Deployment method and system of e-commerce risk identification model

Abstract

The invention discloses an e-commerce risk identification model deployment method and system, belongs to the technical field of e-commerce data processing, can realize a customized risk identification task by obtaining an e-commerce risk identification requirement input by a worker, and can identify the risk of e-commerce according to the e-commerce risk identification requirement. And the improved deep learning algorithm is adopted to perform parameter optimization and deployment on the target e-commerce risk identification model, so that the deployment effect and the deployment efficiency of the e-commerce risk identification model can be effectively improved, and the target e-commerce risk identification model can accurately identify the user-defined e-commerce risk.

Description

Technical Field

[0001] The present invention belongs to the technical field of e-commerce data processing, and particularly relates to a method and system for deploying an e-commerce risk identification model. Background Art

[0002] E-commerce risk identification is a process of using technologies such as big data and machine learning to deeply analyze transaction behaviors, product information, and user data on e-commerce platforms to identify potential risks. Its main goal is to prevent risks such as false transactions, fraud, and illegal products, and to ensure the healthy development of the platform and the rights and interests of users. By constructing a risk identification model, monitoring transaction activities in real time, detecting abnormal behaviors in a timely manner and giving early warnings, the transaction security and trust of e-commerce platforms can be improved. E-commerce risk identification is an important means to maintain the e-commerce ecosystem and is of great significance for promoting the sustainable development of the e-commerce industry. With the rapid development of the Internet, e-commerce has become an important part of modern business. However, various risks existing in e-commerce transactions, such as false transactions, fraud, and illegal products, have seriously affected the healthy development of e-commerce platforms. In order to effectively identify and prevent these risks, e-commerce platforms usually need to deploy risk identification models. However, the existing methods for deploying risk identification models often have problems such as complex deployment, low efficiency, and poor deployment effects, and are difficult to adapt to user-defined risk identification tasks. Summary of the Invention

[0003] The present invention provides a method and system for deploying an e-commerce risk identification model to solve the problems of low efficiency and poor deployment effects that often exist in the existing methods for deploying risk identification models.

[0004] On the one hand, the present invention provides a method for deploying an e-commerce risk identification model, including:

[0005] Obtaining the e-commerce risk identification requirements input by the staff; wherein, the e-commerce risk identification requirements at least include e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data;

[0006] Obtaining the selection instruction corresponding to the e-commerce risk identification model input by the staff, and based on the selection instruction, selecting a target e-commerce risk identification model from the deep learning model library;

[0007] Based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, using an improved deep learning algorithm to optimize the parameters of the target e-commerce risk identification model, and obtaining the target parameters corresponding to the target e-commerce risk identification model;

[0008] Based on the target parameters corresponding to the target e-commerce risk identification model, deploying the target e-commerce risk identification model to complete the deployment process.

[0009] Further, the deep learning model library includes one or more of an LSTM model, a CNN model, an RNN model, a CNN-LSTM model, a CNN-BP model, a BP model, a GRU model, and an SVM model.

[0010] Further, based on the e-commerce risk identification sample data and the corresponding e-commerce risk labels of the e-commerce risk identification sample data, an improved deep learning algorithm is used to optimize the parameters of the target e-commerce risk identification model, and the target parameters corresponding to the target e-commerce risk identification model are obtained, including:

[0011] The parameters of the target e-commerce risk identification model are initialized by using a chaotic mapping initialization method to obtain multiple different parameter individuals;

[0012] Based on the e-commerce risk identification sample data and the corresponding e-commerce risk labels of the e-commerce risk identification sample data, the fitness corresponding to each parameter individual is obtained;

[0013] According to the fitness corresponding to all parameter individuals, the optimal parameter individual is determined;

[0014] Based on the optimal parameter individual, a first variable curve search strategy is used to perform multi-information fusion search on the parameter individuals to obtain the parameter individuals after multi-information fusion search;

[0015] For the parameter individuals after multi-information fusion search, a second variable curve search strategy is used to perform neighborhood information search on the parameter individuals to obtain the parameter individuals after neighborhood information search;

[0016] For the parameter individuals after neighborhood information search, a good point attraction search strategy is used to perform guiding search on the parameter individuals to obtain the parameter individuals after guiding search;

[0017] For the parameter individuals after guiding search, a Cauchy global mutation search strategy is used to perform global search on the parameter individuals to obtain the parameter individuals after global search;

[0018] It is judged whether the optimization end condition is satisfied. If so, the optimal parameter individual is re-determined according to the parameter individuals after global search, and the re-determined optimal parameter individual is used as the target parameter corresponding to the target e-commerce risk identification model. Otherwise, the step of obtaining the fitness is returned.

