Training method and device for personal loan amount assessment model based on blockchain

Abstract

The application discloses a training method and device of a personal loan amount evaluation model based on a blockchain, and the method comprises the following steps: each training node writes the training information thereof into the blockchain according to a training invitation sent by a task publishing node; the task publishing node determines one of the plurality of training nodes as a central training node according to the training information of each training node, and sends a training task to the central training node, and the rest of the training nodes are participating training nodes; the central training node divides the training task into a plurality of training subtasks according to the training capability information of each participating training node, and distributes each training subtask to the corresponding participating training node, so that each participating training node performs training; and the central training node obtains a personal loan amount evaluation model through integration. The application can coordinate a plurality of participants to jointly build a personal loan amount evaluation model without leaking customer privacy data.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a training method and device for a blockchain-based personal loan amount assessment model. Background Art

[0002] Determining the maximum individual loan amount requires comprehensive consideration of multiple data sources, including personal deposit information, borrowing information, purchasing power information, medical and health information, and basic personal information. This information is stored within various departments or businesses. Because these departments or businesses protect the privacy of citizens, it is difficult to share this information with local banks. Internal banks, financial institutions, hospitals, and other corporate departments are also reluctant to disclose relevant data to protect customer privacy or for commercial confidentiality. Consequently, when assessing the amount of personal credit loans, the "data silo" problem is encountered, making it difficult to accurately estimate the amount of personal credit loans. Summary of the Invention

[0003] To solve the above problems, the embodiments of the present application provide a training method and device for a blockchain-based personal loan amount assessment model, which aims to solve the above-mentioned "data island" problem.

[0004] The embodiments of this application adopt the following technical solutions:

[0005] In a first aspect, a method for training a personal loan amount assessment model based on a blockchain is provided, wherein the blockchain includes a task publishing node and multiple training nodes, and the method includes:

[0006] Each training node writes its own training information into the blockchain according to the training invitation sent by the task issuing node, wherein the training information includes: training capability information and user information;

[0007] The task issuing node determines one of the multiple training nodes as a central training node based on the training information of each training node, and sends the training task to the central training node, while the rest serve as participating training nodes;

[0008] The central training node divides the training task into a plurality of training subtasks according to the training capability information of each participating training node, and assigns each training subtask to a corresponding participating training node, so that each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model;

[0009] The central training node integrates the partial training models obtained by each participating training node, as well as the input layer and output layer deployed on the central training node, to obtain a personal loan amount assessment model.

[0010] Optionally, the method further includes:

[0011] The central training node sends the obtained personal loan amount evaluation model to each participating training node, so that each participating training node evaluates the personal loan amount based on the personal loan amount evaluation model.

[0012] Optionally, the task issuing node determines one of the multiple training nodes as a central training node based on the training information of each training node, including:

[0013] The task issuing node determines a training evaluation score for each training node based on the training task and the training capability information of each training node;

[0014] The training node with the lowest training evaluation score is used as the central training node.

[0015] Optionally, the task issuing node determines a training evaluation score for each training node based on the training task and the training capability information of each training node, including:

[0016] Determining a computing power requirement score and a storage capacity requirement score for the training task, wherein the sum of the computing power requirement score and the storage capacity requirement score is one;

[0017] Determining a computing power score and a storage capacity score for each training node, wherein the sum of the computing power score and the storage capacity score is one;

[0018] Determine a first product of the computing power requirement score and the computing power score, and a second product of the storage capacity requirement score and the storage capacity score, and use the sum of the first product and the second product as the training evaluation score of each training node.

[0019] Optionally, the training task includes a multi-layered personal loan amount assessment model;

[0020] The central training node divides the training task into multiple training subtasks based on the training capability information of each participating training node, including:

[0021] The central training node determines a training evaluation score for each participating training node based on the training task and the training capability information of each participating training node;

[0022] Determine the number of training layers for each participating training node based on the training evaluation score of each participating training node and the total number of model layers of the personal loan amount evaluation model;

[0023] The personal loan amount assessment model is divided into multiple training subtasks according to the number of training layers of each participating training node.

