Method, device, equipment and storage medium for updating business process model

Abstract

The application discloses a method, device and equipment for updating a business processing model and a storage medium, and relates to the technical field of artificial intelligence. A specific embodiment of the method comprises the following steps: obtaining newly added business data of an initial business processing model, and obtaining reference business data generated by a preset first generative model; taking the newly added business data and the reference business data as training samples, training the initial business processing model, and obtaining an updated business processing model; taking the newly added business data and the reference business data as training samples, training a preset second generative model, and obtaining an updated second generative model; taking the updated second generative model as an updated first generative model; taking the updated business processing model as the initial business processing model, and taking the updated first generative model as the preset first generative model.

Description

Technical Field

[0001] This application relates to the field of computer technology, specifically to the field of artificial intelligence technology, and in particular to methods, apparatus, devices, and storage media for updating business processing models. Background Technology

[0002] With the development of artificial intelligence technology, deep learning technology has been widely applied in many fields due to its ability to deeply extract data from business data. In current practical applications, to ensure the online performance of deep learning models, it is often necessary to train the model on all new and old business data together when the incremental business data reaches a certain volume, and then update and deploy the model. However, as the amount of business data continues to increase, the cost of data storage and the resources required for model training also increase dramatically; at the same time, the efficiency of model training decreases significantly, making the contradiction between model update time and online performance increasingly serious, affecting application experience and product competitiveness. Therefore, incremental learning technology has received widespread attention.

[0003] Currently, incremental learning includes generative incremental learning; however, with the arrival of new data, the memory-generative model itself also needs to learn from the new data. This requires the memory-generative model to generate old data while simultaneously training and updating itself with new data. Since the learning and generation processes are coupled together, this leads to model drift, i.e., forgetting the memory of old data, which in turn causes the overall incremental learning effect to continuously decrease. Summary of the Invention

[0004] Embodiments of this application propose methods, apparatus, devices, and storage media for updating business processing models.

[0005] In a first aspect, embodiments of this application provide a method for updating a business processing model. The method includes: acquiring new business data of an initial business processing model and acquiring reference business data generated by a preset first generative model; using the new business data and the reference business data together as training samples to train the initial business processing model to obtain an updated business processing model; using the new business data and the reference business data together as training samples to train a preset second generative model to obtain an updated second generative model; using the updated second generative model as an updated first generative model; using the updated business processing model as the initial business processing model, and using the updated first generative model as a preset first generative model.

[0006] In some embodiments, the preset first generative model and the preset second generative model are models with the same structure.

[0007] In some embodiments, the preset update conditions are: the accuracy of the initial business processing model meets a preset accuracy threshold; and / or, the number of new business data in the initial business processing model meets a preset quantity threshold.

[0008] In some embodiments, the initial business processing model is determined based on the following steps: obtaining second initial business data; and training a machine learning model using the second initial business data to obtain the initial business processing model.

[0009] In some embodiments, the initial business processing model is determined based on the following steps: obtaining second initial business data; and training a machine learning model using the second initial business data to obtain the initial business processing model.

[0010] In some embodiments, obtaining new business data from an initial business processing model and obtaining reference business data generated by a preset first generative model includes: determining whether the initial business model meets preset update conditions; if so, obtaining new business data from the initial business processing model and obtaining reference business data generated by a preset first generative model.

[0011] In some embodiments, the method for updating the business processing model further includes: repeatedly executing the above steps until the initial business processing model no longer meets the preset update conditions.

[0012] Secondly, embodiments of this application provide an apparatus for updating a business processing model. The apparatus includes: a data acquisition module configured to acquire new business data of an initial business processing model and acquire reference business data generated by a preset first generative model; a first update module configured to use the new business data and the reference business data together as training samples to train the initial business processing model to obtain an updated business processing model; and a second update module configured to use the new business data and the reference business data together as training samples to train a preset second generative model to obtain an updated second generative model; and to use the updated second generative model as an updated first generative model; and a model as a module configured to use the updated business processing model as the initial business processing model and the updated first generative model as a preset first generative model.

[0013] In some embodiments, the preset first generative model and the preset second generative model are models with the same structure.

[0014] In some embodiments, the preset update conditions include: the accuracy of the initial business processing model meets a preset accuracy threshold; and / or, the number of new business data in the initial business processing model meets a preset quantity threshold.

[0015] In some embodiments, the apparatus for updating the business processing model further includes: a data acquisition module configured to acquire second initial business data; and a model training module configured to train a machine learning model using the second initial business data to obtain an initial business processing model.

[0016] In some embodiments, the apparatus for updating the business model further includes: a data acquisition module configured to acquire second initial business data; and a model training module configured to train a machine learning model using the second initial business data to obtain an initial business processing model.

[0017] In some embodiments, the data acquisition module includes: a condition judgment unit configured to judge whether the initial business model meets the preset update conditions; and a data acquisition unit configured to, if so, acquire the new business data of the initial business processing model and acquire the reference business data generated by the preset first generative model.

[0018] In some embodiments, the apparatus for updating the business processing model further includes:

[0019] The loop module is configured to execute the above steps repeatedly until the initial business processing model no longer meets the preset update conditions.

