KR102531299B1 - A learning model recommendation device based on similarity and a cloud integrated operating system comprising the same - Google Patents

Abstract

An apparatus for recommending a learning model according to an embodiment of the present invention is a learning model recommendation apparatus for recommending a learning model calculated by machine learning or deep learning according to a request of a consumer, wherein the request of the consumer and the learning model requested by the consumer a receiving module for receiving calculation conditions of the target learning model; a storage module in which a stored learning model, which is a learning model created in the past, and calculation conditions of the stored learning model are stored; a similarity determination module for determining a degree of similarity between the target learning model and the stored learning model; and a recommendation module for recommending, as a recommendation learning model, a learning model that satisfies a predetermined recommendation condition among the stored learning models stored in the storage module to a consumer, wherein the predetermined recommendation condition includes the target learning model and the second It may be a similar condition with a degree of similarity of 1 or higher.

Description

A learning model recommendation device based on similarity and a cloud integrated operating system including the same {A LEARNING MODEL RECOMMENDATION DEVICE BASED ON SIMILARITY AND A CLOUD INTEGRATED OPERATING SYSTEM COMPRISING THE SAME}

The present invention relates to a technology for recommending a learning model, and more particularly, to a learning model recommendation device based on similarity for recommending a learning model that meets user needs and a cloud integrated operating system including the same.

Recently, as the cloud market gradually grows, various services are being built based on the cloud, and the scale of the cloud is also growing explosively. Since it is realistically difficult for an administrator to manage hundreds or thousands of virtual machines individually, a technology that systematically and automatically operates and manages them is being developed. Here, the key to automatic operation is to implement functions necessary for cloud operation by using a learning model through machine learning/deep learning.

Unlike the existing method of setting all analysis algorithms, machine learning overcomes the limitations of existing computer algorithms in that it creates a model suitable for the purpose by itself based on given data. Because of these excellent functions and convenience, it is being developed and commercialized in various industrial fields.

However, machine learning or deep learning incurs a lot of time and cost, such as collecting and labeling a lot of data, and learning models that require a lot of resources are treated as the own assets of each development company and are stored and used independently without being disclosed. It is becoming. As a result, learning models with the same purpose are produced and utilized in many places, resulting in a social problem of wasting resources.

In addition, as the number of learning models and the amount of analysis used to efficiently operate the cloud increase, the operational load increases, resulting in a problem of not efficiently managing the cloud.

The present invention is to solve the above problems, and relates to a learning model recommendation device capable of recommending the most similar learning model to suit user needs and activating distribution of the learning model, and a cloud integrated operating system including the same.

An apparatus for recommending a learning model according to an embodiment of the present invention is a learning model recommendation apparatus for recommending a learning model calculated by machine learning or deep learning according to a request of a consumer, wherein the request of the consumer and the learning model requested by the consumer a receiving module for receiving calculation conditions of the target learning model; a storage module in which a stored learning model, which is a learning model created in the past, and calculation conditions of the stored learning model are stored; a similarity determination module for determining a degree of similarity between the target learning model and the stored learning model; and a recommendation module for recommending, as a recommendation learning model, a learning model that satisfies a predetermined recommendation condition among the stored learning models stored in the storage module to a consumer, wherein the predetermined recommendation condition includes the target learning model and the second It may be a similar condition with a degree of similarity of 1 or higher.

In addition, the recommendation module calculates the recommendation learning model according to a predetermined calculation method when a first additional condition is satisfied even if the predetermined recommendation condition is not satisfied, and the first additional condition is determined. , There are a predetermined number or less of the clusters of the stored learning models that are similar to the target learning model by a similarity of at least the second similarity, are similar to the learning model by less than the first similarity, and exist at a similarity of less than the second based on the target learning model. may be a condition.

In addition, in the predetermined calculation method, when there is one cluster of the stored learning models existing at the second similarity or higher based on the target learning model, the stored learning model closest to the average of the clusters is selected as the recommended learning model. may be a method of selection.

In addition, in the predetermined calculation method, when the number of clusters of the stored learning models that exist with the second similarity or higher based on the target learning model exceeds one and is less than or equal to a predetermined number, the stored learning model closest to the average of each cluster. It may be a method of selecting the recommendation learning model with a combination of

In addition, the recommendation module selects the recommended learning model based on the stored learning models for which a second additional condition is satisfied even if the predetermined recommendation condition and the first additional condition are not satisfied, The predetermined second additional condition may be a condition that, when the stored learning models that are similar to the target learning model by less than the second similarity are connected, the similarity with the target learning model is equal to or greater than the first similarity.

