, and training a machine learning model to obtain the source financial strategy generation model. Sample data can be collected from historical financial data, and the results of the determined financial strategy can be used as result labels, which can be input into the machine learning model together with the sample data, and multiple models of the model can be obtained through multiple iterations to obtain a complete model. In the analysis of financial strategy, it is usually based on the user's personal information or behavioral data, and it is input into the model in the form of feature vectors after statistics based on multiple features. Therefore, machine learning models can include neural network models, regression models, support vector product models, etc. These machine learning models are very suitable for processing multi-dimensional vector data analysis, so they can be used as the initial model of the source financial strategy generation model.

Step S22: Adjust the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and result labels of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy generation model.

Due to the short launch time of the intelligent investment advisory products, the sample data of the intelligent investment advisory strategy with the result label is very small, and a large amount of available data has no corresponding result label. If the initial machine learning model is trained with a small amount of sample data and result labels, it is difficult to guarantee a high degree of training and accuracy. Therefore, the source financial strategy generation model obtained in step S21 can be used as the source model in the transfer learning. Considering that the intelligent investment advisory strategy may involve more characteristic information of the user, the dimension of the input vector of the source financial strategy generation model can be adjusted. At the same time, the parameters, including the number of parameters and specific values, can be adjusted manually based on the source financial strategy generation model based on experience, or iteratively adjusted during the training process. The parameters that have been trained in the source financial strategy generation model have important reference functions for the training of the intelligent investment advisory strategy generation model. Due to the high commonality between the financial management strategy and the intelligent investment advisory strategy in algorithm principles, this part of the parameters are usually in The intelligent investment advisory strategy generation model will not change much during the training process, so it can reduce the amount of training.

In step S23, the obtained basic data of the target object is input into an intelligent investment advisory policy generation model for analysis, and an intelligent investment advisory policy of the target object is generated.

Among them, the target object can be investor users, enterprise customers, etc. The basic data refers to various data related to the analysis of intelligent investment advisory strategies, including personal information, behavior data, and so on. Through the intelligent investment advisory strategy generation model obtained in the above steps, and analyzing the basic data of the target object, an intelligent investment advisory strategy suitable for the target object can be generated.

In the above method, based on the source financial strategy generation model, the parameters of the source model and the dimensions of the input vector are adjusted based on a small amount of smart investment strategy sample data and result labels, and the smart investment strategy generation model is obtained through training to complete the Transfer learning from the source financial strategy generation model to the intelligent investment advisory strategy generation model; and then analyze the basic data of the target object through the intelligent investment advisory strategy generation model to generate an intelligent investment advisory strategy for the target object. On the one hand, this embodiment solves the cold start problem of the intelligent investment advisory strategy generation model by migrating the wealth management strategy generation model to the intelligent investment advisory strategy generation model, sharing knowledge and experience between the wealth management product and the intelligent investment advisory product, The training volume of the model is reduced, and the short board of insufficient sample data is made up, so that the generated intelligent investment advisory strategy has a higher accuracy rate. On the other hand, after obtaining the basic data of the target object in this embodiment, the intelligent investment advisory policy generation model can be used to generate the intelligent investment advisory policy for the target object through training, thereby realizing the automatic generation of the intelligent investment advisory policy, thereby reducing manpower. Cost, improving the efficiency of the intelligent investment advisory strategy generation process.

In an exemplary embodiment, referring to FIG. 3, step S22 may be implemented through steps S31 to S34:

Step S31: Obtain a wealth management feature set from the sample data of the source wealth management strategy;

Step S32: Obtain the intelligent investment advisory feature set from the sample data of the intelligent investment advisory strategy, and adjust the dimension of the input vector of the source financial strategy generation model according to the intelligent investment advisory feature set;

Step S33: Set the initial value of the weighting coefficients of the features that belong to the smart investment advisory feature set and not the wealth management feature set to 0 to obtain an intermediate model;

In step S34, the intermediate model is trained according to the sample data and result labels of the intelligent investment advisory strategy, and the intelligent investment advisory strategy generation model is acquired.