[0020] Thirdly, embodiments of this application provide a business processing method, including: acquiring a target business; and performing preset business processing on the target business according to any of the business processing models in the first aspect.

[0021] In some embodiments, the preset business processing includes at least one of the following: identification of target business, recommendation of target business, and prediction of demand for target business.

[0022] Fourthly, embodiments of this application provide a service processing apparatus, including: a service acquisition module configured to acquire a target service; and a service processing module configured to perform preset service processing on the target service according to any of the service processing models in the first aspect.

[0023] In some embodiments, the preset business processing includes at least one of the following: identification of target business, recommendation of target business, and prediction of demand for target business.

[0024] Fifthly, embodiments of this application provide an electronic device including at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the methods described in the first or second aspect.

[0025] In a sixth aspect, embodiments of this application provide a non-transitory computer-readable storage medium storing computer instructions, characterized in that the computer instructions are used to cause a computer to perform the methods described in the first or second aspect.

[0026] The method, apparatus, device, and storage medium for updating a business processing model provided in the embodiments of this application first acquire newly added business data of the initial business processing model within a preset time period, and acquire reference business data generated by a preset first generative model; then, the newly added business data and the reference business data are used together as training samples to train the initial business processing model to obtain an updated business processing model; then, the newly added business data and the reference business data are used together as training samples to train a preset second generative model to obtain an updated second generative model; then, the updated second generative model is used as an updated first generative model; finally, the updated business processing model is used as the initial business processing model, and the updated first generative model is used as the preset first generative model. This allows the generation of old business data using the preset first generative model, and the memory generation capability of the preset second generative model for learning new and old business data, resulting in an updated second generative model. Before the next incremental learning, the updated second generative model is replaced by the preset first generative model, thus decoupling the process of generating old business data and learning new and old business data, thereby completely solving the catastrophic forgetting problem in incremental learning. Attached Figure Description

[0027] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0028] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0029] Figure 2 This is a flowchart of one embodiment of a method for updating a business processing model according to this application;

[0030] Figure 3 This is a schematic diagram of the method for updating the business processing model in this application;

[0031] Figure 4 This is a flowchart of one embodiment of a method for updating a business processing model according to this application;

[0032] Figure 5 This is a flowchart of one embodiment of the business processing method according to this application;

[0033] Figure 6This is a schematic diagram of an embodiment of an apparatus for updating a business processing model according to this application;

[0034] Figure 7 This is a schematic diagram of the structure of an embodiment of the business processing apparatus according to this application;

[0035] Figure 8 This is a schematic diagram of the structure of an electronic device suitable for implementing embodiments of this application. Detailed Implementation

[0036] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0037] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0038] Figure 1 An exemplary system architecture 100 is shown, in which embodiments of the navigation information prompting method or navigation information prompting device of this application can be applied.

[0039] like Figure 1 As shown, system architecture 100 may include terminal devices 101 and 102, network 103, and server 104. Network 103 serves as the medium for providing a communication link between terminal devices 101 and 102 and server 104. Network 103 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0040] Users can use terminal devices 101 and 102 to interact with server 104 via network 103, for example, to transmit target services.

[0041] Terminal devices 101 and 102 can be either hardware or software. When terminal devices 101 and 102 are hardware, they can be electronic products that allow human-computer interaction with users through one or more methods such as a keyboard, touchpad, touchscreen, remote control, voice interaction, or handwriting devices. Examples include PCs (Personal Computers), mobile phones, smartphones, PDAs (Personal Digital Assistants), wearable devices, PPCs (Pocket PCs), tablets, smart car systems, smart TVs, smart speakers, laptops, and desktop computers. When terminal devices 101 and 102 are software, they can be installed in the aforementioned electronic devices. They can be implemented as multiple software programs or software modules, or as a single software program or software module. No specific limitations are made here.

[0042] Server 104 can provide various services. For example, server 104 can obtain new business data from the initial business processing model and obtain reference business data generated by a preset first generative model; use the new business data and reference business data together as training samples to train the initial business processing model to obtain an updated business processing model; use the new business data and reference business data together as training samples to train a preset second generative model to obtain an updated second generative model; use the updated second generative model as the updated first generative model; use the updated business processing model as the initial business processing model, and use the updated first generative model as the preset first generative model.

[0043] It should be noted that server 104 can be either hardware or software. When server 104 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When server 104 is software, it can be implemented as multiple software programs or software modules (e.g., used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0044] It should be noted that the method or business processing method for updating the business processing model provided in this application embodiment is generally executed by the server 104, and correspondingly, the device or business processing device for updating the business processing model is generally located in the server 104.

[0045] It should be understood that Figure 1 The number of electronic devices, networks, and servers shown is merely illustrative. Any number of electronic devices, networks, and servers can be included depending on implementation needs.

[0046] like Figure 2The flowchart 200 illustrates an embodiment of a method for updating a business processing model according to this application, which may include the following steps:

[0047] Step 201: Obtain new business data from the initial business processing model and obtain reference business data generated by the preset first generative model.

[0048] In this embodiment, the execution entity of the method for updating the business processing model (e.g.) Figure 1 The terminal devices 101 and 102 (or server 105) shown can obtain newly added business data within a preset time period of running the initial business processing model, and obtain reference business data generated by the preset first generative model. The newly added business data can be business data newly added by the initial business processing model during a certain period of operation, which can be the time period from the time after the last update of the initial business processing model to the time before executing step 201.