[0019] Further, the parameters of the target e-commerce risk identification model are initialized by using a chaotic mapping initialization method to obtain multiple different parameter individuals, including:

[0020] Generate a random position selection individual using a random initialization method; among them, the total dimension of the parameters of the random position selection individual is the same as the total dimension of the parameters of the target e-commerce risk identification model;

[0021] Based on the random position selection individual, obtain other position selection individuals as follows:

[0022]

[0023] where y i represents the i-th position selection individual, and when i = 1, y i is set as the random position selection individual; y i+1 represents the (i + 1)-th position selection individual, mod represents the remainder function, π represents the pi, and sin represents the sine function;

[0024] Map the position selection individual to the solution space and determine the parameter individual as:

[0025] x i,d = y i+1,d *(ub d - lb d ) + lb d

[0026] where y i+1,d represents the d-th parameter of the (i + 1)-th position selection individual, x i,d represents the d-th parameter of the i-th parameter individual, d = 1, 2, …, D, D represents the total dimension of the parameters of the target e-commerce risk identification model, lb d represents the lower limit of the d-th parameter, and ub d represents the upper limit of the d-th parameter.

[0027] Furthermore, based on the e-commerce risk identification sample data and the corresponding e-commerce risk labels of the e-commerce risk identification sample data, obtain the fitness corresponding to each parameter individual, including:

[0028] For any parameter individual, after applying the hyperparameters included in the parameter individual to the e-commerce risk identification model, use the e-commerce risk identification sample data as the input to obtain the actual output of the e-commerce risk identification model;

[0029] Use the e-commerce risk label corresponding to the e-commerce risk identification sample data as the expected output of the e-commerce risk identification model, and based on the actual output and the expected output of the e-commerce risk identification model, obtain the cross-entropy loss function value;

[0030] Add the cross-entropy loss function value to a preset constant to obtain a non-zero fitness calculation parameter, and take the reciprocal of the non-zero fitness calculation parameter to obtain the fitness corresponding to the parameter individual;

[0031] Traverse all parameter individuals to obtain the fitness corresponding to each parameter individual.

[0032] Further, based on the optimal parameter individual, use the first variable curve search strategy to perform multi-information fusion search on the parameter individuals, and obtain the parameter individuals after multi-information fusion search, including:

[0033]

[0034] λ = e hcos(π(1-t / T))

[0035] where represents the k-th parameter individual in the t-th optimization process, k = 1, 2, …, N, and N represents the total number of parameter individuals, represents the parameter individual after multi-information fusion search represents the optimal parameter individual, e represents the natural constant, cos represents the cosine function, λ represents the variable curve control parameter, r1 represents the first random number uniformly distributed between [-1, 1], h represents the constant term and is set to 5, T represents the preset maximum number of optimization times, π represents pi, r2 represents the second random number uniformly distributed between [-2, 2], R represents a random positive integer less than 2N / 3, represents the j-th parameter individual in the t-th optimization process.

[0036] Further, for the parameter individuals after multi-information fusion search, use the second variable curve search strategy to perform neighborhood information search on the parameter individuals, and obtain the parameter individuals after neighborhood information search, including:

[0037]

[0038] where represents the m-th parameter individual after multi-information fusion search in the t-th optimization process, m = 1, 2, …, N, and N represents the total number of parameter individuals, represents the parameter individual after neighborhood information search ω represents the inertia weight, sin represents the sine function, r3 represents the third random number between (0, 1), r4 represents the fourth random number between (0, 1), represents the random parameter individual except the parameter individual c1 represents the first weighting factor between (0, 1), c2 represents the second weighting factor between (0, 1), α1 represents the first inertia weight adjustment factor, α2 represents the second inertia weight adjustment factor, ω max represents the preset maximum value of the inertia weight, ω minrepresents the preset minimum value of the inertia weight, cos represents the cosine function, T represents the preset maximum number of optimization times, and π represents pi.

[0039] Furthermore, for the parameter individuals after the neighborhood information search, a good point attraction search strategy is used to guide the search of the parameter individuals, and the parameter individuals after the guided search are obtained, including:

[0040]

[0041] in, represents the parameter individuals after the nth neighborhood information search in the tth optimization process, n = 1, 2, ..., N, N represents the total number of parameter individuals, represents the parameter individuals after guided search, represents the optimal parameter individual, r5 represents the fifth random number between (0,1), r6 represents the sixth random number uniformly distributed between [-1,1], Indicates that the individual Random parameter individuals other than .