In addition, when an arbitrary learning model is connected to each other, a simulation module for simulating the state of the final learning model; wherein the simulation module calculates the state of the final learning model when a plurality of learning models are input A simulation model can be created through deep learning.

In addition, when a predetermined learning condition is satisfied, a learning suggestion module for proposing to the manager to create a learning model; the predetermined learning condition may be a condition in which the predetermined recommendation condition is not satisfied. .

The price calculation module further includes a price calculation module that calculates a cost to be charged while recommending the stored learning model, wherein the price calculation module includes a cost to be charged based on a degree of similarity between the target learning model and the recommended learning model to be recommended. can be calculated.

A learning model recommendation method according to an embodiment of the present invention is implemented by a learning model recommendation device and recommends a learning model calculated by machine learning or deep learning according to a request of a consumer. A request receiving step in which a request of a consumer and a calculation condition of a target learning model, which is a learning model requested by the consumer, are received by the module; a condition storage step of storing a stored learning model, which is a learning model generated in the past, and a calculation condition of the stored learning model by a storage module; a similarity determination step of determining a similarity between the target learning model and the stored learning model by a similarity determination module; and a recommending step of recommending, by a recommendation module, a learning model satisfying a predetermined recommendation condition among the stored learning models stored in the storage module to a consumer, wherein the predetermined recommendation condition includes the target learning model and the second learning model. It may be a similar condition with a degree of similarity of 1 or higher.

In addition, in the recommending step, even if the predetermined recommendation condition is not satisfied, a predetermined first additional condition for determining whether the condition is satisfied based on a second similarity lower than the first similarity and a plurality of the stored learning models It may be a step of selecting or calculating a learning model to be recommended to a consumer by determining whether a predetermined second additional condition for determining whether the condition is satisfied is satisfied based on the virtual learning model created by combining the virtual learning model.

A monitoring device according to an embodiment of the present invention, in the monitoring device for monitoring the operation of at least one virtual machine running on a physical server within the allocated resources of a cloud integrated operating system, the operation occurring in the virtual machine Transmitting and receiving module for collecting information; a monitoring module for monitoring whether the virtual machine operates non-ideally based on the operation information collected by the transmission/reception module; Based on the request of the monitoring module, a model operating module that activates a monitoring learning model that helps the monitoring function of the monitoring module; and a schedule module for determining an activation timing of the monitoring learning model through a predetermined timing determination method so that the monitoring learning model can be operated within the allocated resource.

In addition, the monitoring learning model includes a first monitoring learning model that predicts a load generated in the cloud integrated operating system for a predetermined future period, and the schedule module includes a load predicted by the first monitoring learning model. Based on, the operation timing can be determined.

In addition, the method for determining the predetermined time point may be a method for determining the operation time at a time when a load is generated at a predetermined rate or less based on allocated resources.

Also, the method for determining the predetermined time point may be a method of determining the operation time at a time when the least load is generated among the predetermined future periods.

In addition, a model selection module for selecting the monitoring learning model satisfying a predetermined operating condition may be further included, and the predetermined operating condition may be a condition in which a load lower than a predetermined threshold value is applied.

In addition, the learning model includes a first monitoring learning model that predicts a load generated in the cloud integrated operating system for a predetermined future period, and the model selection module includes a load predicted by the first monitoring learning model. It is possible to change the predetermined threshold based on.

Cloud integrated operating system according to an embodiment of the present invention, the management device for managing the virtual machine running on the physical server; a monitoring device for monitoring the operation of the virtual machine; and a learning model recommendation device for recommending a monitoring learning model necessary for the monitoring device to implement a monitoring function, wherein the monitoring device includes a transmission/reception module that collects information generated from the virtual machine, and the transmission/reception module collects information. A monitoring module that monitors whether the virtual machine operates abnormally based on the information of the virtual machine, and a model operation that activates the monitoring learning model that helps the monitoring function of the monitoring module based on the request of the monitoring module. module, a schedule module that determines the operation timing of the monitoring learning model through a predetermined time point determination method so that the monitoring learning model can be operated within the allocated resources, and operation in the model operation module when predetermined operation conditions are satisfied and a model selection module for selecting the monitoring learning model to be selected, wherein the model selection module requests the learning model recommendation device to recommend a monitoring learning model for which the predetermined operation condition is satisfied, and the learning model The recommendation device may recommend a learning model that satisfies a predetermined recommendation condition, which is similar to the monitoring learning model by a first degree of similarity or higher, among pre-stored stored learning models.