[0049] In one example, the preset first generative model in step 201 can be determined based on the following steps: obtaining first initial business data; training the preset second generative model using the first initial business data to obtain the preset first initial business processing model.

[0050] In this embodiment, the preset first generative model has the ability to learn from both new and old business data. The aforementioned execution entity can use the initial business data as training samples for the preset second generative model to iteratively adjust the parameters of the preset second generative model until a preset iteration cutoff condition is met, resulting in an updated first generative model. This updated first generative model can then be replaced with the preset first generative model before the next incremental learning iteration, thereby completely solving the catastrophic forgetting problem in incremental learning. The aforementioned iteration cutoff condition can be set by the user or based on the accuracy of the updated first generative model in processing business data.

[0051] Step 202: Use the newly added business data and the reference business data together as training samples to train the initial business processing model and obtain the updated business processing model.

[0052] In this embodiment, the aforementioned execution entity can use newly added business data and reference business data as training samples for the initial business processing model to iteratively adjust the parameters of the initial business processing model until a preset iteration cutoff condition is met, thereby obtaining an updated business processing model. The aforementioned iteration cutoff condition can be set by the user or based on the processing accuracy of the updated business processing model.

[0053] Step 203: Use the newly added business data and the reference business data together as training samples to train the preset second generative model and obtain the updated second generative model.

[0054] In this embodiment, the aforementioned execution entity can use newly added business data and reference business data as training samples for a preset second generative model to iteratively adjust the parameters of the preset second generative model until a preset iteration cutoff condition is met, thereby obtaining an updated second generative model. The aforementioned iteration cutoff condition can be set by the user or based on the processing accuracy of the updated second generative model.

[0055] In this embodiment, a preset second generative model can be used to learn the memory generation capability of new and old business data to obtain an updated second generative model. Before the next incremental learning, the updated second generative model can be replaced with a preset first generative model, thus decoupling the process of generating old business data and learning new and old business data, thereby completely solving the catastrophic forgetting problem in incremental learning.

[0056] Step 204: Use the updated second generative model as the preset first generative model.

[0057] In this embodiment, after the execution entity completes step 204, it can use the updated second generative model as the preset first generative model.

[0058] In one example, the parameters of the updated second generative model are migrated or copied to the updated first generative model.

[0059] Step 205: Use the updated business processing model as the initial business processing model, and use the updated first generative model as the preset first generative model.

[0060] In this embodiment, the aforementioned execution entity can use the updated business processing model as the initial business processing model in step 201, and the updated first generative model as the preset first generative model in step 202, so as to ensure that when the initial business processing model is updated next time, it is updated based on the updated business processing model obtained in the previous update.

[0061] During the operation of the updated business processing model, its business processing performance will decrease over time. At this point, it is necessary to re-determine whether the updated business processing model in step 205 needs to be updated again, that is, to execute steps 201 to 205 again.

[0062] It should be noted that before taking the business processing model offline, the timing of updating the business processing model can be dynamically adjusted based on the updated business processing model to improve the processing performance.

[0063] It should be noted that steps 202 and 203 can be executed simultaneously. For example, the newly added business data and reference business data can be used together as training samples to train the initial business processing model and the preset second generative model, respectively, to obtain the updated business processing model and the updated second generative model. Alternatively, step 203 can be executed first, followed by step 202. Or, a portion of the training in step 202 can be performed before starting step 203. Or, a portion of the update training in step 203 can be performed first, before starting step 202.

[0064] In addition, since the update time of the initial business processing model may not be the same as the update time of the preset second generative model; for example, when the update of the business processing model is partially executed in step 202, if the update of the second generative model is completed (corresponding to step 203), the updated second generative model in step 204 can be used as the updated first generative model, and the updated first generative model in step 205 can be used as the preset first generative model. Then, the process of updating the business processing model corresponding to the remaining part in step 202 can continue to be executed.

[0065] In this embodiment, before completing the training in step 202, the updated first generative model can be replaced with a preset first generative model in advance. In the subsequent update of the initial business processing model in step 202, since the replaced first generative model has a stronger ability to learn new and old business data than the previous generative model, the forgotten business data in the old business data generated is also reduced, thereby shortening the time to solve the catastrophic forgetting problem in incremental learning.

[0066] It should be noted that steps 202 and 203 can be executed simultaneously. For example, the newly added business data and reference business data can be used together as training samples to train the initial business processing model and the preset second generative model, respectively, to obtain the updated business processing model and the updated second generative model. Alternatively, step 203 can be executed first, followed by step 202. Or, a portion of the training in step 202 can be performed first, followed by step 203. Or, a portion of the update training in step 203 can be performed first, followed by step 202.

[0067] In addition, since the update time of the initial business processing model may not be the same as the update time of the preset first generative model, after obtaining the updated second generative model and using the updated second generative model as the updated first generative model, step 205 is executed first, using the updated first generative model as the preset first generative model, and then step 201 is executed again to obtain the reference business data generated by the preset first generative model, and then participates in the remaining update steps of updating the initial business processing model in step 202.