[0042] Furthermore, for the parameter individuals after the guided search, the Cauchy global mutation search strategy is used to perform a global search on the parameter individuals to obtain the parameter individuals after the global search, including:

[0043] Based on the current number of optimization attempts, the adaptive mutation factor is obtained as follows:

[0044]

[0045] Among them, p is the adaptive variation factor, sin represents the sine function, π represents the pi, T represents the preset maximum number of optimization times, and t represents the current number of optimization times;

[0046] Based on the adaptive mutation factor, a global search is performed on the parameter individuals, and the parameter individuals obtained after the global search are:

[0047]

[0048] in, represents the parameter individuals after the qth guided search in the tth optimization process, q = 1, 2, ..., N, N represents the total number of parameter individuals, Represents the parameter individual after global search cauchy(0,1) represents random numbers generated using a standard cauchy distribution.

[0049] On the other hand, the present invention provides a deployment system for an e-commerce risk identification model, including: an identification requirement acquisition module, a model selection module, a parameter optimization module, and a model deployment module;

[0050] The recognition requirement acquisition module is used to acquire the e-commerce risk recognition requirements input by the staff; wherein, the e-commerce risk recognition requirements at least include e-commerce risk recognition sample data and the corresponding e-commerce risk labels of the e-commerce risk recognition sample data;

[0051] The model selection module is used to acquire the selection instruction corresponding to the e-commerce risk recognition model input by the staff, and based on the selection instruction, select the target e-commerce risk recognition model in the deep learning model library;

[0052] The parameter optimization module is used to optimize the parameters of the target e-commerce risk recognition model by using an improved deep learning algorithm based on the e-commerce risk recognition sample data and the corresponding e-commerce risk labels of the e-commerce risk recognition sample data, and obtain the target parameters corresponding to the target e-commerce risk recognition model;

[0053] The model deployment module is used to deploy the target e-commerce risk recognition model based on the target parameters corresponding to the target e-commerce risk recognition model to complete the deployment process.

[0054] A method and system for deploying an e-commerce risk recognition model provided by the present invention can realize a customized risk recognition task by acquiring the e-commerce risk recognition requirements input by the staff, and optimize and deploy the parameters of the target e-commerce risk recognition model by using an improved deep learning algorithm according to the e-commerce risk recognition requirements, which can effectively improve the deployment effect and deployment efficiency of the e-commerce risk recognition model, so that the target e-commerce risk recognition model can accurately identify the user-defined e-commerce risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The drawings herein are incorporated into the specification and form a part of the specification, showing embodiments consistent with the present invention and used together with the specification to explain the principles of the present invention.

[0056] Figure 1 It is a schematic flowchart of a method for deploying an e-commerce risk recognition model provided by an embodiment of the present invention.

[0057] Figure 2 It is a schematic structural diagram of a system for deploying an e-commerce risk recognition model provided by an embodiment of the present invention.

[0058] Through the above drawings, specific embodiments of the present invention have been shown, and there will be more detailed descriptions hereinafter. These drawings and text descriptions are not intended to limit the scope of the inventive concept in any way, but to illustrate the concept of the present invention to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0060] Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0061] As Figure 1 shown, the present invention provides a method for deploying an e-commerce risk identification model, including:

[0062] S11. Obtain the e-commerce risk identification requirements input by the staff; wherein, the e-commerce risk identification requirements at least include e-commerce risk identification sample data and e-commerce risk labels corresponding to the e-commerce risk identification sample data;

[0063] For example, when the staff needs to identify prohibited items or sensitive items to avoid e-commerce illegal risks, the e-commerce risk identification sample data can be input as e-commerce product images, and the e-commerce risk labels corresponding to the e-commerce risk identification sample data can be input as image types, and the image types include normal categories and various abnormal categories. Thus, a custom identification task can be achieved.

[0064] It should be noted that the above e-commerce risk identification sample data and e-commerce risk labels corresponding to the e-commerce risk identification sample data are only examples of the embodiments of the present invention, and the staff can also input other types of data for the deployment of the e-commerce risk identification model to achieve different e-commerce risk identifications.

