Implementation Method, Device, Electronic Device and Storage Medium for Collaborative Modeling

Abstract

An embodiment of the present invention discloses a method, device, electronic device, and storage medium for realizing collaborative modeling. The method includes: obtaining a current collaborative end model through evolutionary computation; sending, based on a collaborative end modeling sample, a current model calculation result output by the current collaborative end model to an initiating end of collaborative modeling, so that the initiating end evaluates the current collaborative end model based on the current model calculation result; controlling whether to perform evolutionary computation on the current collaborative end model according to the evaluation result of the current collaborative end model obtained from the initiating end, to generate a new collaborative end model for collaborative modeling, until the initiating end obtains a collaborative end model that meets the modeling requirements. By adopting the solution of this application, with evolutionary computation as the main framework, enabling the collaborative end to send the model calculation result to the initiating end and the initiating end to send the evaluation result to the collaborative end, collaborative modeling between the two parties is achieved by exchanging statistics without the need for the collaborative end and the initiating end to interact with the original data of the modeling sample.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of artificial intelligence technology, and in particular, to a method, device, electronic device, and storage medium for realizing collaborative modeling. Background Art

[0002] In the process of joint modeling, due to data security supervision requirements, especially in the financial field, the original modeling data of both parties in joint modeling need to be kept confidential before joint modeling can be carried out.

[0003] Currently, the commonly used technical solution in the industry is the federated learning framework, that is: based on the homomorphic encryption algorithm, the modeling sample data is encrypted, and the encrypted result is calculated and model training is carried out, and finally an available model is obtained. However, in the process of joint modeling, it is necessary to transform the encryption algorithm for the model training algorithm, data statistics algorithm, etc., which will increase the amount of calculation, and thus make it more difficult to implement the technology on the ground. Summary of the Invention

[0004] Embodiments of the present invention provide a method, device, electronic device, and storage medium for realizing collaborative modeling, so as to realize joint modeling without encrypting and transforming the existing modeling algorithms.

[0005] In a first aspect, embodiments of the present invention provide a method for realizing collaborative modeling, which is executed by a collaborative end of collaborative modeling. The method includes:

[0006] Obtaining the current collaborative end model through evolutionary computation;

[0007] Based on the collaborative end modeling samples, sending the current model calculation result output by the current collaborative end model to the initiating end of collaborative modeling, so that the initiating end evaluates the current collaborative end model based on the current model calculation result;

[0008] Controlling whether to perform evolutionary computation on the current collaborative end model according to the evaluation result of the current collaborative end model obtained from the initiating end, to generate a new collaborative end model for collaborative modeling until the initiating end obtains a collaborative end model that meets the modeling requirements.

[0009] In a second aspect, embodiments of the present invention also provide a device for realizing collaborative modeling, which is configured at the collaborative end of collaborative modeling. The device includes:

[0010] A model evolutionary computation module, configured to obtain the current collaborative end model through evolutionary computation;

[0011] A model result sending module, configured to send the current model calculation result output by the current collaborative end model to the initiating end of collaborative modeling based on the collaborative end modeling samples, so that the initiating end evaluates the current collaborative end model based on the current model calculation result;

[0012] A model evaluation processing module, configured to control whether to perform evolutionary computation on the current collaborative end model according to the evaluation result of the current collaborative end model obtained from the initiating end, generate a new collaborative end model for collaborative modeling until the initiating end obtains a collaborative end model that meets the modeling requirements.

[0013] In a third aspect, an embodiment of the present invention further provides an electronic device, including:

[0014] One or more processors;

[0015] A storage device for storing one or more programs;

[0016] The one or more programs are executed by the one or more processors, so that the one or more processors implement the collaborative modeling implementation method provided in any embodiment of the present invention.

[0017] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the collaborative modeling implementation method provided in any embodiment of the present invention.

[0018] An embodiment of the present invention provides a collaborative modeling implementation method. The collaborative end of collaborative modeling obtains the current collaborative end model through evolutionary computation, and sends the current model calculation result output by the current collaborative end model to the initiating end of collaborative modeling based on the collaborative end modeling sample, so that the initiating end evaluates the current collaborative end model based on the current model calculation result to obtain an evaluation result; furthermore, the collaborative end will determine whether to perform evolutionary computation on the current collaborative end model according to the evaluation result of the initiating end on the current collaborative end model, and continue to optimize to generate a new collaborative end model for collaborative modeling. Adopting the solution of this application, with evolutionary computation as the main framework, the collaborative end sends the model calculation result to the initiating end and the initiating end sends the evaluation result to the collaborative end. There is no need to encrypt the modeling algorithm, and without the need for the collaborative end and the initiating end to exchange the original data of the modeling sample, collaborative modeling between the two parties is achieved by exchanging statistics.