[0068] In this embodiment, before completing the training in step 202, the updated first generative model can be replaced with a preset first generative model in advance. In the subsequent update of the initial business processing model in step 202, since the replaced first generative model has a stronger ability to learn new and old business data than the previous generative model, the forgotten business data in the old business data generated is also reduced, thereby shortening the time to solve the catastrophic forgetting problem in incremental learning.

[0069] In one example, Figure 3 In this context, M refers to the initial business processing model (i.e., it can be any machine learning model with learning capabilities). Before updating the initial business processing model, the method for updating the business processing model may further include the following steps: First, iteratively train the first generative model G1 until convergence; Second, deploy the initial business processing model M online and run it for time t; Third, incrementally train the initial business processing model M using newly added business data within time t to obtain the updated business processing model; Fourth, incrementally train the second generative model G2 using newly added business data to obtain the updated second generative model G2, and then use the updated second generative model G2 as the updated first generative model G1; Fifth, replace the first generative model G1 in step one with the updated first generative model from step four; Repeat steps three and four until the initial business processing model M is taken offline.

[0070] The incremental training of the initial business processing model M using newly added business data within time t may include the following steps: (a) generating reference business data using the first generative model G1; (b) training the initial business processing model M using the newly added business data and the reference business data; (c) repeating steps (a) and (b) until the model converges.

[0071] The incremental training of the second generative model G2 using newly added business data within time t may include the following steps: (A) Training the second generative model G2 using reference business data generated by the first generative model G1 and newly added business data until G2 converges to obtain the updated second generative model G2, and then using the updated second generative model G2 as the updated first generative model G1.

[0072] It should be noted that the execution order of steps one through four above can be freely combined according to specific circumstances. For example, steps three and four can be executed simultaneously, or step two can be executed first, followed by step one. The time t can be determined based on the storage capacity of the storage service and the precision of the initial business processing model.

[0073] The method for updating the business processing model provided in this embodiment first determines whether the initial business processing model meets preset update conditions; then, when the initial business processing model meets the preset update conditions, it acquires new business data of the initial business processing model within a preset time period and acquires reference business data generated by a preset first generative model; then, it uses the new business data and the reference business data together as training samples to train the initial business processing model to obtain the updated business processing model; then, it uses the new business data and the reference business data together as training samples to train a preset second generative model to obtain the updated second generative model; then, it uses the updated second generative model as the updated first generative model; finally, it uses the updated business processing model as the initial business processing model and uses the updated second generative model as the preset first generative model. It can generate old business data using a preset first generative model and use a preset second generative model to learn the memory generation capability of new and old business data to obtain an updated second generative model. Before the next incremental learning, the updated second generative model is replaced with the preset first generative model, thus decoupling the process of generating old business data and learning new and old business data, thereby completely solving the catastrophic forgetting problem in incremental learning.

[0074] In some optional implementations of this embodiment, the preset update conditions include: the accuracy of the initial business processing model meets a preset accuracy threshold; and / or the number of new business data in the initial business processing model meets a preset quantity threshold.

[0075] In this implementation, the precision of the initial business processing model can be the precision of the initial business processing model in business processing. The new business data of the initial business processing model can be all new business data generated between the time of the last update of the initial business processing model and the time of this update of the initial business processing model (execution of step 201). The quantity of new business data of the initial business processing model can be used to characterize the storage service capacity of the initial business processing model.

[0076] In one example, the accuracy of the initial business model being less than a preset accuracy threshold and / or the number of new business data in the initial business processing model being greater than or equal to a preset quantity threshold can be converted into a corresponding preset time. For example, if the accuracy of the initial business model is less than the preset accuracy threshold and / or the number of new business data in the initial business processing model is greater than or equal to the preset quantity threshold after the last update of the preset time of the initial business model, then the initial business processing model can be updated again after the last update of the preset time of the initial business model, without needing to check the accuracy of the initial business model and the number of new business data again, making the operation simpler and faster.

[0077] It should be noted that the above preset time can be set according to the user's requirements for the processing performance of the initial business model.

[0078] It should be noted that the new business data in step 201 above can be part or all of the new business data in the initial business processing model.

[0079] In this implementation, the preset update conditions can be used to dynamically update the initial business processing model so that the updated business processing model has higher precision business processing capabilities.

[0080] In some optional implementations of this embodiment, the preset first generative model and the preset second generative model have the same structure.

[0081] In this implementation, the preset first generative model and the preset second generative model can have the same structure, so that the parameters of the preset second generative model can be better updated to the preset first generative model. This allows the preset second generative model to learn new and old business data, which is more conducive to solving the catastrophic forgetting problem in the incremental learning process.

[0082] In some optional implementations of this embodiment, the initial business processing model can be determined based on the following steps: obtaining second initial business data; training a machine learning model using the second initial business data to obtain the initial business processing model.

[0083] In this implementation, before executing steps 201 to 206, the aforementioned executing entity can pre-train the machine learning model using the second initial business data to iterate the parameters of the initial business processing model until a preset iteration cutoff condition is met. The second initial business data can be business-related data, such as images, text, or audio. The machine learning model can be a neural network model, such as a classification model, a probabilistic model, or another classifier. The second initial business data can be the same as or different from the first initial business data.