[0065] S12. Obtain the selection instruction corresponding to the e-commerce risk identification model input by the staff, and based on the selection instruction, select a target e-commerce risk identification model in the deep learning model library;

[0066] The deep learning model library can include a variety of deep learning models, and these deep learning models at least include models for data identification and models for image identification, which can enable the staff to select different models to achieve custom e-commerce risk identification tasks, thereby improving the flexibility of e-commerce risk identification.

[0067] S13. Based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, use an improved deep learning algorithm to optimize the parameters of the target e-commerce risk identification model, and obtain the target parameters corresponding to the target e-commerce risk identification model;

[0068] In the prior art, optimization algorithms such as particle swarm optimization algorithm and genetic algorithm are generally used to deploy deep learning models, which often have technical problems of poor deployment effect and poor deployment efficiency, resulting in the inability to accurately complete the e-commerce risk identification task finally. Therefore, the embodiments of the present invention adopt an improved deep learning algorithm to optimize the parameters of the target e-commerce risk identification model to solve the technical problems existing in the prior art and improve the accuracy of e-commerce risk identification.

[0069] S14. Deploy the target e-commerce risk identification model with the target parameters corresponding to the target e-commerce risk identification model to complete the deployment process.

[0070] A deployment method and system for an e-commerce risk identification model provided by the present invention can realize a customized risk identification task by obtaining the e-commerce risk identification requirements input by the staff, and optimize and deploy the parameters of the target e-commerce risk identification model by using an improved deep learning algorithm according to the e-commerce risk identification requirements, which can effectively improve the deployment effect and deployment efficiency of the e-commerce risk identification model, so that the target e-commerce risk identification model can accurately identify the e-commerce risks customized by the user.

[0071] In the embodiments of the present invention, the deep learning model library includes one or more of an LSTM (Long Short-Term Memory) model, a CNN (Convolutional Neural Networks) model, an RNN (Recurrent Neural Network) model, a CNN-LSTM model, a CNN-BP (Back Propagation) model, a BP model, a GRU (Gate Recurrent Unit) model, and an SVM (Support Vector Machine) model.

[0072] It should be noted that the above deep learning models are only preferred cases of the embodiments of the present invention, and the deep learning model library may also include other deep learning models. When it is detected that the e-commerce risk identification sample data input by the staff is image data, only the deep learning models that can identify images are allowed to be selected by the staff.

[0073] In the embodiments of the present invention, based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, an improved deep learning algorithm is used to optimize the parameters of the target e-commerce risk identification model to obtain the target parameters corresponding to the target e-commerce risk identification model, including:

[0074] Initialize the parameters of the target e-commerce risk identification model using the chaotic mapping initialization method to obtain multiple different parameter individuals;

[0075] Based on the e-commerce risk identification sample data and the corresponding e-commerce risk labels of the e-commerce risk identification sample data, obtain the fitness corresponding to each parameter individual;

[0076] Determine the optimal parameter individual according to the fitness corresponding to all parameter individuals;

[0077] Based on the optimal parameter individual, use the first variable curve search strategy to perform multi-information fusion search on the parameter individuals to obtain the parameter individuals after multi-information fusion search;

[0078] For the parameter individuals after multi-information fusion search, use the second variable curve search strategy to perform neighborhood information search on the parameter individuals to obtain the parameter individuals after neighborhood information search;

[0079] For the parameter individuals after neighborhood information search, use the good point attraction search strategy to perform guided search on the parameter individuals to obtain the parameter individuals after guided search;

[0080] For the parameter individuals after guided search, use the Cauchy global mutation search strategy to perform global search on the parameter individuals to obtain the parameter individuals after global search;

[0081] Judge whether the optimization end condition is satisfied. If so, re-determine the optimal parameter individual according to the parameter individual after global search, and use the re-determined optimal parameter individual as the target parameter corresponding to the target e-commerce risk identification model. Otherwise, return to the step of obtaining the fitness.

[0082] In the prior art, optimization algorithms such as particle swarm algorithm and genetic algorithm are generally used to deploy deep learning models, which often have technical problems such as poor deployment effect and poor deployment efficiency, resulting in the inability to accurately complete the e-commerce risk identification task finally. Therefore, the embodiments of the present invention use an improved deep learning algorithm to perform parameter optimization on the target e-commerce risk identification model to solve the technical problems existing in the prior art and improve the accuracy of e-commerce risk identification.