[0019] The above invention content is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present invention more obvious and understandable, the following specifically illustrates the specific implementation manners of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments read in conjunction with the accompanying drawings. The drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Also, throughout the drawings, the same reference numerals are used to denote the same components. In the drawings:

[0021] Figure 1 is a flowchart of an implementation method for collaborative modeling provided in an embodiment of the present invention;

[0022] Figure 2a is an interaction schematic diagram for collaborative modeling between a collaborative end and an initiating end provided in an embodiment of the present invention;

[0023] Figure 2b is another interaction schematic diagram for collaborative modeling between a collaborative end and an initiating end provided in an embodiment of the present invention;

[0024] Figure 3 is a flowchart of another implementation method for collaborative modeling provided in an embodiment of the present invention;

[0025] Figure 4 is a structural block diagram of an implementation device for collaborative modeling provided in an embodiment of the present invention;

[0026] Figure 5 is a structural schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Embodiments

[0027] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It can be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention. Additionally, it should be noted that, for the sake of convenience of description, only parts related to the present invention rather than all structures are shown in the drawings.

[0028] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts depict the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operations are completed, but there can also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0029] Before understanding the technical solution of this application, the existing federated learning framework is elaborated. Currently, the existing joint modeling technology is mainly based on the homomorphic encryption algorithm. Among them, the homomorphic encryption algorithm needs to encrypt the original data during the model calculation process and perform calculations based on the encrypted data. This requires optimizing the existing model training algorithm to support the calculations after homomorphic encryption. Moreover, during model training, for some encrypted data, due to homomorphic encryption exceeding the limit, it needs to be decrypted and re-encrypted. This not only increases the computational amount but also requires decryption and then re-encryption. When the training data is large, for the encryption and decryption service requests, the service response pressure is relatively large, and the so-called "network flood" may occur during the process of training the model.

[0030] Based on the homomorphic encryption algorithm, some data still needs to be calculated after being decrypted in the memory, which requires a relatively fair and trustworthy third-party service, and has relatively high requirements for the third-party technology. This not only increases the technical complexity of joint modeling but also brings up the problem of trustworthy authentication service for the third party. Therefore, how to overcome at least part of the above problems during the collaborative modeling process becomes particularly important.

[0031] The following elaborates in detail on the implementation methods, devices, electronic devices, and storage media for collaborative modeling provided in the embodiments of the present invention through various embodiments and optional solutions of each embodiment.

[0032] Figure 1 It is a flowchart of an implementation method for collaborative modeling provided in the embodiments of the present invention. The technical solution of this embodiment can be applied to the situation of joint collaborative modeling among multiple parties. This method can be executed by an implementation device for collaborative modeling. The device can be implemented by software and / or hardware and integrated on any electronic device with network communication functions. Among them, the electronic device can be a collaborative end device for collaborative modeling, etc.

[0033] As Figure 1 shown, the implementation method for collaborative modeling in the embodiments of the present invention may include the following steps:

[0034] S110. Obtain the current collaborative end model through evolutionary computation.

[0035] In this embodiment, evolutionary computation is influenced by the natural selection mechanism of "survival of the fittest" and the transmission law of genetic information during the biological evolution process. Evolutionary computation is a class of stochastic search optimization algorithms that simulate the principles of biological evolution and heredity. By continuously iterating and simulating this optimization process, among a population composed of some possible solutions, the optimal solution is sought through natural evolution. Based on the above theory, this application introduces "evolutionary computation" into the multi-party collaborative modeling process, and takes evolutionary computation as the main framework at the collaborative end to construct a collaborative modeling framework between the collaborative end and the initiating end for collaborative modeling.

[0036] In this embodiment, the solution of the present application can be applied to general modeling scenarios in the financial field, etc. For example, taking the credit risk control scenario in the financial field as an example, in order to meet the requirements for customer credit risk control, a customer credit evaluation model can be established. On the basis of ensuring the interpretability of the model, the customer credit rating can be predicted through the constructed credit evaluation model to distinguish good customers from bad customers. Of course, in order to build a credit evaluation model that meets the requirements, it is usually necessary to carry out joint collaborative modeling among multiple parties to ensure the reliability of the model.

[0037] In this embodiment, Figure 2a is an interaction schematic diagram for collaborative modeling between a collaborative end and an initiating end provided in an embodiment of the present invention. Figure 2b is another interaction schematic diagram for collaborative modeling between a collaborative end and an initiating end provided in an embodiment of the present invention. Refer to Figure 2a and Figure 2b After the initiating end of the collaborative modeling sends a collaborative modeling request to the collaborative end, the collaborative end can obtain the current collaborative end model required for the current round of collaborative modeling by performing evolutionary calculation on the specified model in the collaborative end. Among them, the current collaborative end model is a rough model established at the collaborative end during the collaborative modeling process, rather than the final required model, and a better model needs to be optimized through multiple rounds of collaborative modeling operations through evolutionary calculation.

[0038] S120. Based on the collaborative end modeling sample, send the current model calculation result output by the current collaborative end model to the initiating end of the collaborative modeling, so that the initiating end can evaluate the current collaborative end model based on the current model calculation result.