Financial characteristics, that is, financial strategies, need to consider information about which characteristics of users, and organize them into a set of financial characteristics; intelligent investment advisory strategies need to consider the characteristics of users into intelligent investment advisory feature sets. Generally, financial management features need to be considered in the analysis of intelligent investment advisory strategies, but the converse is not necessarily the case, that is, the financial management feature set can be a subset of the intelligent investment advisory feature set. Taking Table 1 as an example, the characteristics of financial management can include 8 characteristics such as age, occupation, income, etc. In addition to these 8 characteristics, the characteristics of intelligent investment consulting can also include 4 characteristics such as historical borrowing behavior and health status. It can be seen that intelligent investment consulting The number of features is generally more and includes all financial management features. Generally, after the sample data is sorted and collected, a feature set can be obtained.

Based on the situation shown in Table 1, in the source financial strategy generation model, the input vector should be a vector of 8 dimensions, and the intelligent investment advisory strategy generation model should be a vector of 12 dimensions. Therefore, when performing model transfer learning, you can first adjust the dimensions of the input vector of the source financial strategy generation model to be the same as the number of intelligent investment advisory features. The extra dimensions correspond to the features that need not be considered in the financial strategy. The initial value of the weight coefficient of the feature is set to 0, and the existing weight coefficient in the source financial strategy generation model can be retained, so that the source financial strategy generation model is adjusted according to the final form of the intelligent investment advisory strategy generation model to obtain an intermediate model. After that, the intermediate model is trained based on the sample data and result tags of the intelligent investment advisory strategy. The model parameters can be adjusted and optimized, and finally the intelligent investment advisory strategy generation model is obtained.

Further, the output form of the source financial strategy generation model can be adjusted to conform to the specific form of the intelligent investment advisory strategy. The output of the source financial strategy generation model can be in two forms: multi-dimensional vectors or classification results. A multi-dimensional vector refers to a financial strategy that is output in the form of a ratio of various financial products. Each dimension in the multi-dimensional vector represents each type of financial product, and its value represents the proportion. The classification result refers to the determination of specific various financial strategies as Various classifications, and the financial management strategies of each classification have been determined in advance. The model only needs to determine which classification belongs to, and then decide which financial management strategy to use. Correspondingly, the output of the intelligent investment advisory strategy generation model can also be in the above two forms: multi-dimensional vectors or classification results. If the source financial generation model outputs multi-dimensional vectors, in the intelligent investment advisory strategy generation model, the dimension of the output vector can be increased according to the number of financial products; if the source financial generation model outputs the classification results, in the intelligent investment advisory strategy generation model, you can Increase the number of classifications based on the combination or allocation of financial products.

In an exemplary embodiment, the source financial strategy generation model may be a neural network model. When adjusting the parameters of the source financial strategy generation model, specifically, the weight coefficient of the neural network model may be adjusted, or at least one intermediate layer may be added. .

The following uses a specific example for further explanation. Based on the features in Table 1, a neural network model as shown in Figure 4 can be constructed. The dotted line is the source financial strategy generation model, including the input layer Input (S), the first intermediate layer Mid (S) 1 (the source financial strategy generation model contains only one intermediate layer), and the output layer Output (S).

Input a vector of 8 dimensions,

, Where x1, x2, etc. represent the input value of each feature;

In this embodiment, the output of the source financial strategy generation model is a multi-dimensional vector, which represents the ratio between wealth management A, wealth management B, wealth management C, funds, and bonds. By inputting sample data and result labels of the financial strategy, the values of W (S) 1 and W (S) 2 can be trained to determine the source financial strategy generation model.

When performing model transfer learning, because the dimensions of the input layer and output layer change, usually the dimensions of the intermediate layer can also be changed accordingly. For example, as shown in Figure 4, the dimensions of the first intermediate layer are generated by the source financial strategy in the model. Four become eight of the intelligent investment advisory strategy generation models.

The complete model in FIG. 4 is the intelligent investment advisory strategy generation model, which includes an input layer Input (T), a first intermediate layer Mid (T) 1, and an output layer Output (T).