[0083] In the embodiments of the present invention, the parameters of the target e-commerce risk identification model are initialized using the chaotic mapping initialization method to obtain multiple different parameter individuals, including:

[0084] Generate a random position selection individual using the random initialization method; wherein, the total parameter dimension of the random position selection individual is the same as the total parameter dimension of the target e-commerce risk identification model; for example, an individual can be randomly generated in the solution space and directly used as the random position selection individual.

[0085] Based on randomly selecting individuals at positions, obtaining individuals selected at other positions as follows:

[0086]

[0087] where y i represents the individual selected at the i-th position, and when i = 1, y i is set as the randomly selected individual at the position; y i+1 represents the individual selected at the (i + 1)-th position, mod represents the remainder function, π represents the pi, and sin represents the sine function;

[0088] Mapping the individuals selected at positions to the solution space, determining the parameter individuals as:

[0089] x i,d = y i+1,d *(ub d - lb d ) + lb d

[0090] where y i+1,d represents the d-th dimension parameter of the individual selected at the (i + 1)-th position, x i,d represents the d-th dimension parameter of the i-th parameter individual, d = 1, 2,..., D, D represents the total dimension of the parameters of the target e-commerce risk identification model, lb d represents the lower limit of the d-th dimension parameter, and ub d represents the upper limit of the d-th dimension parameter.

[0091] The chaos mapping initialization method provided by the embodiments of the present invention is more uniformly distributed compared with the conventional chaos mapping, and can effectively make the initial solutions more uniformly distributed in the solution space. It can not only effectively improve the deployment efficiency of the target e-commerce risk identification model, but also effectively avoid the algorithm from getting stuck.

[0092] In the embodiments of the present invention, based on the e-commerce risk identification sample data and the corresponding e-commerce risk labels of the e-commerce risk identification sample data, obtaining the fitness corresponding to each parameter individual includes:

[0093] For any parameter individual, after applying the hyperparameters included in the parameter individual to the e-commerce risk identification model, using the e-commerce risk identification sample data as the input, obtaining the actual output of the e-commerce risk identification model;

[0094] Using the e-commerce risk label corresponding to the e-commerce risk identification sample data as the expected output of the e-commerce risk identification model, and according to the actual output and the expected output of the e-commerce risk identification model, obtaining the cross-entropy loss function value;

[0095] Add the cross - entropy loss function value to a preset constant to obtain a non - zero fitness calculation parameter, and take the reciprocal of the non - zero fitness calculation parameter to obtain the fitness corresponding to the parameter individual;

[0096] Traverse all parameter individuals to obtain the fitness corresponding to each parameter individual.

[0097] In the embodiment of the present invention, based on the optimal parameter individual, a first variable curve search strategy is used to perform multi - information fusion search on the parameter individuals to obtain the parameter individuals after multi - information fusion search, including:

[0098]

[0099] λ = e hcos(π(1-t / T))

[0100] Among them, represents the k - th parameter individual in the t - th optimization process, k = 1, 2, …, N, where N represents the total number of parameter individuals, represents the parameter individual after multi - information fusion search represents the optimal parameter individual, e represents the natural constant, cos represents the cosine function, λ represents the variable curve control parameter, r1 represents the first random number uniformly distributed between [-1, 1], h represents a constant term and is set to 5, T represents the preset maximum number of optimization times, π represents the pi, r2 represents the second random number uniformly distributed between [-2, 2], R represents a random positive integer less than 2N / 3, represents the j - th parameter individual in the t - th optimization process.

[0101] The first variable curve search strategy provided by the embodiment of the present invention can enable parameter individuals to fuse the information of other parameter individuals, enabling outlier parameter individuals to obtain higher search efficiency, effectively improving the search speed of the algorithm. At the same time, combining the optimal parameter individual for variable curve search can provide the ability to jump out of the local optimum during the search towards the optimal position, and in the later stage of the algorithm, it can ensure the convergence of the algorithm and improve the deployment effect of the target e - commerce risk identification model.

[0102] In the embodiment of the present invention, for the parameter individuals after multi - information fusion search, a second variable curve search strategy is used to perform neighborhood information search on the parameter individuals to obtain the parameter individuals after neighborhood information search, including:

[0103]

[0104] Among them, represents the m - th parameter individual after multi - information fusion search in the t - th optimization process, m = 1, 2, …, N, where N represents the total number of parameter individuals, Denote the parameter individuals after neighborhood information search ω represents the inertia weight, sin represents the sine function, r3 represents the third random number between (0, 1), and r4 represents the fourth random number between (0, 1). Denote the random parameter individuals except the parameter individuals c1 represents the first weighting factor between (0, 1), c2 represents the second weighting factor between (0, 1), α1 represents the first inertia weight adjustment factor, α2 represents the second inertia weight adjustment factor, and ω max represents the preset maximum value of the inertia weight, ω min represents the preset minimum value of the inertia weight, cos represents the cosine function, T represents the preset maximum number of optimization iterations, and π represents the pi.