[0084] In this implementation, the device learning model can be pre-trained using the second initial service data to obtain the initial service processing model.

[0085] In some optional implementations of this embodiment, obtaining new business data from the initial business processing model and obtaining reference business data generated by the preset first generative model includes: determining whether the initial business model meets preset update conditions; if so, obtaining new business data from the initial business processing model and obtaining reference business data generated by the preset first generative model.

[0086] In this implementation, the aforementioned execution entity can determine whether the initial business processing model meets preset update conditions. The initial business processing model can be used to process business operations. This initial business processing model can be the business processing model before the update, and the business operations can include speech recognition, image recognition, item recommendation, and business demand prediction, etc. The preset update conditions can be set based on the processing performance of the initial business processing model and / or set by the user. The initial business processing model can be an untrained model or a trained model.

[0087] In one example, after the initial business processing model has been running for a period of time, its performance can be monitored to determine whether an update is needed. For instance, if the initial business processing model's performance falls below a preset performance threshold, it can be updated. This preset performance threshold can be set based on the user's performance requirements.

[0088] It should be noted that the above processing performance can be determined by the current storage capacity and / or the accuracy of the processing business in the initial business processing model.

[0089] In this embodiment, the aforementioned execution entity can, upon determining that the preset update conditions for updating the initial business processing model are met, acquire newly added business data within a preset time period of running the initial business processing model, and acquire reference business data generated by a preset first generative model. The newly added business data can be business data newly added during the initial business processing model's operation for a period of time, which can be the time period from the end of the last update of the initial business processing model to the time before execution step 201. The preset first generative model can be determined based on the following steps: acquiring first initial business data; training a machine learning model using the first initial business data to obtain the preset first generative model.

[0090] It should be noted that the aforementioned time period can also be any time period between the time after the last update of the initial business processing model ends and the time before the execution of step 201, such as any time period t in a time period T.

[0091] In one example, the above-mentioned acquisition of new business data in the initial business processing model may include: acquiring new business data in the initial business processing model within any time period t.

[0092] In this embodiment, a portion of the newly added business data from all the newly added business data before the execution step 201 can be used to update the initial business processing model, thereby accelerating the update speed of the initial business processing model.

[0093] Furthermore, the aforementioned reference business data can be reference business data generated during the operation of a preset first generative model. This preset first generative model can be used to generate older business data, such as GAN (Genericaccess networks) or VEA (Variational Auto-Encoder). The aforementioned preset first generative model can be used to generate reference business data, which can be business data generated during the operation of the preset first generative model.

[0094] It should be noted that the aforementioned reference business data can be business data generated during the operation of the preset first generative model within the aforementioned period, or it can be reference business data generated within a period unrelated to the aforementioned period.

[0095] In this implementation, the business processing model and the generative model can be updated flexibly according to the update conditions.

[0096] In some optional implementations of this embodiment, the method for updating the business model further includes: repeatedly executing the above steps until the initial business processing model no longer meets the preset update conditions.

[0097] In this implementation, the performance of the updated business processing model will decrease over time during operation. At this point, it is necessary to re-determine whether the updated business processing model in step 205 needs to be updated again, that is, to execute steps 201 to 205 again.

[0098] It should be noted that before taking the business processing model offline, the timing of updating the business processing model can be dynamically adjusted based on the updated business processing model to improve the processing performance.

[0099] like Figure 4 The flowchart 400 illustrates an embodiment of a method for updating a business processing model according to this application, which may include the following steps:

[0100] Step 401: Determine whether the initial business processing model meets the preset update conditions.

[0101] In this embodiment, the execution entity of the method for updating the business processing model (e.g.) Figure 1 The terminal devices 101 and 102 (or server 105) shown can determine whether the initial service processing model meets the preset update conditions. The initial service processing model can be used to process services. This initial service processing model can be the service processing model before the update. The services can include voice recognition, image recognition, item recommendation, and service demand prediction, etc. The preset update conditions can be set based on the processing performance settings of the initial service processing model and / or set by the user.

[0102] Step 402: If yes, obtain the new business data of the initial business processing model within the preset time period, and obtain the reference business data generated by the preset first generative model.

[0103] In this embodiment, the aforementioned execution entity can, upon determining that the preset update conditions for updating the initial business processing model are met, obtain newly added business data within a preset time period of running the initial business processing model, and obtain reference business data generated by the preset first generative model. The newly added business data can be business data newly added by the initial business processing model during a certain period of operation. This period can be the time interval from the end of the last update of the initial business processing model to the time before executing step 401.

[0104] Step 403: Use the newly added business data and the reference business data as training samples to train the initial business processing model and the preset second generative model respectively, so as to obtain the updated business processing model and the updated second generative model.

[0105] In this embodiment, the aforementioned execution entity can use newly added business data and reference business data as training samples for the initial business processing model, and train the initial business processing model and the preset second generative model respectively, to iteratively adjust the parameters of the initial business processing model until a preset iteration cutoff condition is met, thereby obtaining an updated business processing model; and iteratively adjust the parameters of the preset second generative model until a preset iteration cutoff condition is met, thereby obtaining an updated second generative model. The aforementioned iteration cutoff condition can be set by the user or based on the accuracy of the updated business processing model in processing business.