[0024] Optionally, each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model, including:

[0025] Each participating training node follows the order of the model training layers in the assigned training subtasks in the personal loan amount assessment model, uses the output of the previous participating training node as the input of the subsequent participating training node, and performs training based on the local user information of each participating training node to obtain a partial training model corresponding to each participating training node.

[0026] Optionally, the multiple training nodes include bank training nodes, financial institution training nodes, medical training nodes, and consumer training nodes.

[0027] Optionally, the user information includes: personal credit information, personal basic information, personal loan information, personal deposit information, and personal expenditure flow information provided by the bank training node;

[0028] Personal financial product purchase information provided by financial institution training nodes;

[0029] Personal medical records, hospitalization information, and physical examination information provided by medical training nodes;

[0030] Personal consumption information and purchased product information provided by consumer companies.

[0031] In a second aspect, a training device for a blockchain-based personal loan amount assessment model is provided. The blockchain includes a task publishing node and multiple training nodes. The training device is deployed in each node of the blockchain, and the device includes:

[0032] a writing unit, configured to write respective training information into the blockchain according to the training invitation sent by the task issuing node, wherein the training information includes training capability information and user information;

[0033] a task issuing unit, configured to determine one of the plurality of training nodes as a central training node based on the training information of each training node, and send the training task to the central training node, with the remaining nodes serving as participating training nodes;

[0034] a task assignment unit, configured to divide the training task into a plurality of training subtasks according to the training capability information of each participating training node, and assign each training subtask to a corresponding participating training node, so that each participating training node performs training according to the assigned training subtask and based on its own user information to obtain a partial training model;

[0035] The integration unit is used to integrate the partial training models obtained by each participating training node and the input layer and output layer deployed on the central training node to obtain a personal loan amount assessment model.

[0036] In a third aspect, a blockchain is provided, comprising a task publishing node and multiple training nodes;

[0037] Each training node is configured to write its own training information into the blockchain according to the training invitation sent by the task issuing node, wherein the training information includes training capability information and user information;

[0038] The task issuing node is configured to determine one of the plurality of training nodes as a central training node based on the training information of each training node, and send the training task to the central training node, with the remaining nodes serving as participating training nodes;

[0039] The central training node is configured to divide the training task into a plurality of training subtasks according to the training capability information of each participating training node, and assign each training subtask to a corresponding participating training node, so that each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model;

[0040] The central training node is used to integrate the partial training models obtained by each participating training node, as well as the input layer and output layer deployed on the central training node, to obtain a personal loan amount assessment model.

[0041] In a fourth aspect, an embodiment of the present application further provides an electronic device, comprising: a processor; and a memory arranged to store computer-executable instructions, wherein the executable instructions, when executed, enable the processor to perform any of the above methods.

[0042] In a fifth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple applications, the electronic device executes any of the above methods.

[0043] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0044] The present application is based on the idea of blockchain and split learning, splits the training task of the whole personal loan amount evaluation model into multiple training sub-tasks, and makes the participants participating in the construction of the personal loan amount evaluation model train according to the sub-tasks based on the local user personal information of each participant. Finally, the partial training models obtained by each participant are integrated together, and the whole personal loan amount evaluation model can be obtained. The present application can coordinate multiple participants to jointly build a personal loan amount evaluation model without revealing customer privacy data. The data and related information of each participant are all on-chain, realizing traceability and tamper resistance. The local data of each participant is not trained outside the local area, ensuring privacy and meeting the demand of model training task. In addition, the present application allocates training tasks according to the capabilities of each participant, so that the participant cannot complete the model training due to insufficient capability, and the participant does not have capability redundancy due to excessive capability. In addition, the present application can dynamically change according to the needs of each participant and the training task, and flexibly adjust the deployment and training of the training model. BRIEF DESCRIPTION OF DRAWINGS

[0045] The drawings described herein are used to provide further understanding of the present application, and form part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute improper limitations on the present application. In the drawings:

[0046] Figure 1 A structural schematic diagram of a blockchain according to an embodiment of the present application is shown;