[0105] The second variable curve search strategy provided by the embodiments of the present invention can perform collaborative search by combining different parameter individuals, and perform adaptive inertia weight adjustment during the collaborative search process to gradually improve the convergence ability of the algorithm as the number of optimization iterations increases, and can effectively search the area between two parameter vectors, improving the ability to search for the global optimal solution.

[0106] In the embodiments of the present invention, for the parameter individuals after neighborhood information search, a good point attraction search strategy is adopted to guide the search for the parameter individuals, and the parameter individuals after the guided search are obtained, including:

[0107]

[0108] wherein represents the nth parameter individual after neighborhood information search in the tth optimization process, n = 1, 2, …, N, and N represents the total number of parameter individuals represents the parameter individual after the guided search represents the optimal parameter individual, r5 represents the fifth random number between (0, 1), and r6 represents the sixth random number uniformly distributed between [-1, 1].[[]] represents the random parameter individuals except the parameter individuals

[0109] The good point attraction search strategy provided by the embodiments of the present invention can enable the parameter individuals to perform information interaction with other parameter individuals and search around the optimal position at the same time, thereby improving the optimization accuracy of the algorithm. In the later stage of the algorithm, all parameter individuals search around together, which can greatly increase the probability of finding the global optimal solution.

[0110] In the embodiments of the present invention, for the parameter individuals after the guided search, a Cauchy global mutation search strategy is adopted to perform global search on the parameter individuals, and the parameter individuals after the global search are obtained, including:[[]] ​

[0111] Based on the current optimization iteration count, the adaptive mutation factor is obtained as follows:

[0112]

[0113] where p is the adaptive mutation factor, sin represents the sine function, π represents the pi, T represents the preset maximum optimization iteration count, and t represents the current optimization iteration count;

[0114] Based on the adaptive mutation factor, global search is performed on the parameter individuals, and the parameter individuals after global search are obtained as follows:

[0115]

[0116] where represents the parameter individual after the q-th guided search in the t-th optimization process, q = 1, 2,..., N, and N represents the total number of parameter individuals, represents the parameter individual after global search cauchy(0, 1) represents a random number generated by the standard Cauchy distribution.

[0117] The Cauchy global mutation search strategy provided by the embodiments of the present invention can effectively improve the global search ability of the algorithm, thereby avoiding the algorithm from falling into local optimum. In the later stage of the algorithm, the global search ability gradually decreases, which can ensure the convergence of the algorithm.

[0118] In summary, the embodiments of the present invention improve the deployment efficiency and deployment effect of the e-commerce risk identification model, can effectively increase the e-commerce risk identification ability, assist the staff in discovering e-commerce risks, and improve work efficiency.

[0119] As Figure 2 shown, the present invention provides a deployment system for an e-commerce risk identification model, including: an identification requirement acquisition module 21, a model selection module 22, a parameter optimization module 23, and a model deployment module 24;

[0120] The identification requirement acquisition module 21 is configured to acquire the e-commerce risk identification requirements input by the staff; wherein, the e-commerce risk identification requirements at least include e-commerce risk identification sample data and e-commerce risk labels corresponding to the e-commerce risk identification sample data;

[0121] The model selection module 22 is configured to acquire the selection instruction corresponding to the e-commerce risk identification model input by the staff, and based on the selection instruction, select a target e-commerce risk identification model from the deep learning model library;

[0122] The parameter optimization module 23 is configured to optimize the parameters of the target e-commerce risk identification model by using an improved deep learning algorithm based on the e-commerce risk identification sample data and the corresponding e-commerce risk labels of the e-commerce risk identification sample data, so as to obtain the target parameters corresponding to the target e-commerce risk identification model;

[0123] The model deployment module 24 is configured to deploy the target e-commerce risk identification model with the target parameters corresponding to the target e-commerce risk identification model to complete the deployment process.

[0124] The deployment system of an e-commerce risk identification model provided by an embodiment of the present invention can execute the above method technical solution, and its principle and beneficial effects are similar, which will not be elaborated here.