Enter a vector of 12 dimensions,

, Where y1, y2, etc. represent the input values of each feature;

It can be seen that the weight coefficients W (T) 1, W (T) 2 are different from the weight coefficients W (S) 1, W (S) 2 in the source financial strategy generation model, so the source financial strategy generation needs to be adjusted The weight coefficient in the model.

For example, the initial value of the weight coefficient can be set for the neural network model of FIG. 4 through step S33 in FIG. 3. At the input layer, since the last 4 features belong to the intelligent investment advisory feature set and not to the financial feature set, the The initial values of the weight coefficients of these four features are set to 0, and the initial value of the weight coefficient W (T) 1 can be:

In addition, as shown in FIG. 4, one or more intermediate layers can be added to reflect the more complex relationship between the characteristics of intelligent investment advisory and intelligent investment advisory strategies. When adding an intermediate layer, you can copy an existing intermediate layer and gradually adjust its dimensions and weight coefficients during the training process to solve the problem of initial values assigned to the intermediate layer. Because increasing the middle layer will increase the amount of training, in the transfer learning process of the source financial strategy generation model, the weight coefficient can be adjusted first. When the adjustment of the weight coefficient cannot make the model's accuracy reach the requirement, the middle layer is added.

It should be noted that FIG. 4 is only an example of a source financial strategy generation model and an intelligent investment advisory strategy generation model, and is not limited to the feature names, the number of dimensions, the number of intermediate layers, and the specific forms of each layer shown in FIG. 4. In Figure 4, the actual meaning of each dimension of the middle layer (customer positioning, basic needs, etc.) is used to facilitate the initial value of the weight coefficient at the initial stage of the model, and to increase the transparency of the model to facilitate parameter debugging. The neural network model It is a black box model. Generally, the dimensions of the intermediate layer do not need to have corresponding actual meanings. During the training process, the intermediate layer may be adjusted to a large degree to obtain the optimal model parameters and make the output as accurate as possible. Therefore, the text labeling of each dimension in the neural network model shown in FIG. 4 does not constitute a limitation on the protection scope of the present disclosure.

In an exemplary embodiment, the method for generating an intelligent investment advisory policy may further include periodically acquiring changes in basic data of the target object, and updating the intelligent investment advisory policy of the target object. Changes in the basic data of the target object can usually include two situations: changes in the user's basic information or behavior data, and changes in assets due to income in the process of using intelligent investment consulting products. When the basic data changes, the intelligent investment advisory strategy obtained based on the analysis of the basic data may also change accordingly. Therefore, the intelligent investment advisory strategy of the target object can be periodically updated to achieve the dynamic optimal allocation of assets and improve returns.

During the actual use of the intelligent investment advisory strategy generation model, the use results can also be fed back to the model, such as fine-tuning its intelligent investment advisory strategy based on the actual application of the intelligent investment advisory strategy, and fine-tuning the intelligent investment advisory strategy. The strategy is used as the result label of the sample data to iteratively optimize the parameters of the model, thereby further improving the accuracy of the intelligent investment advisory strategy generated by the model.

Exemplary embodiments of the present disclosure also provide an intelligent investment advisory policy generating device based on transfer learning. Referring to FIG. 5, the device 500 may include: a source model acquisition module 510 for acquiring a source financial policy generation model; A model migration learning module 520 is used to adjust the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and classification labels of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy generation model; the target object analysis module 530 is used to The obtained basic data of the target object is input into the intelligent investment advisory strategy generation model for analysis, and the intelligent investment advisory strategy of the target object is generated. The specific details of each module have been described in detail in the embodiments of the method section, and therefore will not be described again.

An exemplary embodiment of the present disclosure also provides an electronic device capable of implementing the above method.

Those skilled in the art can understand that various aspects of the present disclosure may be implemented as a system, method, or program product. Therefore, various aspects of the present disclosure can be embodied in the following forms: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, which can be collectively referred to herein "Circuit," "Module," or "System."