[0039] In this embodiment, refer to Figure 2a and Figure 2b The collaborative end includes a collaborative end modeling sample. After obtaining the current collaborative end model through evolutionary calculation, the collaborative end modeling sample can be input into the current collaborative end model, and the current model calculation result can be output through the calculation of the current collaborative end model. Furthermore, the collaborative end can send the current model calculation result to the initiating end, and the initiating end will evaluate the current collaborative end model established by the collaborative end based on the current model calculation result, so that the collaborative end can continue to optimize the established model and generate a better model.

[0040] In the above manner of this embodiment, during each round of evolutionary computation in collaborative modeling, for the initiating end, the current model computation result output by the current collaborative end model of the collaborative end is a black-box model result data. Since the initiating end does not know how many features, what features, how the features are combined, and the value ranges of the features are used by the collaborative end to output the current model computation result, it is impossible to confirm the specific details of the data provided by the initiating end. Therefore, the security of the data provided by the collaborative end to the initiating end is ensured. Moreover, during the collaborative modeling process, it is not necessary for the collaborative end to perform encryption algorithm transformation on the model training algorithm, data statistics algorithm, etc., avoiding increasing the technical implementation difficulty and significantly increasing the computation amount.

[0041] In this embodiment, referring to Figure 2a and Figure 2b , the initiating end can evaluate the current collaborative end model obtained by the collaborative end through evolutionary computation based on the initiating end modeling samples and the current collaborative end model computation sent by the collaborative end, and obtain the evaluation result of the current collaborative end model under the target evaluation index. Optionally, the initiating end can select the target evaluation index according to requirements. For example, the target evaluation index can include indicators such as model discrimination and model precision measurement.

[0042] In the above manner of this embodiment, after the initiating end obtains the collaborative end model computation result, without any encryption modification, it can combine the collaborative end model computation result of the collaborative end to calculate the evaluation result under relevant statistical evaluation indicators, and return the evaluation result to the collaborative end. During this process, since the collaborative end only obtains the evaluation result of the initiating end, the evaluation result of the initiating end is also black-box data to the collaborative end. The collaborative end also does not know how many features, what features, how the features are combined, and the value ranges of other features are used by the initiating end, and it is also impossible to confirm the sample data details of the initiating end. Therefore, the security of the data provided by the initiating end to the collaborative end is ensured. Moreover, after the initiating end obtains the collaborative end model computation result of the collaborative end, there is no need for decryption and re-encryption, so the computation amount will not be significantly increased.

[0043] S130. According to the evaluation result of the current collaborative end model obtained from the initiating end, control whether to perform evolutionary computation on the current collaborative end model to generate a new collaborative end model for collaborative modeling until the initiating end obtains a collaborative end model that meets the modeling requirements.

[0044] In this embodiment, referring to Figure 2a and Figure 2b , after determining the evaluation result of the current collaborative end model obtained by the collaborative end, the initiating end will send the evaluation result of the current collaborative model to the collaborative end. The collaborative end can judge whether to perform the next round of evolutionary computation on the current collaborative end model obtained in the current round according to the evaluation result and enter the next round of evolutionary computation process, so that the collaborative end can generate a new collaborative end model.

[0045] In this embodiment, referring to Figure 2a and Figure 2b , in each round of collaborative modeling process, when the collaborative end receives the evaluation result of the collaborative end model sent by the initiating end, the collaborative end determines whether to continue the next round of evolutionary calculation operation according to the evaluation result, and optimizes to generate a better collaborative end model until a collaborative end model satisfactory to the initiating end is obtained. Through the data exchange mode between the collaborative end and the initiating end, the security of the basic data of both parties is guaranteed, and at the same time, the collaborative end will gradually obtain a collaborative end model suitable for the modeling requirements of the initiating end based on evolutionary calculation.

[0046] In the embodiment of the present invention, a method for realizing collaborative modeling is provided. With evolutionary calculation as the main framework, the collaborative end sends the model calculation result to the initiating end and the initiating end sends the evaluation result to the collaborative end. There is no need to encrypt the modeling algorithm, and in the case where the collaborative end and the initiating end do not need to interact with the original data of the modeling sample, collaborative modeling between the two parties is realized by exchanging statistics. It can be seen that the collaborative modeling framework of this application achieves the effects of not requiring encryption transformation of the existing modeling algorithm, not requiring data exchange with encryption and decryption, and not requiring the participation of a third-party trusted institution, and has the advantage of relatively simple technical implementation difficulty.

[0047] Figure 3 is a flowchart of another method for realizing collaborative modeling provided in the embodiment of the present invention. The embodiment of the present invention further optimizes the foregoing embodiment on the basis of the above embodiment, and the embodiment of the present invention can be combined with each optional solution in one or more of the above embodiments. As Figure 3 shown, the method for realizing collaborative modeling provided in this embodiment may include the following steps:

[0048] S310. Determine the model to be processed currently retained at the collaborative end.