[0125] Those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include known common knowledge or conventional technical means in the technical field not disclosed by the present invention. It should be understood that the present invention is not limited to the precise structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A method for deploying an e-commerce risk identification model, characterized in that: include: Obtaining e-commerce risk identification requirements input by a staff member; wherein the e-commerce risk identification requirements at least include e-commerce risk identification sample data and e-commerce risk labels corresponding to the e-commerce risk identification sample data; Obtaining a selection instruction corresponding to the e-commerce risk identification model input by the staff, and selecting a target e-commerce risk identification model in the deep learning model library based on the selection instruction; Based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, an improved deep learning algorithm is used to optimize the parameters of the target e-commerce risk identification model to obtain the target parameters corresponding to the target e-commerce risk identification model; The target e-commerce risk identification model is deployed with the target parameters corresponding to the target e-commerce risk identification model to complete the deployment process.

2. The method for deploying an e-commerce risk identification model according to claim 1, characterized in that: The deep learning model library includes one or more of an LSTM model, a CNN model, an RNN model, a CNN-LSTM model, a CNN-BP model, a BP model, a GRU model, and an SVM model.

3. The method for deploying an e-commerce risk identification model according to claim 1, characterized in that: Based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, an improved deep learning algorithm is used to optimize the parameters of the target e-commerce risk identification model to obtain the target parameters corresponding to the target e-commerce risk identification model, including: The parameters of the target e-commerce risk identification model are initialized using the chaotic mapping initialization method to obtain multiple different parameter individuals; Based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, the fitness corresponding to each parameter individual is obtained; According to the fitness corresponding to all parameter individuals, determine the optimal parameter individual; Based on the optimal parameter individual, a first variable curve search strategy is used to perform multiple information fusion search on the parameter individual to obtain the parameter individual after multiple information fusion search; For the parameter individuals after multiple information fusion searches, the second variable curve search strategy is used to search the neighborhood information of the parameter individuals to obtain the parameter individuals after the neighborhood information search; For the parameter individuals after the neighborhood information search, the best point attraction search strategy is used to guide the search of the parameter individuals to obtain the parameter individuals after the guided search; For the parameter individuals after the guided search, the Cauchy global mutation search strategy is used to perform a global search on the parameter individuals to obtain the parameter individuals after the global search; Determine whether the optimization end condition is met. If so, redetermine the optimal parameter individual based on the parameter individual after the global search, and use the redetermined optimal parameter individual as the target parameter corresponding to the target e-commerce risk identification model. Otherwise, return to the step of obtaining fitness.

4. The method for deploying an e-commerce risk identification model according to claim 3, characterized in that: The chaotic mapping initialization method is used to initialize the parameters of the target e-commerce risk identification model to obtain multiple different parameter individuals, including: A random initialization method is used to generate a random position selection individual; wherein the total dimension of the parameters of the random position selection individual is the same as the total dimension of the parameters of the target e-commerce risk identification model; Based on the random position selection individual, the other position selection individuals are obtained as follows: Among them, y i represents the i-th position selection individual, and when i=1, y i Set to select individuals at random positions; y i+1 represents the individual selected at the i+1th position, mod represents the remainder function, π represents the pi, and sin represents the sine function; Map the location selection individuals to the solution space and determine the parameter individuals as: x i,d =y i+1,d *(ub d -lb d )+lb d Among them, y i+1,d represents the d-th dimension parameter of the individual selected at the i+1th position, x i,d represents the d-th dimension parameter of the i-th parameter individual, d=1,2,…,D, D represents the total dimension of the parameters of the target e-commerce risk identification model, lb d Represents the lower limit of the d-th dimension parameter, ub d Indicates the upper limit of the d-th dimension parameter.

5. The method for deploying an e-commerce risk identification model according to claim 3, characterized in that: Based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, the fitness corresponding to each parameter individual is obtained, including: For any parameter individual, after applying the hyperparameters contained in the parameter individual to the e-commerce risk identification model, the e-commerce risk identification sample data is used as input to obtain the actual output of the e-commerce risk identification model; The e-commerce risk label corresponding to the e-commerce risk identification sample data is used as the expected output of the e-commerce risk identification model, and the cross entropy loss function value is obtained according to the actual output and expected output of the e-commerce risk identification model; Add the cross entropy loss function value to the preset constant to obtain a non-zero fitness calculation parameter, and then take the inverse of the non-zero fitness calculation parameter to obtain the fitness corresponding to the parameter individual; Traverse all parameter individuals and obtain the fitness corresponding to each parameter individual.