[0047] Figure 2 A flowchart of a training method of a personal loan amount evaluation model based on a blockchain according to an embodiment of the present application is shown;

[0048] Figure 3 A structural schematic diagram of a training device of a personal loan amount evaluation model based on a blockchain according to an embodiment of the present application is shown;

[0049] Figure 4 A structural schematic diagram of an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0050] To make the purpose, technical scheme and advantages of the present application clearer, the technical scheme of the present application will be described clearly and completely below in combination with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0051] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0052] The concept of this application is to provide a training method for a blockchain-based personal loan amount assessment model based on blockchain technology and the idea of ​​split learning, based on the problem that the existing technology cannot break the technical barriers to user information sharing among multiple institutions. This method enables multiple participants to jointly train and establish a personal loan amount assessment model based on blockchain without leaking customer privacy data.

[0053] Figure 1 The schematic diagram of the block chain structure according to an embodiment of the present application is shown, but the implementation of the present application is not limited to Figure 1 The blockchain shown can be any blockchain system that can implement the training method of the personal loan amount assessment model provided in this application.

[0054] from Figure 1 As can be seen, Figure 1 The blockchain 100 shown includes a task issuing node 101 and multiple training nodes, one of which is a central training node 102, and the remaining training nodes are participating training nodes 103. Task issuing node 101 is in communication with each training node, and central training node 102 is in communication with each participating training node 103.

[0055] It should be noted that each training node can serve as a central training node or a participating training node. This application does not impose any restrictions. This indicates that during model training, the central training node and participating training nodes have different responsibilities. The determination of which training node is the central training node will be made during the execution of this application.

[0056] Figure 2 The training method of the personal loan amount assessment model based on blockchain in one embodiment of the present application is shown. Figure 2 It can be seen that this application includes at least steps S210 to S240:

[0057] Step S210: Each training node writes its own training information into the blockchain according to the training invitation sent by the task issuing node, where the training information includes: training capability information and user information.

[0058] First, the task issuing node sends a training invitation to each training node that wants to participate in the co-construction of the personal loan amount assessment model. In some embodiments of the present application, each training node may include banks, financial institutions, medical institutions, consumer enterprises, etc. These training nodes are respectively referred to as bank training nodes, financial institution training nodes, medical training nodes, and consumer training nodes.

[0059] After receiving the training invitation from the task issuing node, each training node writes its own training information into the blockchain. Blockchain technology can achieve data traceability and tamper-proofing. Writing this information into the blockchain as an endorsement for model training ensures the reliability of the data. Furthermore, it ensures the accuracy of the data used for model training, and thus ensures the accuracy of the trained personal loan amount assessment model.

[0060] In some embodiments of the present application, the training information of each training node includes, but is not limited to, training capability information and user information. The training capability information includes the computing power and storage capacity of each training node, specifically, the number of CPU cores of the training node's server, the server's cache size, and the memory size. User information includes, but is not limited to, user information stored locally by each participant, such as personal credit information, basic personal information, loan information, deposit information, and expenditure flow information provided by a bank training node; personal financial product purchase information provided by a financial institution training node; personal medical records, hospitalization information, and physical examination information provided by a medical training node; and personal consumption information and product purchase information provided by consumer companies.

[0061] In some embodiments of the present application, each training node may also sort the local user information in order of identity ID size, and transmit the last two digits of the identity ID to the central bank for secondary confirmation before model training. Based on the feedback information from the central bank, the training data is distinguished between true and false to ensure the accuracy of the data.

[0062] Step S220: The task issuing node determines one of the multiple training nodes as the central training node based on the training information of each training node, and sends the training task to the central training node, while the rest serve as participating training nodes.

[0063] The task issuing node selects one node from multiple training nodes as the central training node, and treats all training nodes except the central training node as ordinary participating training nodes.

[0064] In some embodiments of the present application, the selection of the central training node can be random, or the training node with the strongest training capability can be selected as the central training node according to the training capability information in the training information of each training node, and the central training node coordinates the entire training process. The central training node needs to maintain a communication connection with other training nodes participating in the training.