[0049] In an optional solution of this embodiment, determining the model to be processed currently retained at the collaborative end may include the following steps A1 - A2:

[0050] Step A1. If the evolutionary calculation is performed for the first time after receiving the collaborative modeling request sent by the initiating end, use the collaborative end modeling sample for model training and randomly establish the initial model required by the collaborative end.

[0051] In this embodiment, when the evolutionary calculation is performed for the first time after receiving the collaborative modeling request sent by the initiating end, the modeling sample of the collaborative end is used for simple model training to obtain a random initial model of the collaborative end. This initial model is a rough model, which can be continuously optimized in subsequent multiple rounds of evolutionary calculation based on evolutionary calculation, and gradually obtain a collaborative end model suitable for the modeling requirements of the initiating end.

[0052] Step A2: Determine the initial model as the model to be processed currently reserved for use at the collaborative end.

[0053] In another alternative of this embodiment, determining the model to be processed currently reserved for use at the collaborative end may include the following steps B1 - B2:

[0054] Step B1: If it is not the first time to perform evolutionary computation after receiving the collaborative modeling request sent by the initiating end, obtain the evaluation result of the previous collaborative end model determined by the previous evolutionary computation.

[0055] Step B2: Based on the evaluation result of the previous collaborative end model, perform model screening and elimination on the previous collaborative end model to obtain the model to be processed.

[0056] In this embodiment, if it is not the first time to perform evolutionary computation after receiving the collaborative modeling request sent by the initiating end, it indicates that during the previous round of evolutionary computation, the collaborative end has obtained the previous collaborative end model through evolutionary computation. For the previous collaborative end model, in this round of evolutionary computation, not all previous collaborative end models will be evolved because this will cause a large amount of computation and introduce collaborative end models that do not meet the requirements. Therefore, based on the evaluation result of the previous collaborative end model, the models that do not meet the modeling requirements in the previous collaborative end model can be screened and eliminated.

[0057] S320: Through evolutionary computation on the model to be processed, heuristically obtain a preset number of current collaborative end models; where the evolutionary computation includes at least one of the following operations: selection, crossover, and mutation genetic operations.

[0058] S330: Based on the collaborative end modeling samples, send the current model calculation results output by the current collaborative end model to the initiating end of the collaborative modeling, so that the initiating end can evaluate the current collaborative end model based on the current model calculation results.

[0059] In an alternative of this embodiment, based on the collaborative end modeling samples, sending the current model calculation results output by the current collaborative end model to the initiating end of the collaborative modeling may include the following steps C1 - C2:

[0060] Step C1: Input the collaborative end modeling samples into a preset number of current collaborative end models for calculation respectively to obtain the current model calculation results output by different current collaborative end models under the collaborative end modeling samples.

[0061] Step C2: Send the current model calculation results output by different current collaborative end models under the collaborative end modeling samples to the initiating end.

[0062] In this embodiment, refer to Figure 2a and Figure 2b, optionally, the collaborative end can batch compress the current model calculation results output by different current collaborative end models respectively, and send the batch compressed current model calculation results to the initiating end. In this way, the number of data exchanges for collaborative modeling based on evolutionary computation is related to the number of evolutionary computation batches. By interacting between the collaborative end and the initiating end through batch compressed files, the number of interactions can be significantly reduced, the request pressure on the collaborative end and the initiating end can be reduced, and the technical implementation is relatively simple.

[0063] In an alternative solution of this embodiment, evaluating the current collaborative end model based on the current model calculation result may include the following steps D1 - D2:

[0064] Step D1: Combine the initiating end Y label, the initiating end modeling sample, and the current model calculation results output by different current collaborative end models under the collaborative end modeling sample to obtain combined modeling samples under different current collaborative end models.

[0065] Step D2: According to the combined modeling samples under different current collaborative end models, evaluate different current collaborative end models respectively to obtain evaluation results for each current collaborative end model.

[0066] In this embodiment, referring to Figure 2a and Figure 2b , the initiating end of collaborative modeling is the ultimate decision - maker for model establishment. The initiating end can provide evaluation metrics and the Y label required for modeling. Optionally, as the ultimate decision - maker of the model, the initiating end can select the evaluation metrics of the model. For example, the evaluation metrics include: model discrimination, model precision measurement, etc. Optionally, the modeling samples of the initiating end are labeled through the Y label. For example, the customer credit level in the labeled samples is used to distinguish good and bad customers.

[0067] In this embodiment, referring to Figure 2a and Figure 2b , specifically evaluating different current collaborative end models may include: using a preset model evaluation algorithm, training models respectively based on the combined modeling samples under different current collaborative end models to obtain initiating end models under different current collaborative end models; determining the evaluation results of different current collaborative end models under the target evaluation metrics by evaluating the established different initiating end models. Among them, the preset model evaluation algorithm includes at least one of the following: logistic regression algorithm and xgboost algorithm.