6. The method for deploying an e-commerce risk identification model according to claim 5, characterized in that: Based on the optimal parameter individual, the first variable curve search strategy is used to perform multiple information fusion search on the parameter individual to obtain the parameter individual after multiple information fusion search, including: λ=e hcos(π(1-t / T)) in, represents the kth parameter individual in the tth optimization process, k = 1, 2, ..., N, N represents the total number of parameter individuals, Represents the parameter individual after multiple information fusion search represents the optimal parameter individual, e represents the natural constant, cos represents the cosine function, λ represents the variable curve control parameter, r1 represents the first random number uniformly distributed between [-1,1], h represents the constant term, and is set to 5, T represents the preset maximum number of optimization times, π represents the circumference of a circle, r2 represents the second random number uniformly distributed between [-2,2], and R represents a random positive integer less than 2N / 3. Represents the jth parameter individual in the tth optimization process.

7. The method for deploying an e-commerce risk identification model according to claim 6, characterized in that: For the parameter individuals after multiple information fusion searches, the second variable curve search strategy is used to search the neighborhood information of the parameter individuals to obtain the parameter individuals after the neighborhood information search, including: in, represents the parameter individual after the mth multi-information fusion search in the tth optimization process, m = 1, 2, ..., N, N represents the total number of parameter individuals, Represents the parameter individual after neighborhood information search ω represents the inertia weight, sin represents the sine function, r3 represents the third random number between (0,1), and r4 represents the fourth random number between (0,1). Indicates that the individual The random parameter individuals outside, c1 represents the first weighting factor between (0,1), c2 represents the second weighting factor between (0,1), α1 represents the first inertia weight adjustment factor, α2 represents the second inertia weight adjustment factor, ω max Represents the preset maximum value of the inertia weight, ω min represents the preset minimum value of the inertia weight, cos represents the cosine function, T represents the preset maximum number of optimization times, and π represents pi.

8. The method for deploying an e-commerce risk identification model according to claim 7, characterized in that: For the parameter individuals after the neighborhood information search, the best point attraction search strategy is used to guide the search of the parameter individuals, and the parameter individuals after the guided search are obtained, including: in, represents the parameter individuals after the nth neighborhood information search in the tth optimization process, n = 1, 2, ..., N, N represents the total number of parameter individuals, represents the parameter individuals after guided search, represents the optimal parameter individual, r5 represents the fifth random number between (0,1), r6 represents the sixth random number uniformly distributed between [-1,1], Indicates that the individual Random parameter individuals other than .

9. The method for deploying an e-commerce risk identification model according to claim 8, characterized in that: For the parameter individuals after the guided search, the Cauchy global mutation search strategy is used to perform a global search on the parameter individuals to obtain the parameter individuals after the global search, including: Based on the current number of optimization attempts, the adaptive mutation factor is obtained as follows: Among them, p is the adaptive variation factor, sin represents the sine function, π represents the pi, T represents the preset maximum number of optimization times, and t represents the current number of optimization times; Based on the adaptive mutation factor, a global search is performed on the parameter individuals, and the parameter individuals obtained after the global search are: in, represents the parameter individuals after the qth guided search in the tth optimization process, q = 1, 2, ..., N, N represents the total number of parameter individuals, Represents the parameter individual after global search cauchy(0,1) represents random numbers generated using a standard cauchy distribution.

10. A deployment system for an e-commerce risk identification model, characterized in that: include: Identify requirement acquisition module, model selection module, parameter optimization module, and model deployment module; The identification requirement acquisition module is used to acquire the e-commerce risk identification requirements input by the staff; wherein the e-commerce risk identification requirements at least include e-commerce risk identification sample data and e-commerce risk labels corresponding to the e-commerce risk identification sample data; The model selection module is used to obtain the selection instruction corresponding to the e-commerce risk identification model input by the staff, and select the target e-commerce risk identification model in the deep learning model library based on the selection instruction; The parameter optimization module is used to optimize the parameters of the target e-commerce risk identification model using an improved deep learning algorithm based on the e-commerce risk identification sample data and the e-commerce risk labels corresponding to the e-commerce risk identification sample data, so as to obtain the target parameters corresponding to the target e-commerce risk identification model; The model deployment module is used to deploy the target e-commerce risk identification model with the target parameters corresponding to the target e-commerce risk identification model to complete the deployment process.

Images