[0106] Step 404: Use the updated second generative model as the preset first generative model.

[0107] In this embodiment, after the execution entity completes step 404, it can use the updated second generative model as the preset first generative model.

[0108] Step 405: Use the updated business processing model as the initial business processing model, and use the updated first generative model as the preset first generative model.

[0109] In this embodiment, the specific operations of steps 402, 404, and 405 are the same as... Figure 2 The operations of steps 202, 204, and 205 in the illustrated embodiments are basically the same, and will not be described again here.

[0110] from Figure 4 It can be seen from this that, with Figure 2Compared to the corresponding embodiments, the method 400 for updating the business processing model in this embodiment first determines whether the initial business processing model meets the preset update conditions; then, when the initial business processing model meets the preset update conditions, it acquires the new business data of the initial business processing model within a preset time period and acquires the reference business data generated by the preset first generative model; then, it uses the new business data and the reference business data together as training samples to train the initial business processing model and the preset second generative model respectively, obtaining the updated business processing model and the updated second generative model; then, it uses the updated second generative model as the updated first generative model; then, it uses the updated business processing model as the initial business processing model and uses the updated second generative model as the preset first generative model. This method can utilize the preset first generative model to generate old business data and utilize the preset second generative model to learn the memory generation capability of new and old business data, obtaining the updated second generative model. Before the next incremental learning, the updated second generative model is replaced with the preset first generative model, thus decoupling the process of generating old business data and learning new and old business data, thereby completely solving the catastrophic forgetting problem in incremental learning.

[0111] like Figure 5 The flowchart 500 illustrates an embodiment of a business processing method according to this application, which may include the following steps:

[0112] Step 501: Obtain the target business.

[0113] In this embodiment, the execution body of the business processing method (e.g. Figure 1 The terminal in the process can retrieve the target service from its storage or receive the target service input from external sources, such as user-inputted target services or target services transmitted from other terminals. Alternatively, the execution body of the service processing method (e.g., Figure 1 The server in the middle can retrieve the target service from its storage, or receive the terminal (e.g., Figure 1 The target service is sent by the terminal in the process. The target service can be the service to be executed.

[0114] Step 502: Perform preset business processing on the target business according to the updated business processing model.

[0115] In this embodiment, the aforementioned execution entity can, according to Figure 2 , Figure 4 The updated business processing model (which can be the business processing model after each update of the initial business processing model) or the initial business processing model performs preset business processing on the target business.

[0116] In one example, the aforementioned execution entity inputs the target business into the updated business processing model to perform a preset business processing on the target business.

[0117] In this embodiment, the business processing method 500 first obtains the target business; then, using the above-mentioned business processing model, while ensuring the high accuracy of the business processing model, it performs preset business processing on the target business.

[0118] In some optional implementations of this embodiment, the preset business processing includes at least one of the following: identification of target business, recommendation of target business, and prediction of demand for target business.

[0119] In this implementation, the identification of the target service can be achieved by using the service processing model in step 502 to identify the target service in step 501, thereby obtaining the identification result of the target service. The target service may include at least one of the following: image, audio, and text.

[0120] In one example, the target image is input into the business processing model in step 502 to obtain the category of the target object in the target image, and also to obtain the location of the target object in the target image.

[0121] In this implementation, the recommendation of the target service can be achieved by recommending the target service to the user in step 501 through the business processing model in step 502. The target service can be any service that can be shared with other users, such as a product.

[0122] In one example, user behavior data related to the target product is input into the business processing model in step 502 to obtain product recommendation results. The aforementioned user behavior can be actions performed by the user on the product, such as browsing, adding to favorites, or purchasing.

[0123] In this implementation, the demand for the target service can be predicted using the business processing model in step 502, based on the target service in step 501. The target service can be any service that can be shared with other users, such as a product.

[0124] In one example, the demand characteristics of the target product within a preset time period are input into the business processing model in step 502 to obtain the demand quantity of the target product.

[0125] Further reference Figure 6 As an implementation of the methods shown in the above figures, this application discloses an embodiment of an apparatus for updating a business processing model, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0126] like Figure 6 As shown, an embodiment of this application provides an apparatus 600 for updating a business processing model. The apparatus 600 includes: a data acquisition module 601, a first update module 602, a second update module 603, and a model-as-a-model module 604. Specifically, the data acquisition module 601 is configured to acquire new business data from an initial business processing model and acquire reference business data generated by a preset first generative model; the first update module 602 is configured to use the new business data and reference business data together as training samples to train the initial business processing model, obtaining an updated business processing model; and the second update module 603 is configured to use the new business data and reference business data together as training samples to train a preset second generative model, obtaining an updated second generative model; and use the updated second generative model as an updated first generative model; the model-as-a-model module 604 is configured to use the updated business processing model as the initial business processing model and use the updated first generative model as a preset first generative model.

[0127] In this embodiment, the specific processing of the data acquisition module 601, the first update module 602, the second update module 603, and the model as module 604 in the device 600 for updating the business processing model, and the resulting technical effects, can be referred to respectively. Figure 2 Steps 201 to 205 in the corresponding embodiment.

[0128] In some embodiments, the preset first generative model and the preset second generative model are models with the same structure.