[0065] After determining the central training node, the task publishing node sends the training task to the central training node, wherein the training task includes the architecture of the personal loan amount evaluation model that has not been trained. In some embodiments of the present application, the architecture of the personal loan amount evaluation model is multi-layered.

[0066] Step S230: According to the training capability information of each training node, the central training node divides the training task into multiple training sub-tasks, and assigns each training sub-task to a corresponding training node, so that each training node trains based on its own user information according to the assigned training sub-task to obtain a partial training model.

[0067] After obtaining the training task, the central training node divides the entire training task into multiple training sub-tasks. Then, the central training node assigns each training sub-task to each training node, and assigns one training sub-task to one training node.

[0068] The division of the training sub-tasks can be based on the training capability information of each training node. For training nodes with strong training capability, a training sub-task with high training capability requirement can be assigned to them. Similarly, for training nodes with weak training capability, a training sub-task with low training capability requirement can be assigned to them, thereby achieving flexible adjustment and rational utilization of the computing resources of each training node.

[0069] In the case of a multi-layer architecture of the personal loan amount evaluation model, the specific division method of the training sub-tasks can be to divide one or more adjacent training layers of the personal loan amount evaluation model into one training sub-task, so that one training node only needs to train one model training layer or several model training layers, and does not need to train the entire model, thereby greatly avoiding the situation that a training node cannot complete the training due to insufficient computing capability.

[0070] Each training node trains based on its own local user information according to the assigned training sub-task to obtain a partial training model. In this way, the data of each participant can be used to train the model without the need for data sharing.

[0071] For the central training node, it can participate in the training or not participate in the training. In some embodiments of the present application, usually only the input layer and the output layer of the personal loan amount evaluation model are deployed in the central training node, and the input layer and the output layer also need to be trained.

[0072] For the specific method of model training, reference can be made to the machine learning method in the prior art, and no limitation is applied.

[0073] Each participating training node obtains a part of the personal loan amount evaluation model through training, which is recorded as a part of the training model. Each participating training node can write the obtained part of the training model to the blockchain.

[0074] In step S240, the central training node integrates the part of the training model obtained by each participating training node and the input layer and the output layer deployed in the central training node to obtain the personal loan amount evaluation model.

[0075] Finally, the central training node integrates the part of the training model obtained by each participating training node and the input layer and the output layer deployed in the central training node to obtain the entire personal loan amount evaluation model, which can be used for personal loan amount evaluation.

[0076] For obtaining the part of the training model, the central training node can obtain it from each participating training node or from the blockchain. For the specific method of integration, the central training node can adjust the initial personal loan amount evaluation model parameters to be consistent with the parameters of each part of the training model.

[0077] In some embodiments of the present application, the central training node performs evaluation after integration into the personal loan amount evaluation model. If the accuracy meets the preset requirement, the training is ended to obtain the final personal loan amount evaluation model. If the accuracy does not meet the preset requirement, multiple training can be performed until the final preset requirement is met.

[0078] By Figure 2The method shown can be seen that the present application is based on the idea of blockchain technology blockchain and split learning, splits the training task of the whole personal loan amount evaluation model into multiple training sub-tasks, and enables the participants participating in the co-construction of the personal loan amount evaluation model to train according to the sub-tasks based on the local user personal information of each participant, and finally integrates the partial training models obtained by training of each participant to obtain the whole personal loan amount evaluation model. The present application can coordinate multiple participants to co-construct the personal loan amount evaluation model without revealing the client privacy data, the data and related information of each participant are all chained, traceability and tamper resistance are realized, the local data of each participant is not trained outside the local area, the privacy is ensured, and the demand of the model training task is met; and the present application allocates the training task according to the ability of each participant, so that the situation that the participant cannot complete the model training due to insufficient ability does not occur, and the situation that the participant has redundant ability due to excessive ability does not occur; in addition, the present application can dynamically change according to the participants and the training task demand, and flexibly adjust the deployment and training of the training model.