[0068] In the above - mentioned manner of this embodiment, without the need to exchange the original modeling sample data of both parties, the modeling collaboration end provides the available model calculation result data to the initiating end. After the initiating end combines the Y - label, the initiating - end modeling sample data, and the available model calculation result data of the collaboration end, it evaluates the collaboration - end model obtained through evolutionary calculation by the collaboration end, obtains the model - related evaluation results, and returns the evaluation results to the collaboration end. The collaboration end continues to complete the evolutionary calculation based on the evaluation results, optimizes to generate a better collaboration - end model until a collaboration - end model satisfactory to the initiating end is obtained.

[0069] S340. According to the evaluation result of the current collaboration - end model obtained from the initiating end, control whether to perform evolutionary calculation on the current collaboration - end model, enter the next - round collaborative - modeling process of evolutionary calculation to generate a new collaboration - end model for collaborative modeling until the initiating end obtains a collaboration - end model that meets the modeling requirements.

[0070] In the embodiment of the present invention, a method for realizing collaborative modeling is provided. With evolutionary calculation as the main framework, the collaboration end sends model calculation results to the initiating end and the initiating end sends evaluation results to the collaboration end. There is no need to encrypt the modeling algorithm, and without the need to exchange the original modeling sample data between the collaboration end and the initiating end, collaborative modeling between the two parties is achieved by exchanging statistics. It can be seen that the collaborative - modeling framework of this application realizes the effects of not requiring encryption transformation of the existing modeling algorithm, not requiring data exchange through encryption and decryption, and not requiring the participation of a third - party trusted institution. It has the advantage of relatively simple technical implementation difficulty and can avoid the problems of relatively large data requests and interaction pressure in homomorphic - encryption - based modeling as much as possible.

[0071] Based on the above - mentioned embodiment, optionally, before obtaining the current collaboration - end model through evolutionary calculation, the following operations may further be included:

[0072] Encrypt the data index of the collaboration - end modeling sample, and perform a collision count with the encrypted result of the encrypted data index of the initiating - end modeling sample to align the collaboration - end modeling sample with the initiating - end modeling sample.

[0073] In this embodiment, referring to Figure 2a and Figure 2b , since the modeling samples of the collaboration end and the initiating end do not completely overlap, it is necessary to determine the common users of both parties without the collaboration end and the initiating end disclosing their respective modeling samples, and not expose the overlapping users of both parties to achieve sample alignment. Optionally, between the collaboration end and the initiating end, after encrypting the data index of their respective modeling samples with sh256, the modeling samples between the collaboration end and the initiating end can be aligned by means of "collision count" to ensure that both parties in the modeling have the same modeling samples.

[0074] In this embodiment, referring to Figure 2aand Figure 2b Optionally, after aligning the modeling samples between the collaborative end and the initiating end, the collaborative end can also perform data analysis on the aligned modeling samples of the collaborative end, specifically including: data analysis such as overall missing rate, sample distribution homomorphism index, etc., to facilitate the initiating end to preliminarily select its own modeling samples before modeling.

[0075] Based on the above embodiments, optionally, the implementation method of collaborative modeling in this embodiment may further include the following operations:

[0076] After the initiating end obtains the target collaborative end model that meets the modeling requirements, according to the target collaborative end model selected by the initiating end, send the business explanation of the target collaborative end model to the initiating end, so that the initiating end can select the collaborative end model for publication based on the business explanation of the target collaborative end model.

[0077] In this embodiment, refer to Figure 2b , in the model selection stage, the initiating end selects the corresponding collaborative end model according to the corresponding evaluation indicators, and obtains the business explanation of the collaborative end model by the collaborative end, and finally selects the target model that meets the modeling requirements. In the model reporting stage, the collaborative end can explain the generated target collaborative end model according to the specific target collaborative end model selected by the initiating end, and issue an explanatory note. In the model publishing stage, the initiating end selects the collaborative end model, publishes the combined model, and notifies the collaborative end to publish the selected collaborative end model based on the business explanation of the target collaborative end model.

[0078] Figure 4 is the structural block diagram of an implementation device for collaborative modeling provided in an embodiment of the present invention. The technical solution of this embodiment is applicable to the situation of joint collaborative modeling among multiple parties. This device can be implemented by software and / or hardware and integrated on any electronic device with network communication functions. Among them, this electronic device can be a collaborative end device for collaborative modeling, etc. As Figure 4 shown, the implementation device for collaborative modeling in the embodiment of the present invention may include the following: a model evolution calculation module 410, a model result sending module 420, and a model evaluation processing module 430. Among them:

[0079] The model evolution calculation module 410 is used to obtain the current collaborative end model through evolutionary calculation;

[0080] The model result sending module 420 is used to send the current model calculation result output by the current collaborative end model to the initiating end of the collaborative modeling based on the collaborative end modeling samples, so that the initiating end can evaluate the current collaborative end model based on the current model calculation result;

[0081] The model evaluation processing module 430 is configured to control whether to perform evolutionary computation on the current collaborative end model according to the evaluation result of the current collaborative end model obtained from the initiating end, generate a new collaborative end model for collaborative modeling, until the initiating end obtains a collaborative end model that meets the modeling requirements.