[0129] In some embodiments, the preset update conditions include: the accuracy of the initial business processing model meets a preset accuracy threshold; and / or, the number of new business data in the initial business processing model meets a preset quantity threshold.

[0130] In some embodiments, the apparatus for updating the business processing model further includes: a data acquisition module configured to acquire second initial business data; and a model training module configured to train a machine learning model using the second initial business data to obtain an initial business processing model.

[0131] In some embodiments, the data acquisition module 601 includes: a condition judgment unit configured to judge whether the initial business model meets the preset update conditions; and a data acquisition unit configured to, if so, acquire the new business data of the initial business processing model and acquire the reference business data generated by the preset first generative model.

[0132] In some embodiments, the apparatus for updating the business processing model further includes:

[0133] The loop module is configured to execute the above steps repeatedly until the initial business processing model no longer meets the preset update conditions.

[0134] The apparatus for updating a business processing model provided in the embodiments of this application firstly obtains new business data of the initial business processing model within a preset time period by a data acquisition module 601, and obtains reference business data generated by a preset first generative model; then, a first update module 602 uses the new business data and the reference business data together as training samples to train the initial business processing model to obtain an updated business processing model; then, a second update module 603 uses the new business data and the reference business data together as training samples to train a preset second generative model to obtain an updated second generative model; the updated second generative model is used as the updated first generative model; then, a model-as-a-model module 604 uses the updated business processing model as the initial business processing model, and uses the updated first generative model as the preset first generative model. It can generate old business data using a preset first generative model and use a preset second generative model to learn the memory generation capability of new and old business data to obtain an updated second generative model. Before the next incremental learning, the updated second generative model is replaced with the preset first generative model, thus decoupling the process of generating old business data and learning new and old business data, thereby completely solving the catastrophic forgetting problem in incremental learning.

[0135] Further reference Figure 7 As an implementation of the methods shown in the above figures, this application discloses an embodiment of a service processing apparatus, which is similar to... Figure 5 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0136] like Figure 7 As shown, an embodiment of this application provides a service processing apparatus 700, which includes a service acquisition module 701 and a service processing module 702. The service acquisition module 701 is configured to acquire a target service; the service processing module 702 is configured to perform preset service processing on the target service according to any of the aforementioned service processing models.

[0137] In this embodiment, the specific processing of the service acquisition module 701 and the service processing module 702 in the service processing device 700 and the resulting technical effects can be referred to respectively. Figure 5 Steps 501 to 502 in the corresponding embodiment.

[0138] In some embodiments, the preset business processing includes at least one of the following: identification of target business, recommendation of target business, and prediction of demand for target business.

[0139] The service processing apparatus provided in the embodiments of this application acquires a target service through a service acquisition module 701; then, through a service processing module 702, it performs preset service processing on the target service while ensuring the high accuracy of the service processing model described above.

[0140] like Figure 8 The diagram shown is a block diagram of an electronic device for a method of updating a business processing model according to an embodiment of this application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present application described and / or claimed herein.

[0141] like Figure 8 As shown, the electronic device includes one or more processors 801, a memory 802, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components are interconnected via different buses and can be mounted on a common motherboard or otherwise as required. The processors can process instructions executed within the electronic device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In other embodiments, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple electronic devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 8 Take the 801 processor as an example.

[0142] The memory 802 is the non-transitory computer-readable storage medium provided in this application. The memory stores instructions executable by at least one processor to cause at least one processor to perform the method or business processing method for updating the business processing model provided in this application. The non-transitory computer-readable storage medium of this application stores computer instructions for causing a computer to perform the method or business processing method for updating the business processing model provided in this application.

[0143] Memory 802, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the method for updating the business processing model or the program instructions / modules corresponding to the business processing method in the embodiments of this application (e.g., appendix). Figure 6 The data acquisition module 601, the first update module 602, the second update module 603, and the model are shown as module 604; or Figure 7 (The business acquisition module 701 and business processing module 702 are shown). The processor 801 executes various functional applications and data processing of the server by running non-transient software programs, instructions and modules stored in the memory 802, that is, implementing the method or business processing method for updating the business processing model in the above method embodiments.

[0144] The memory 802 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the blockchain-based information processing electronic device. Furthermore, the memory 802 may include high-speed random access memory and may also include non-transient memory, such as at least one disk storage device, flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 802 may optionally include memory remotely located relative to the processor 801, and these remote memories can be connected to the blockchain-based information processing electronic device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0145] The electronic device for updating the business processing model or business processing method may further include: an input device 803 and an output device 804. The processor 801, memory 802, input device 803, and output device 804 can be connected via a bus or other means. Figure 8 Taking the example of a connection between China and Israel via a bus.

[0146] Input device 803 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of blockchain-based information processing electronic devices, such as touchscreens, keypads, mice, trackpads, touchpads, joysticks, one or more mouse buttons, trackballs, joysticks, etc. Output device 804 may include display devices, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibration motors). The display device may include, but is not limited to, liquid crystal displays (LCDs), light-emitting diode (LED) displays, and plasma displays. In some embodiments, the display device may be a touchscreen.

[0147] Various implementations of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, application-specific integrated circuits (ASICs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transferring data and instructions to the storage system, the at least one input device, and the at least one output device.