[0079] In some embodiments of the present application, the above method further comprises: the central training node sends the obtained personal loan amount evaluation model to each training participating node, so that each training participating node evaluates the personal loan amount based on the personal loan amount evaluation model.

[0080] The personal loan amount evaluation model can be deployed in systems such as banks, financial institutions, etc., for evaluating the personal loan amount.

[0081] For the evaluation of the personal loan amount, the staff can input the personal information of the user who wants to take a loan, and the personal loan amount evaluation model can automatically simulate the personal loan amount through the personal information.

[0082] In some embodiments of the present application, the task publishing node determines one from the multiple training nodes as a central training node according to the training information of each training node, comprising: the task publishing node determines the training evaluation score of each training node according to the training task and the training ability information of each training node; and the training node with the lowest training evaluation score is determined as the central training node.

[0083] The central training node needs to be responsible for the overall integration work of the whole training process, and therefore it is best to have certain computing power and storage capacity, and has the highest matching degree with the training task.

[0084] When a task issuing node selects a central training node from multiple training nodes, how to select the most suitable central training node? Some embodiments of the present application recommend the following method: using a training evaluation score to represent the training node, where a lower training evaluation score indicates that the training node is more suitable as a central training node, and a higher training evaluation score indicates that the training node is less suitable as a central training node.

[0085] The following recommends a scoring method for training evaluation scores. In some embodiments of the present application, the training evaluation score includes two parts. One part is the computing power requirement score and storage capacity requirement score of the training task. The larger the computing power requirement score or the storage capacity requirement score, the greater the demand of the training task in this aspect; the other part is the computing power score and storage capacity score of the training node. The larger the computing power score or the storage capacity score, the stronger the training node's ability in this aspect.

[0086] The task issuing node determines the training evaluation score of each training node based on the training task and the training capacity information of each training node, including: determining the computing power requirement score and storage capacity requirement score of the training task, wherein the sum of the computing power requirement score and the storage capacity requirement score is one; determining the computing power score and storage capacity score of each training node, wherein the sum of the computing power score and the storage capacity score is one; determining the first product of the computing power requirement score and the computing power score, and the second product of the storage capacity requirement score and the storage capacity score, and taking the sum of the first product and the second product as the training evaluation score of each training node. Since the above scores are

[0087] The lower the training evaluation value of a training node, the higher its matching degree with the training task. Finally, the training node with the lower training evaluation value is selected as the central training node.

[0088] In some embodiments of the present application, the training task includes a personal loan amount assessment model with a multi-layer architecture; the central training node divides the training task into multiple training sub-tasks based on the training capability information of each participating training node, including: the central training node determines the training evaluation score of each participating training node based on the training task and the training capability information of each participating training node; determines the number of training layers of each participating training node based on the training evaluation score of each participating training node and the total number of model layers of the personal loan amount assessment model; and divides the personal loan amount assessment model into multiple training sub-tasks according to the number of training layers of each participating training node.

[0089] When the central training node divides the training task into subtasks, it can also refer to the above-mentioned "training evaluation score" method. First, determine the training evaluation score of a participating training node. The training evaluation score is a decimal greater than zero and less than 1. Multiply the training evaluation score by the total number of model layers of the personal loan amount assessment model to obtain the number of model layers that can be trained by the participating training node. According to the calculated number of model layers, the personal loan amount assessment model is divided into multiple training subtasks. A training subtask includes one or more layers. It should be noted that during the calculation, if the calculated number of model layers that can be trained by a participating training node is not an integer, the number of model layers of the participating training node can be calculated by rounding.

[0090] The training evaluation scores of the nodes participating in the training can be determined using the aforementioned method, which will not be repeated here.

[0091] In some embodiments of the present application, each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model, including: each participating training node uses the output of the previous participating training node as the input of the subsequent participating training node according to the order of the model training layers in the assigned training subtask in the personal loan amount assessment model, and performs training based on the local user information of each participating training node to obtain a partial training model corresponding to each participating training node.