[0082] Based on the above embodiments, optionally, the model evolutionary computation module 410 includes:

[0083] The model to be processed determination unit is configured to determine the model to be processed that is currently retained and used at the collaborative end;

[0084] The model evolutionary computation unit is configured to heuristically obtain a preset number of current collaborative end models through evolutionary computation on the model to be processed; the evolutionary computation includes at least one of the following operations: selection, crossover, and mutation genetic operations.

[0085] Based on the above embodiments, optionally, the model to be processed determination unit includes:

[0086] If the evolutionary computation is performed for the first time after receiving the collaborative modeling request sent by the initiating end, model training is performed using the collaborative end modeling samples, and an initial model required by the collaborative end is randomly established;

[0087] The initial model is determined as the model to be processed that is currently retained and used at the collaborative end.

[0088] Based on the above embodiments, optionally, the model to be processed determination unit includes:

[0089] If the evolutionary computation is not performed for the first time after receiving the collaborative modeling request sent by the initiating end, the evaluation result of the previous collaborative end model determined in the previous evolutionary computation is obtained;

[0090] Based on the evaluation result of the previous collaborative end model, the previous collaborative end model is screened and eliminated to obtain the model to be processed.

[0091] Based on the above embodiments, optionally, the model result sending module 420 includes:

[0092] The model calculation result determination unit is configured to input the collaborative end modeling samples into the preset number of current collaborative end models respectively for calculation, and obtain the current model calculation results output by different current collaborative end models under the collaborative end modeling samples;

[0093] The model result sending unit is configured to send the current model calculation results output by different current collaborative end models under the collaborative end modeling samples to the initiating end.

[0094] Based on the above embodiments, optionally, the model result sending unit includes:

[0095] Batch compress the current model calculation results output by different current collaboration - end models respectively, and send the batch - compressed current model calculation results to the initiating end.

[0096] Based on the above - mentioned embodiments, optionally, the model result sending unit specifically includes:

[0097] Combine the initiating - end Y - label, the initiating - end modeling samples, and the current model calculation results output by different current collaboration - end models under the collaboration - end modeling samples to obtain combined modeling samples under different current collaboration - end models;

[0098] According to the combined modeling samples under different current collaboration - end models, perform model evaluation on different current collaboration - end models respectively to obtain the evaluation results of each current collaboration - end model.

[0099] Based on the above - mentioned embodiments, optionally, according to the combined modeling samples under different current collaboration - end models, perform model evaluation on different current collaboration - end models respectively to obtain the evaluation results of each current collaboration - end model, including:

[0100] Adopt a preset model evaluation algorithm, and perform model training respectively based on the combined modeling samples under different current collaboration - end models to obtain the initiating - end models under different current collaboration - end models;

[0101] Determine the evaluation results of different current collaboration - end models under the target evaluation metrics by evaluating the established different initiating - end models.

[0102] Based on the above - mentioned embodiments, optionally, the preset model evaluation algorithm includes at least one of the following: logistic regression algorithm and xgboost algorithm.

[0103] Based on the above - mentioned embodiments, optionally, the target evaluation metrics include at least one of the following: model discrimination and model precision metric.

[0104] Based on the above - mentioned embodiments, optionally, the device further includes:

[0105] A sample - pair alignment module, configured to encrypt the data index of the collaboration - end modeling samples before obtaining the current collaboration - end model through evolutionary computation, and perform a collision count on the encrypted result and the encrypted result of the data index of the initiating - end modeling samples encrypted, so as to align the collaboration - end modeling samples and the initiating - end modeling samples.

[0106] Based on the above - mentioned embodiments, optionally, the device further includes:

[0107] The model release module 440 is configured to, after obtaining a target collaborative end model that meets the modeling requirements at the initiating end, send a business explanation of the target collaborative end model to the initiating end according to the selected target collaborative end model at the initiating end, so that the initiating end can select a collaborative end model for release based on the business explanation of the target collaborative end model.

[0108] The collaborative modeling implementation device provided in the embodiments of the present invention can execute the collaborative modeling implementation method provided in any of the above embodiments of the present invention, and has the corresponding functions and beneficial effects of executing the collaborative modeling implementation method. For the detailed process, refer to the relevant operations of the collaborative modeling implementation method in the foregoing embodiments.

[0109] Figure 5 It is a schematic structural diagram of an electronic device provided in the embodiments of the present invention. As Figure 5 shown in the structure, the electronic device provided in the embodiments of the present invention includes: one or more processors 510 and a storage device 520; the processors 510 in the electronic device can be one or more, Figure 5 and one processor 510 is taken as an example here; the storage device 520 is used to store one or more programs; the one or more programs are executed by the one or more processors 510, so that the one or more processors 510 implement the collaborative modeling implementation method described in any one of the embodiments of the present invention.

[0110] The electronic device may further include: an input device 530 and an output device 540.