[0148] These computational programs (also referred to as programs, software, software applications, or code) include machine instructions for a programmable processor and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms “machine-readable medium” and “computer-readable medium” refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) used to provide machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term “machine-readable signal” refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0149] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0150] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0151] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0152] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this application can be achieved, and this is not limited herein.

[0153] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for updating a business processing model, comprising: obtaining new business data of an initial business processing model, and obtaining reference business data generated by a preset first generative model, wherein the new business data and the reference business data correspond to at least one of demand prediction of a commodity, identification of the commodity, and recommendation of the commodity, and the business data comprises at least one of image, audio, and text of the commodity, and the initial business processing model is used for corresponding identification of at least one of a target commodity, demand prediction of the target commodity, and recommendation of the target commodity; training the initial business processing model by taking the new business data and the reference business data as training samples to obtain an updated business processing model, comprising: taking the new business data and the reference business data as training samples of the initial business processing model to iteratively adjust parameters of the initial business processing model until a preset iteration stop condition is met to obtain the updated business processing model; and training a preset second generative model by taking the new business data and the reference business data as training samples to obtain an updated second generative model, comprising: taking the new business data and the reference business data as training samples of the preset second generative model to iteratively adjust parameters of the preset second generative model until a preset iteration stop condition is met to obtain the updated second generative model; taking the updated second generative model as the updated first generative model; taking the updated business processing model as the initial business processing model, and taking the updated first generative model as the preset first generative model.

2. The method of claim 1, wherein, The preset first generative model and the preset second generative model are models with the same structure.

3. The method of claim 1, wherein, The preset update condition comprises: accuracy of the initial business processing model satisfying a preset accuracy threshold; and / or, quantity of the new business data of the initial business processing model satisfying a preset quantity threshold.

4. The method of claim 1, wherein, The initial business processing model is determined based on the following steps: obtaining second initial business data; training a machine learning model by using the second initial business data to obtain the initial business processing model.

5. The method according to any one of claims 1 to 4, wherein, The obtaining of the new business data of the initial business processing model and the reference business data generated by the preset first generative model comprises: determining whether the initial business model satisfies a preset update condition; if yes, obtaining the new business data of the initial business processing model and the reference business data generated by the preset first generative model.

6. The method of claim 5, further comprising: recursively performing the above steps until the initial business processing model does not satisfy the preset update condition.

7. A business processing method, comprising: obtaining a target business; performing a preset business processing on the target business according to any one of the business processing models of claims 1-6.

8. The method of claim 7, wherein, The preset business processing includes at least one of the following: identification of a target business, recommendation of the target business, and demand prediction of the target business.

9. An apparatus for updating a business processing model, comprising: a data acquisition module configured to acquire new business data of an initial business processing model and to acquire reference business data generated by a preset first generative model, wherein the business corresponding to the new business data and the reference business data is at least one of demand prediction of a commodity, identification of the commodity, and recommendation of the commodity, and the business data includes at least one of an image, audio, and text of the commodity, and the initial business processing model is used to correspondingly identify at least one of a target commodity, demand prediction of the target commodity, and recommendation of the target commodity; a first update module configured to jointly use the new business data and the reference business data as training samples to train the initial business processing model to obtain an updated business processing model, including: using the new business data and the reference business data as training samples of the initial business processing model to iteratively adjust parameters of the initial business processing model until a preset iteration stop condition is met to obtain the updated business processing model; and a second update module configured to jointly use the new business data and the reference business data as training samples to train a preset second generative model to obtain an updated second generative model, and to use the updated second generative model as the updated first generative model, including: using the new business data and the reference business data as training samples of the preset second generative model to iteratively adjust parameters of the preset second generative model until a preset iteration stop condition is met to obtain the updated second generative model; a model as module configured to use the updated business processing model as the initial business processing model and to use the updated first generative model as the preset first generative model.

10. The apparatus of claim 9, wherein, The preset first generative model and the preset second generative model are models with the same structure.

11. The apparatus of claim 9, wherein, The preset update condition includes: the accuracy of the initial business processing model meets a preset accuracy threshold; and / or, the number of new business data of the initial business processing model meets a preset number threshold.

12. The apparatus of claim 9, further comprising: the data acquisition module is configured to acquire second initial business data; a model training module configured to train a machine learning model using the second initial business data to obtain the initial business processing model.

13. The apparatus of any of claims 9-12, wherein, the data acquisition module includes: a condition judgment unit configured to judge whether the initial business model meets a preset update condition; a data acquisition unit configured to, if so, acquire new business data of the initial business processing model and to acquire reference business data generated by a preset first generative model.

14. A business processing apparatus, comprising: a business acquisition module configured to acquire a target business; a business processing module configured to perform a preset business processing on the target business according to any one of the business processing models in claims 1-6.

15. The apparatus of claim 14, wherein, The preset business processing comprises at least one of the following: identification of the target business, recommendation of the target business, and prediction of demand volume of the target business. 16.An electronic device, comprising: one or more processors; a memory device having stored thereon 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 method according to any one of claims 1-8.

17. A non-transitory computer readable storage medium having computer instructions stored thereon, having a computer program stored thereon, wherein, The program, when executed by a processor, implements the method according to any one of claims 1-8.

Images