An electronic device 600 according to such an exemplary embodiment of the present disclosure is described below with reference to FIG. 6. The electronic device 600 shown in FIG. 6 is only an example, and should not impose any limitation on the functions and scope of use of the embodiments of the present disclosure.

As shown in FIG. 6, the electronic device 600 is expressed in the form of a general-purpose computing device. The components of the electronic device 600 may include, but are not limited to, the at least one processing unit 610, the at least one storage unit 620, a bus 630 connecting different system components (including the storage unit 620 and the processing unit 610), and a display unit 640.

The storage unit stores program code, and the program code can be executed by the processing unit 610, so that the processing unit 610 executes the steps according to various exemplary embodiments of the present disclosure described in the "exemplary method" section of the present specification. For example, the processing unit 610 may perform steps S21 to S23 shown in FIG. 2, and may also perform steps S31 to S34 shown in FIG. 3.

The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 621 and / or a cache storage unit 622, and may further include a read-only storage unit (ROM) 623.

The storage unit 620 may further include a program / utility tool 624 having a set (at least one) of program modules 625, such program modules 625 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data, Each or some combination of these examples may include an implementation of a network environment.

The bus 630 may be one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local area using any bus structure in a variety of bus structures bus.

The electronic device 600 may also communicate with one or more external devices 800 (such as a keyboard, pointing device, Bluetooth device, etc.), and may also communicate with one or more devices that enable a user to interact with the electronic device 600, and / or with Any device (eg, router, modem, etc.) that enables the electronic device 600 to communicate with one or more other computing devices. This communication can be performed through an input / output (I / O) interface 650. Moreover, the electronic device 600 may also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through the network adapter 660. As shown, the network adapter 660 communicates with other modules of the electronic device 600 through the bus 630. It should be understood that although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives And data backup storage systems.

Through the description of the foregoing embodiments, those skilled in the art can easily understand that the example embodiments described herein can be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (which may be a CD-ROM, a U disk, a mobile hard disk, etc.) or on a network Including instructions to cause a computing device (which may be a personal computer, a server, a terminal device, or a network device, etc.) to execute a method according to an exemplary embodiment of the present disclosure.

Exemplary embodiments of the present disclosure also provide a computer-readable storage medium having stored thereon a program product capable of implementing the above method of the present specification. In some possible implementation manners, various aspects of the present disclosure may also be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to cause the terminal device to execute the foregoing description of the specification. The steps according to various exemplary embodiments of the present disclosure are described in the "Exemplary Method" section.

Referring to FIG. 7, a program product 700 for implementing the above method according to an exemplary embodiment of the present disclosure is described, which may adopt a portable compact disk read-only memory (CD-ROM) and include program code, and may be implemented in a terminal. Device, such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, the readable storage medium may be any tangible medium containing or storing a program, and the program may be used by or in combination with an instruction execution system, apparatus, or device.

The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: electrical connection with one or more wires, portable disk, hard disk, random access memory (RAM), read-only memory (ROM), erasable Programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the foregoing.

The computer-readable signal medium may include a data signal in baseband or propagated as part of a carrier wave, which carries readable program code. Such a propagated data signal may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The readable signal medium may also be any readable medium other than a readable storage medium, and the readable medium may send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

The program code contained on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

The program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages. The programming languages include object-oriented programming languages such as Java, C ++, etc., and also include conventional procedural programming. Language—such as "C" or a similar programming language. The program code can be executed entirely on the user computing device, partly on the user device, as an independent software package, partly on the user computing device, partly on the remote computing device, or entirely on the remote computing device or server On. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it may be connected to an external computing device (such as provided by using an Internet service) (Commercially connected via the Internet).

In addition, the above-mentioned drawings are merely a schematic description of processes included in a method according to an exemplary embodiment of the present disclosure, and are not limiting purposes. It is easy to understand that the processes shown in the above drawings do not indicate or limit the chronological order of these processes. In addition, it is also easy to understand that these processes may be performed synchronously or asynchronously in multiple modules, for example.