[0111] The processors 510, the storage device 520, the input device 530, and the output device 540 in the electronic device can be connected through a bus or other means, Figure 5 and taking the connection through a bus as an example here.

[0112] The storage device 520 in the electronic device, as a computer-readable storage medium, can be used to store one or more programs, and the programs can be software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the collaborative modeling implementation method provided in the embodiments of the present invention. The processor 510 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the storage device 520, that is, implements the collaborative modeling implementation method in the above method embodiments.

[0113] The storage device 520 may include a program storage area and a data storage area. The program storage area may store an operating system and application programs required for at least one function. The data storage area may store data created according to the use of the electronic device, etc. In addition, the storage device 520 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some instances, the storage device 520 may further include a memory remotely provided with respect to the processor 510, and these remote memories may be connected to the device through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0114] The input device 530 can be used to receive input digital or character information, and generate key signal inputs related to the user settings and function controls of the electronic device. The output device 540 may include display devices such as a display screen.

[0115] Moreover, when one or more programs included in the above electronic device are executed by the one or more processors 510, the programs perform the following operations:

[0116] Obtain the current collaborative client model through evolutionary computing;

[0117] Based on the collaborative client modeling samples, send the current model calculation results output by the current collaborative client model to the initiating end of the collaborative modeling, so that the initiating end can evaluate the current collaborative client model based on the current model calculation results;

[0118] According to the evaluation results of the current collaborative client model obtained from the initiating end, control whether to perform evolutionary computing on the current collaborative client model to generate a new collaborative client model for collaborative modeling until the initiating end obtains a collaborative client model that meets the modeling requirements.

[0119] Of course, those skilled in the art can understand that when one or more programs included in the above electronic device are executed by the one or more processors 510, the programs can also perform the related operations in the implementation method of collaborative modeling provided in any embodiment of the present invention.

[0120] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it is used to execute the implementation method of collaborative modeling. The method includes:

[0121] Obtain the current collaborative client model through evolutionary computing;

[0122] Based on the collaborative end modeling sample, send the current model calculation result output by the current collaborative end model to the initiating end of collaborative modeling, so that the initiating end can evaluate the current collaborative end model based on the current model calculation result;

[0123] According to the evaluation result of the current collaborative end model obtained from the initiating end, control whether to perform evolutionary calculation on the current collaborative end model to generate a new collaborative end model for collaborative modeling until the initiating end obtains a collaborative end model that meets the modeling requirements.

[0124] Optionally, when the program is executed by a processor, it can also be used to execute the implementation method of collaborative modeling provided in any embodiment of the present invention.

[0125] The computer storage medium of the embodiments of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (Random Access Memory, RAM), a read-only memory (Read Only Memory, ROM), an erasable programmable read-only memory (Erasable Programmable Read Only Memory, EPROM), a flash memory, an optical fiber, a portable CD-ROM, an optical storage device, a magnetic storage device, or any suitable combination of the above. The computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device.

[0126] The computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries the computer-readable program code. Such a propagated data signal can take various forms, including but not limited to: electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0127] The program code contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to: wireless, wire, optical cable, radio frequency (Radio Frequency, RF), etc., or any suitable combination of the above.

[0128] Computer program code for performing the operations of the present invention may be written in one or more programming languages or combinations thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and also including conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0129] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0130] Note that the above is only the preferred embodiment of the present invention and the technical principles applied. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and various obvious changes, re-adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the concept of the present invention, more other equivalent embodiments may be included, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for realizing collaborative modeling, characterized in that, executed by the collaborative end, the method includes: Obtaining the current collaborative end model through evolutionary computation; Based on the collaborative end modeling samples, sending the current model calculation results output by the current collaborative end model to the initiating end of collaborative modeling, so that the initiating end evaluates the current collaborative end model based on the current model calculation results; According to the evaluation results of the current collaborative end model obtained from the initiating end, controlling whether to perform evolutionary computation on the current collaborative end model to generate a new collaborative end model for collaborative modeling until the initiating end obtains a collaborative end model that meets the modeling requirements; The obtaining the current collaborative end model through evolutionary computation includes: Determining the to-be-processed model currently retained and used at the collaborative end; Through evolutionary computation on the to-be-processed model, heuristically obtaining a preset number of current collaborative end models; the evolutionary computation includes at least one of the following operations: selection, crossover, and mutation genetic operations; The determining the to-be-processed model currently retained and used at the collaborative end includes: If evolutionary computation is first performed after receiving the collaborative modeling request sent by the initiating end, then model training is performed using the collaborative end modeling samples, and an initial model required by the collaborative end is randomly established; Determining the initial model as the to-be-processed model currently retained and used at the collaborative end; If evolutionary computation is not performed for the first time after receiving the collaborative modeling request sent by the initiating end, then obtaining the evaluation results of the previous collaborative end model determined by the previous evolutionary computation; According to the evaluation results of the previous collaborative end model, performing model screening and elimination on the previous collaborative end model to obtain the to-be-processed model; Before obtaining the current collaborative end model through evolutionary computation, it further includes: Encrypting the data index of the collaborative end modeling samples, and performing a collision count on the encrypted result and the encrypted result of the data index of the initiating end modeling samples after encryption to align the collaborative end modeling samples with the initiating end modeling samples; The method further includes: after aligning the modeling samples between the collaborative end and the initiating end, the collaborative end performs data analysis of the overall missing rate and sample distribution homomorphism index on the aligned modeling samples of the collaborative end.