In addition, even if there is no learning model that satisfies the predetermined recommendation condition among pre-stored stored learning models, the learning model recommendation device determines whether the condition is satisfied based on a second similarity lower than the first similarity. Based on the first additional condition or a virtual learning model generated by combining a plurality of the stored learning models, a learning model satisfying a predetermined second additional condition for determining whether the condition is satisfied may be recommended.

In addition, the predetermined first additional condition is similar to the monitoring learning model to a second similarity or more, similar to the first similarity to less than the first similarity, and based on the monitoring learning model, the pre-stored information that is less than the second similarity. This may be a condition in which there are less than a predetermined number of clusters of stored learning models.

A monitoring method according to an embodiment of the present invention is a monitoring method for monitoring the operation of at least one virtual machine running in a physical server within an allocated resource of a cloud integrated operating system through a monitoring device, By, collecting operation information generated in the virtual machine; Monitoring, by a monitoring module, whether or not the virtual machine operates non-ideally based on the operation information; activating, by a model operation module, a monitoring learning model that helps the monitoring function of the monitoring module, based on a request of the monitoring module; and determining, by a schedule module, an activation timing of the monitoring learning model through a predetermined timing determination method so that the monitoring learning model can be operated within allocated resources.

The apparatus for recommending a learning model based on the degree of similarity according to the present invention and the cloud integrated operating system including the same can reduce social resource costs.

In addition, it is possible to efficiently operate the system.

In addition, development time can be drastically reduced.

In addition, overloading of the system can be prevented.

However, the effects of the present invention are not limited to the above-mentioned effects, and effects not mentioned will be clearly understood by those skilled in the art from this specification and the accompanying drawings.

1 is a relationship diagram of a cloud integrated operating system according to an embodiment of the present invention
2 is a block diagram of a cloud integrated operating system according to an embodiment of the present invention
3 is a configuration relationship diagram of a monitoring device of a cloud integrated operating system according to an embodiment of the present invention
4 is a flowchart of a learning model recommendation method implemented by a learning model recommendation device of a cloud integrated operating system according to an embodiment of the present invention.
5 is a diagram for explaining a predetermined recommendation condition and a predetermined first additional condition in a learning model recommendation method according to an embodiment of the present invention.
6 is a diagram showing a learning model mediation interface screen provided by an interface module of a learning model recommendation device according to an embodiment of the present invention.
7 is a flowchart illustrating a process of selecting a learning model to be recommended to a monitoring device by an apparatus for recommending a learning model according to an embodiment of the present invention.
8 is a view for explaining a method for calculating a first operating time and a second operating time by a model selection module of a monitoring device according to an embodiment of the present invention;

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the drawings. However, the spirit of the present invention is not limited to the presented embodiments, and those skilled in the art who understand the spirit of the present invention may add, change, delete, etc. other elements within the scope of the same spirit, through other degenerative inventions or the present invention. Other embodiments included within the scope of the inventive idea can be easily proposed, but it will also be said to be included within the scope of the inventive concept.

1 is a relationship diagram of a cloud integrated operating system according to an embodiment of the present invention.

Referring to FIG. 1, the cloud integrated operating system 10 according to an embodiment of the present invention manages and operates a virtual machine running on a physical server, and sells, rents, and provides cloud services to cloud users (learning models required by consumers). It may be a system that provides

To this end, the cloud integrated operating system 10 is connected to the consumer M10 and the physical server S10 through a wired/wireless network, and may transmit and receive necessary information.

As an example, the consumer M10 may be a person who is provided with a virtualized cloud service.

For example, the consumer (M100) can transmit conditions and requests for a learning model to be purchased or rented to the learning model recommendation device, and the learning model recommendation device is an interface capable of requesting and transmitting a learning model and learning model to the consumer etc. can be provided.

The cloud integrated operating system 10 may be a system that recommends a necessary cloud service or learning model according to the request when there is a request from a consumer (client).

In addition, the learning model recommendation method may provide an interface so that the learning model can be used in a cloud environment.