[0092] During task allocation, or model deployment, the input and output layers of the personal loan amount assessment model can be deployed on a central training node, which trains the input and output layers. Specifically, the central training node can input local user information, including but not limited to basic personal information, credit information, personal loan information, personal deposit information, and personal expenditure flow information, into the input layer for training. Other participating training nodes can randomly sort and sequentially obtain the corresponding number of untrained personal loan amount assessment model structures for training.

[0093] During training, the training is performed according to the order of the participating training nodes, that is, the order of the personal loan amount assessment model architecture. The output of the previous participating training node is used as the input of the next participating training node. The next participating training node receives the output of the previous participating training node, adds the user information in the local data to the output data, and then uses it as the input of the participating training node for training. In some embodiments of the present application, training can be repeated multiple times until the final evaluation criteria are met, at which point training ends.

[0094] Figure 3A training device for a personal loan amount assessment model based on blockchain according to an embodiment of the present application is shown. The blockchain includes a task issuing node and multiple training nodes. The training device is deployed on each node of the blockchain ( Figure 1 101, 102 and 103), from Figure 3 It can be seen that the device 300 includes:

[0095] A writing unit 310 is configured to write respective training information into the blockchain according to the training invitation sent by the task issuing node, wherein the training information includes training capability information and user information;

[0096] The task issuing unit 320 is configured to determine one of the plurality of training nodes as a central training node based on the training information of each training node, and send the training task to the central training node, with the remaining nodes serving as participating training nodes;

[0097] The task assignment unit 330 is configured to divide the training task into a plurality of training subtasks according to the training capability information of each participating training node, and assign each training subtask to a corresponding participating training node, so that each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model;

[0098] The integration unit 340 is used to integrate the partial training models obtained by each participating training node and the input layer and output layer deployed on the central training node to obtain a personal loan amount assessment model.

[0099] In some embodiments of the present application, the above-mentioned device also includes: a sending unit, used to send the obtained personal loan amount evaluation model to each participating training node, so that each participating training node evaluates the personal loan amount based on the personal loan amount evaluation model.

[0100] In some embodiments of the present application, in the above-mentioned device, the task issuing unit 320 is used to determine the training evaluation score of each training node based on the training task and the training capability information of each training node; and the training node with the lowest training evaluation score is used as the central training node.

[0101] In some embodiments of the present application, in the above device, the task publishing node 320 is configured to determine an algorithm requirement score and a storage capacity requirement score of the training task, wherein the sum of the algorithm requirement score and the storage capacity requirement score is one; determine an algorithm score and a storage capacity score of each training node, wherein the sum of the algorithm score and the storage capacity score is one; determine a first product of the algorithm requirement score and the algorithm score, and a second product of the storage capacity requirement score and the storage capacity score, and take the sum of the first product and the second product as a training evaluation score of each training node.

[0102] In some embodiments of the present application, in the above device, the training task includes a personal loan amount evaluation model of a multi-layer architecture; the task allocation unit 330 is configured to determine a training evaluation score of each participating training node according to the training task and the training capability information of each participating training node; determine a training layer number of each participating training node according to the training evaluation score of each participating training node and the total model layer number of the personal loan amount evaluation model; and divide the personal loan amount evaluation model into a plurality of training sub-tasks according to the training layer number of each participating training node.

[0103] In some embodiments of the present application, the above device further includes a training unit configured to take the output of a preceding participating training node as the input of a following participating training node in the order of the model training layers in the personal loan amount evaluation model, and train based on the local user information of each participating training node to obtain a partial training model corresponding to each participating training node.

[0104] In some embodiments of the present application, in the above device, the plurality of training nodes include a bank training node, a financial institution training node, a medical training node, and a consumption training node.

[0105] In some embodiments of the present application, in the above device, the bank training node provides personal credit information, personal basic information, personal lending information, personal deposit information, and personal expenditure flow information; the financial institution training node provides personal financial product purchase information; the medical training node provides personal case information, hospitalization information, and physical examination information; and the consumption enterprise provides personal consumption information and purchase product information.