2. The method according to claim 1, characterized in that, Based on the collaborative end modeling samples, sending the current model calculation results output by the current collaborative end model to the initiating end of collaborative modeling includes: Inputting the collaborative end modeling samples into the preset number of current collaborative end models respectively for calculation, to obtain the current model calculation results output by different current collaborative end models under the collaborative end modeling samples; Sending the current model calculation results output by different current collaborative end models under the collaborative end modeling samples to the initiating end.

3. The method according to claim 2, characterized in that, Sending the current model calculation results output by different current collaborative end models under the collaborative end modeling samples to the initiating end includes: Batch compressing the current model calculation results output by different current collaborative end models respectively, and sending the batch compressed current model calculation results to the initiating end.

4. The method according to claim 1, characterized in that, Evaluating the current collaborative end model based on the calculation results of the current model, including: Combining the initiating end Y label, the initiating end modeling samples, and the current model calculation results output by different current collaborative end models under the collaborative end modeling samples to obtain combined modeling samples under different current collaborative end models; According to the combined modeling samples under different current collaborative end models, respectively evaluating different current collaborative end models to obtain evaluation results for each current collaborative end model.

5. The method according to claim 4, wherein, According to the combined modeling samples under different current collaborative end models, respectively evaluating different current collaborative end models to obtain evaluation results for each current collaborative end model, including: Using a preset model evaluation algorithm, respectively performing model training based on the combined modeling samples under different current collaborative end models to obtain initiating end models under different current collaborative end models; By evaluating the established different initiating end models, determining the evaluation results for different current collaborative end models under the target evaluation index.

6. The method according to claim 5, wherein, The preset model evaluation algorithm at least includes one of the following: logistic regression algorithm and xgboost algorithm.

7. The method according to claim 5, wherein, The target evaluation index includes at least one of the following: model discrimination and model precision metric.

8. The method according to claim 1, wherein, The method further includes: After obtaining a target collaborative end model that meets the modeling requirements at the initiating end, according to the target collaborative end model selected by the initiating end, sending the business explanation of the target collaborative end model to the initiating end for the initiating end to select a collaborative end model for publishing based on the business explanation of the target collaborative end model.

9. An implementation device for collaborative modeling, wherein, Configured at the collaborative end, the device includes: A model evolution calculation module for obtaining the current collaborative end model through evolution calculation; A model result sending module for sending the current model calculation results output by the current collaborative end model to the initiating end of the collaborative modeling based on the collaborative end modeling samples, so that the initiating end evaluates the current collaborative end model based on the current model calculation results; A model evaluation processing module for controlling whether to perform evolution calculation on the current collaborative end model according to the evaluation results of the current collaborative end model obtained from the initiating end, generating a new collaborative end model for collaborative modeling until the initiating end obtains a collaborative end model that meets the modeling requirements; The model evolution calculation module includes: A to-be-processed model determination unit for determining the to-be-processed model currently retained and used at the collaborative end; A model evolution calculation unit for heuristically obtaining a preset number of current collaborative end models through evolution calculation of the to-be-processed model; the evolution calculation includes at least one of the following operations: selection, crossover, and mutation genetic operations; The to-be-processed model determination unit includes: If the evolution calculation is first performed after receiving the collaborative modeling request sent by the initiating end, then use the collaborative end modeling samples for model training and randomly establish the initial model required by the collaborative end; Determine the initial model as the to-be-processed model currently retained for use at the collaborative end; If it is not the first time to perform evolutionary computation after receiving a collaborative modeling request sent by the initiating end, obtain the evaluation result of the previous collaborative end model determined by the previous evolutionary computation; Based on the evaluation result of the previous collaborative end model, perform model screening and elimination on the previous collaborative end model to obtain the to-be-processed model; The device further includes: A sample pair module, configured to encrypt the data index of the collaborative end modeling sample before obtaining the current collaborative end model through evolutionary computation, and perform a collision count on the encryption result with the encryption result of the data index of the initiating end modeling sample after encryption, so as to align the collaborative end modeling sample with the initiating end modeling sample; A data analysis unit, configured to perform data analysis on the overall missing rate and sample distribution homomorphism index of the collaborative end-aligned modeling samples at the collaborative end after aligning the modeling samples between the collaborative end and the initiating end.

10. An electronic device characterized in that it includes: One or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the collaborative modeling implementation method according to any one of claims 1-8.

11. A computer-readable storage medium, on which a computer program is stored characterized in that when the program is executed by a processor, it implements the collaborative modeling implementation method according to any one of claims 1-8.

Images