The network in the present invention may be a core network integrated with a wired public network, a wireless mobile communication network, or a mobile Internet, etc., and various services existing in the TCP/IP protocol and its upper layer, that is, HTTP (Hyper Text Transfer Protocol), HTTPS (Hyper Text Transfer Protocol Secure), Telnet, FTP (File Transfer Protocol), DNS (Domain Name System), SMTP (Simple Mail Transfer Protocol), etc. It comprehensively means a data communication network capable of transmitting and receiving data in various forms without limitation.

Hereinafter, the cloud integrated operating system will be described in detail.

Figure 2 is a configuration diagram of a cloud integrated operating system according to an embodiment of the present invention, Figure 3 is a configuration relationship diagram of a monitoring device of the cloud integrated operating system according to an embodiment of the present invention.

2 and 3, the cloud integrated operating system according to an embodiment of the present invention includes a learning model recommendation device 100 for recommending a learning model at the request of a consumer, and managing virtual machines running on a physical server. It may include a management device and a monitoring device for monitoring the operation of the virtual machine.

Hereinafter, each device will be described in detail.

Learning model recommendation apparatus 100 according to an embodiment of the present invention, in the learning model recommendation apparatus 100 for recommending a learning model calculated by machine learning or deep learning according to a request of a consumer, A receiving module 110 that receives calculation conditions of a target learning model, which is a learning model requested by a consumer, a storage learning model, which is a learning model created in the past, and a first storage module 120 in which calculation conditions of the storage learning model are stored. ), the similarity determination module 130 for determining the degree of similarity between the target learning model and the stored learning model, and the learning model that satisfies a predetermined recommendation condition among the stored learning models stored in the first storage module 120. It may include a recommendation module 140 that recommends to consumers as a model.

In addition, the learning model recommendation device 100 may further include a simulation module 150 for simulating a state of a final learning model when an arbitrary learning model is connected to each other.

In addition, the learning model recommendation device 100 may further include a learning suggestion module 160 that proposes generation of a learning model to a manager when a predetermined learning condition is satisfied.

In addition, the learning model recommendation device 100 may further include a price calculation module 170 for calculating a cost to be charged while recommending the stored learning model.

In addition, the learning model recommendation device 100 may further include an interface module 180 that calculates and transmits an interface capable of displaying and collecting information necessary for implementing the learning model recommendation method on a manager's or consumer's computing device. there is.

The receiving module 110 may collect information necessary for implementing the learning model recommendation method.

For example, information requested by the consumer and conditions of the learning model required by the consumer may receive calculation conditions.

A consumer may input a learning model request and calculation conditions through an interface provided by the interface module 180 through his or her own computing device.

The calculation conditions may include the learning purpose of the learning model, the type of output, the learning method, the type of learning model, the amount of learning data, the type of learning data, and the learning time.

The learning purpose may refer to a purpose or task to be solved by the learning model.

For example, the learning purpose may be a purpose of identifying a person based on image data, a purpose of identifying lung cancer based on X-RAY data, and the like.

However, it is not limited to this, and the type of learning purpose can be variously modified at a level obvious to those skilled in the art.

The type of output may refer to a specific meaning of the result value.

For example, 'whether human or non-human', 'lung cancer or non-lung cancer', 'predetermined treatment method', etc. may be indicated.

However, it is not limited to this, and the type of output can be variously modified at a level obvious to those skilled in the art.

The learning method may refer to a method of machine learning and/or deep learning of a learning model.

As an example, the learning method may be classified as deep learning or machine learning.

In addition, learning methods can be classified into supervised learning, unsupervised learning, and reinforcement learning.

The learning model type may mean a specific algorithm type used for learning.

For example, the type of learning model may be classified into a k-nearest neighbor algorithm (K-NN), a support vector machine algorithm (SVM), a decision tree algorithm, a random forest algorithm, and the like.

For example, learning model types may be classified into ANN, CNN, DNN, GAN, and RNN.

However, it is not limited thereto, and the type of learning model can be variously modified at a level obvious to those skilled in the art.

The amount of training data may mean the amount of data required to generate a learning model.

For example, the amount of training data may be displayed as a number and unit representing the amount of data such as GB or TB.

The type of training data may mean that it is classified into an image form, a text form, or a video form.

The learning time may mean the time required for machine learning/deep learning.

However, it is not limited thereto, and the calculation conditions can be variously modified at a level obvious to those skilled in the art.

The receiving module 110 may transfer the received information to other modules including the similarity determining module 130 .

The receiving module 110 may receive information necessary to operate the learning model recommendation method and the learning model mediation platform.

The first storage module 120 may store all information necessary to implement the learning model recommendation method.