[0106] Figure 4 FIG. 1 is a structural schematic diagram of an electronic device according to an embodiment of the present application. Please refer to Figure 4At the hardware level, the electronic device includes a processor, and optionally further includes an internal bus, a network interface, and a memory. The memory can include a memory such as a random-access memory (RAM), and can further include a non-volatile memory such as at least one disk memory. Of course, the electronic device can further include other hardware required by a business.

[0107] The processor, the network interface, and the memory can be connected to each other through the internal bus, which can be an industry standard architecture (ISA) bus, a peripheral component interconnect (PCI) bus, an extended industry standard architecture (EISA) bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, Figure 4 Only one bidirectional arrow is used in the figure to represent the bus, but it does not mean that there is only one bus or only one type of bus.

[0108] The memory is used to store a program. Specifically, the program can include program code including computer operation instructions. The memory can include a memory and a non-volatile memory, and provide instructions and data to the processor.

[0109] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs, and forms a training device of a personal loan amount evaluation model based on a block chain at the logical level. The processor executes the program stored in the memory, and is specifically used to perform the following operations:

[0110] Each training node writes its own training information into the block chain according to the training invitation sent by the task publishing node, wherein the training information includes training capability information and user information;

[0111] The task publishing node determines one of the plurality of training nodes as a central training node according to the training information of each training node, and sends a training task to the central training node, and the rest as participating training nodes;

[0112] The central training node divides the training task into a plurality of training subtasks according to the training capability information of each participating training node, and assigns each training subtask to a corresponding participating training node, so that each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model;

[0113] The central training node integrates the partial training models obtained by each participating training node, as well as the input layer and output layer deployed on the central training node, to obtain a personal loan amount assessment model.

[0114] The above application Figure 3 The method performed by the training device for the blockchain-based personal loan amount assessment model disclosed in the illustrated embodiment can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be performed by hardware integrated logic circuits within the processor or by software instructions. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules within the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method.

[0115] The electronic device may also perform Figure 3 The method executed by the training device of the blockchain-based personal loan amount assessment model is realized as the function of the embodiment of the training device of the blockchain-based personal loan amount assessment model shown in Figure 3, and the embodiments of this application are not repeated here.

[0116] The embodiments of the present application also provide a computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by an electronic device comprising a plurality of applications, can cause the electronic device to perform Figure 3 The method executed by the training device of the personal loan amount evaluation model based on the blockchain in the illustrated embodiment, and is specifically used for executing:

[0117] Each training node writes the training information of each training node into the blockchain according to the training invitation sent by the task publishing node, wherein the training information comprises training capability information and user information;

[0118] The task publishing node determines one of the plurality of training nodes as a central training node according to the training information of each training node, and sends the training task to the central training node, and the rest are participating training nodes;

[0119] The central training node splits the training task into a plurality of training sub-tasks according to the training capability information of each participating training node, and allocates each training sub-task to the corresponding participating training node, so that each participating training node performs training based on the allocated training sub-task and the user information of each participating training node to obtain a partial training model;

[0120] The central training node integrates the partial training models obtained by each participating training node and the input layer and output layer deployed in the central training node to obtain a personal loan amount evaluation model.

[0121] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0122] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0123] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0124] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0125] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0126] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0127] Computer-readable media includes permanent and non-permanent, movable and non-movable media that can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules 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 technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0128] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to encompass non-exclusive inclusions, such that a process, method, article or apparatus that comprises a list of elements does not include only those elements in the list, but can also include other elements not expressly listed or inherent to such process, method, article or apparatus. Without more limitations, the element defined by the phrase "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.

[0129] Those skilled in the art will appreciate that embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) containing computer usable program code.

[0130] The above only describes the embodiments of the present application and is not intended to limit the present application. Those skilled in the art can make various modifications and changes to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the scope of the claims of the present application.