It should be noted that although several modules or units of the device for action execution are mentioned in the detailed description above, this division is not mandatory. In fact, according to an exemplary embodiment of the present disclosure, the features and functions of two or more modules or units described above may be embodied in one module or unit. Conversely, the features and functions of a module or unit described above can be further divided into multiple modules or units to be embodied.

Those skilled in the art will readily contemplate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that conform to the general principles of this disclosure and include the common general knowledge or conventional technical means in the technical field not disclosed in this disclosure. . It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It should be understood that the present disclosure is not limited to the precise structure that has been described above and illustrated in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the disclosure is limited only by the following claims.

Claims

A method for generating an intelligent investment advisory strategy based on transfer learning, which includes:

Obtain the source financial strategy generation model;

Adjusting the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and result tags of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy generation model;

The obtained basic data of the target object is input to the intelligent investment consulting strategy generation model for analysis, and the intelligent investment consulting strategy of the target object is generated.
The method according to claim 1, wherein obtaining a source financial strategy generation model comprises:

Obtain sample data and result tags of the source financial strategy, and train a machine learning model to obtain the source financial strategy generation model.
The method according to claim 2, wherein adjusting the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and result tags of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy generation model comprises:

Obtaining a wealth management feature set from sample data of the source wealth management strategy;

Obtaining an intelligent investment advisory feature set from sample data of the intelligent investment advisory strategy, and adjusting a dimension of an input vector of the source financial strategy generation model according to the intelligent investment advisory feature set;

Setting an initial value of a weighting coefficient of a feature that belongs to the intelligent investment advisory feature set and does not belong to the wealth management feature set to 0 to obtain an intermediate model;

The intermediate model is trained according to the sample data and result tags of the intelligent investment advisory strategy, and the intelligent investment advisory strategy generation model is acquired.
The method according to claim 3, wherein the output of the intelligent investment strategy generation model comprises a multi-dimensional vector or a classification result.
The method according to claim 2, wherein the machine learning model comprises one or more of a neural network model, a regression model, and a support vector machine model.
The method according to claim 2, wherein the machine learning model comprises a neural network model, and adjusting parameters of the source financial strategy generation model comprises:

Adjusting a weighting coefficient of the neural network model, and / or adding at least one intermediate layer.
The method according to any one of claims 1-6, further comprising:

Periodically acquire changes in basic data of the target object, and update the smart investment strategy of the target object.
An intelligent investment advisory strategy generating device based on transfer learning is characterized in that it includes:

Source model acquisition module configured to: obtain a source financial strategy generation model;

The model migration learning module is configured to: adjust the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and classification labels of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy generation model;

The target object analysis module is configured to: input the obtained basic data of the target object into the intelligent investment consulting strategy generation model for analysis, and generate the intelligent investment consulting strategy of the target object.
The apparatus according to claim 8, wherein the source model acquisition module is further configured to:

Obtain sample data and result tags of the source financial strategy, and train a machine learning model to obtain the source financial strategy generation model.
The apparatus according to claim 9, wherein the model transfer learning module comprises:

A wealth management feature obtaining unit configured to: obtain a wealth management feature set from sample data of the source wealth management strategy;

A dimension adjustment unit configured to obtain a smart investment advisory feature set from sample data of the smart investment advisory strategy, and adjust a dimension of an input vector of the source financial strategy generation model according to the smart investment advisory feature set;

An intermediate model obtaining unit configured to: set an initial value of a weight coefficient of a feature belonging to the intelligent investment advisory feature set and not belonging to the wealth management feature set to 0 to obtain an intermediate model;

The strategy generation model acquisition unit is configured to: train the intermediate model according to the sample data and result tags of the smart investment advisory strategy to acquire the smart investment advisory strategy generation model.
The device according to claim 10, wherein the output of the intelligent investment strategy generation model comprises a multi-dimensional vector or a classification result.
The device according to claim 9, wherein the machine learning model comprises a neural network model, and parameters for adjusting the source financial strategy generation model include:

Adjusting a weighting coefficient of the neural network model, and / or adding at least one intermediate layer.
The device according to any one of claims 8-12, further comprising:

The update module is configured to periodically obtain changes in basic data of the target object, and update a smart investment strategy of the target object.
An electronic device, comprising:

Processor; and

A memory for storing executable instructions of the processor;

The processor is configured to perform the following steps by executing the executable instruction:

Obtain the source financial strategy generation model;

Adjusting the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and result tags of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy generation model;

The obtained basic data of the target object is input to the intelligent investment consulting strategy generation model for analysis, and the intelligent investment consulting strategy of the target object is generated.
The electronic device according to claim 14, wherein the obtaining a source financial strategy generation model comprises:

Obtain sample data and result tags of the source financial strategy, and train a machine learning model to obtain the source financial strategy generation model.
The electronic device according to claim 15, wherein adjusting the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and result tags of the intelligent investment advisory strategy, and obtaining the intelligent investment advisory strategy generation model comprises :

Obtaining a wealth management feature set from sample data of the source wealth management strategy;

Obtaining an intelligent investment advisory feature set from sample data of the intelligent investment advisory strategy, and adjusting a dimension of an input vector of the source financial strategy generation model according to the intelligent investment advisory feature set;

Setting an initial value of a weighting coefficient of a feature that belongs to the intelligent investment advisory feature set and does not belong to the wealth management feature set to 0 to obtain an intermediate model;

The intermediate model is trained according to the sample data and result tags of the intelligent investment advisory strategy, and the intelligent investment advisory strategy generation model is acquired.
The electronic device according to claim 16, wherein the output of the intelligent investment strategy generation model comprises a multi-dimensional vector or a classification result.
The electronic device according to claim 15, wherein the machine learning model includes a neural network model, and parameters for adjusting the source financial strategy generation model include:

Adjusting a weighting coefficient of the neural network model, and / or adding at least one intermediate layer.
The electronic device according to any one of claims 14-18, wherein the processor is further configured to perform steps by executing the executable instructions:

Periodically acquire changes in basic data of the target object, and update the smart investment strategy of the target object.
A computer-readable storage medium having stored thereon a computer program, characterized in that when the computer program is executed by a processor, the processor causes the processor to perform the following steps:

Obtain the source financial strategy generation model;

Adjusting the parameters of the source financial strategy generation model and the dimensions of the input vector according to the sample data and result tags of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy generation model;

The obtained basic data of the target object is input to the intelligent investment consulting strategy generation model for analysis, and the intelligent investment consulting strategy of the target object is generated.
The computer-readable storage medium of claim 20, wherein obtaining the source financial strategy generation model comprises:

Obtain sample data and result tags of the source financial strategy, and train a machine learning model to obtain the source financial strategy generation model.
The computer-readable storage medium according to claim 21, wherein parameters of the source financial strategy generation model and dimensions of an input vector are adjusted according to sample data and result tags of the intelligent investment advisory strategy to obtain the intelligent investment advisory strategy Generated models include:

Obtaining a wealth management feature set from sample data of the source wealth management strategy;

Obtaining an intelligent investment advisory feature set from sample data of the intelligent investment advisory strategy, and adjusting a dimension of an input vector of the source financial strategy generation model according to the intelligent investment advisory feature set;

Setting an initial value of a weighting coefficient of a feature that belongs to the intelligent investment advisory feature set and does not belong to the wealth management feature set to 0 to obtain an intermediate model;

The intermediate model is trained according to the sample data and result tags of the intelligent investment advisory strategy, and the intelligent investment advisory strategy generation model is acquired.
The computer-readable storage medium of claim 22, wherein the output of the intelligent investment strategy generation model comprises a multi-dimensional vector or a classification result.
The computer-readable storage medium according to claim 21, wherein the machine learning model includes a neural network model, and parameters for adjusting the source financial strategy generation model include:

Adjusting a weighting coefficient of the neural network model, and / or adding at least one intermediate layer.
The computer-readable storage medium according to any one of claims 20-24, wherein when the computer program is executed by a processor, the processor further causes the processor to perform steps:

Periodically acquire changes in basic data of the target object, and update the smart investment strategy of the target object.