Claims

1. A training method for a personal loan amount assessment model based on blockchain, characterized in that: The blockchain includes a task issuing node and multiple training nodes, and the method includes: Each training node writes its own training information into the blockchain according to the training invitation sent by the task issuing node, wherein the training information includes: training capability information and user information; The task issuing node determines one of the multiple training nodes as a central training node based on the training information of each training node, and sends the training task to the central training node, while the rest serve as participating training nodes; The central training node divides the training task into a plurality of training subtasks according to the training capability information of each participating training node, and assigns each training subtask to a corresponding participating training node, so that each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model; The central training node integrates the partial training models obtained by each participating training node, as well as the input layer and output layer deployed on the central training node, to obtain a personal loan amount assessment model; Each participating training node performs training based on the assigned training subtask and its own user information to obtain a partial training model, including: Each participating training node is trained in the order of the model training layers in the assigned training subtasks in the personal loan amount assessment model, adding the local user information of the subsequent participating training node to the output of the previous participating training node and using it as the input of the subsequent participating training node to obtain the partial training model corresponding to each participating training node.

2. The method according to claim 1, characterized in that The method further comprises: The central training node sends the obtained personal loan amount evaluation model to each participating training node, so that each participating training node evaluates the personal loan amount based on the personal loan amount evaluation model.

3. The method according to claim 1, characterized in that The task issuing node determines one of the plurality of training nodes as a central training node based on the training information of each training node, including: The task issuing node determines a training evaluation score for each training node based on the training task and the training capability information of each training node; The training node with the lowest training evaluation score is used as the central training node.

4. The method according to claim 3, characterized in that The task issuing node determines the training evaluation score of each training node based on the training task and the training capability information of each training node, including: Determining a computing power requirement score and a storage capacity requirement score for the training task, wherein the sum of the computing power requirement score and the storage capacity requirement score is 1; Determining a computing power score and a storage capacity score for each training node, where the sum of the computing power score and the storage capacity score is 1; Determine a first product of the computing power requirement score and the computing power score, and a second product of the storage capacity requirement score and the storage capacity score, and use the sum of the first product and the second product as the training evaluation score of each training node.

5. The method according to claim 1, wherein The training task includes a multi-layered personal loan amount assessment model; The central training node divides the training task into multiple training subtasks based on the training capability information of each participating training node, including: The central training node determines a training evaluation score for each participating training node based on the training task and the training capability information of each participating training node; Determine the number of training layers for each participating training node based on the training evaluation score of each participating training node and the total number of model layers of the personal loan amount evaluation model; The personal loan amount assessment model is divided into multiple training subtasks according to the number of training layers of each participating training node.

6. The method according to any one of claims 1 to 5, characterized in that The multiple training nodes include a bank training node, a financial institution training node, a medical training node, and a consumer training node.

7. The method according to claim 6, characterized in that The user information includes: personal credit information, personal basic information, personal loan information, personal deposit information, and personal expenditure flow information provided by the bank training node; Personal financial product purchase information provided by financial institution training nodes; Personal medical records, hospitalization information, and physical examination information provided by medical training nodes; The consumption training node provides personal consumption information and purchased product information.

8. A training device for a personal loan amount assessment model based on blockchain, characterized in that: The blockchain includes a task issuing node and multiple training nodes. The training device is deployed in each node of the blockchain, and the device includes: a writing unit, configured to write respective training information into the blockchain according to the training invitation sent by the task issuing node, wherein the training information includes training capability information and user information; a task issuing unit, configured to determine one of the plurality of training nodes as a central training node based on the training information of each training node, and send the training task to the central training node, with the remaining nodes serving as participating training nodes; a task assignment unit, configured to divide the training task into a plurality of training subtasks according to the training capability information of each participating training node, and assign each training subtask to a corresponding participating training node, so that each participating training node performs training based on its own user information according to the assigned training subtask to obtain a partial training model; An integration unit, configured to integrate the partial training models obtained by each participating training node, and the input layer and output layer deployed on the central training node, to obtain a personal loan amount assessment model; It also includes a training unit for adding local user information of subsequent participating training nodes to the output of previous participating training nodes in the order of the model training layers in the assigned training subtasks in the personal loan amount assessment model, and then training them as input of the subsequent participating training nodes to obtain partial training models corresponding to each participating training node.

Images