In the first storage module 120, a stored learning model, which is a learning model learned in the past, and a calculation condition of the stored learning model may be matched and stored.

The first storage module may include a built-in memory and/or an external memory.

By way of example, built-in memory may include volatile memory (eg, DRAM, SRAM, or SDRAM, etc.), non-volatile memory (eg, one time programmable ROM (OTPROM), PROM, EPROM, EEPROM, mask ROM, flash ROM, flash memory, hard drive memory, etc.) drive, or at least one of a solid state drive (SSD).

External memory is a flash drive, such as a compact flash (CF), secure digital (SD), Micro-SD, Mini-SD, extreme digital (xD), multi-media card (MMC), or memory stick. etc. may be included.

The similarity determination module 130 may be a model that determines the degree of similarity between the target learning model and the stored learning model.

The degree of similarity can be calculated by comparing the calculation conditions of the target learning model and the calculation conditions of the stored learning model.

The similarity determination module 130 may vectorize calculation conditions of the stored learning model by pre-processing them. The vectorization method utilizes a known technique, and detailed description thereof may be omitted.

Similarly, the similarity determination module 130 may preprocess and vectorize the calculation conditions of the target learning model.

The similarity determination module 130 may compare the calculation condition of the target learning model with the calculation condition of the target learning model to calculate the degree of similarity.

For example, the similarity determination module 130 may calculate the degree of similarity based on the distance between the calculation condition vector information of the target learning model and the calculation condition vector information of the stored learning model.

However, it is not limited thereto, and the similarity level determination method of the similarity determination module 130 can be variously modified at a level obvious to those skilled in the art.

The recommendation module 140 may select a recommended learning model by determining whether a predetermined recommendation condition, a predetermined first additional condition, and a predetermined second additional condition are satisfied.

The pre-determined recommendation condition may be a condition similar to the target learning model with a first similarity or higher.

When the distance to the target learning model is less than or equal to the first distance, it may be determined that the model is more similar than the first similarity.

A first additional pre-determined condition is a cluster of the stored learning models that is similar to the target learning model to a second degree of similarity or more, is similar to the target learning model to a degree of less than the first similarity, and exists to a degree of less than the second degree of similarity based on the target learning model. It may be a condition that exists in less than a predetermined number.

When the distance to the target learning model is greater than the first distance and less than or equal to the second distance, the similarity may be determined to be equal to or greater than the second similarity and less than the first similarity.

Here, the second distance may mean a longer distance than the first distance.

Here, the predetermined number of clusters may be three.

However, it is not limited thereto, and the predetermined number of clusters can be variously modified at a level obvious to those skilled in the art.

The predetermined second additional condition may be a condition that, when the stored learning models that are similar to the target learning model by less than the second similarity are connected, the similarity with the target learning model is equal to or greater than the first similarity.

When the distance to the target learning model exceeds the second distance, it may be judged to be similar to less than the second similarity.

The stored learning model itself is similar to the target learning model in terms of the second similarity, but when several stored learning models are combined, there may be cases where the target learning model and the first similarity or more are similar.

As an example, it can be assumed that the target learning model is a model that detects whether or not an intruder is based on image data and calculates a countermeasure according to the type of intrusion. The first stored learning model may be a learning model for determining whether or not an intruder is based on image data, the second stored learning model may be a learning model for determining the type of intrusion based on image data, and the third stored learning model may be a learning model for determining an intrusion type based on image data. The model may be a model that calculates countermeasures based on the intrusion type and form. Each of the first to third stored learning models is similar to the target learning model by less than the second similarity, but when the first to third stored learning models are combined, the target learning model is similar to the target learning model by more than the first similarity. can do.

Through this, it is possible to recommend a learning model that meets the needs of the consumer by maximally utilizing the stored learning module stored in the storage module 120 .

The simulation module 150 may generate a virtual learning model by connecting the storage learning modules stored in the storage module 120 to each other.

When the above-mentioned predetermined recommendation condition and the predetermined first additional condition are not satisfied, the simulation module 150 receives the calculation condition of the target learning model from the receiving module 110, and the stored learning model from the storage module 120. All information about can be delivered.

The simulation module 150 may combine stored learning modules based on data input to the target learning model and data calculated from the target learning model.

As a specific example, the simulation module 150 may select a stored learning model (start stored learning model) in which input data input to a target learning model and input data of an arbitrary stored learning model match.

In addition, the simulation module 150 may select a stored learning model (final stored learning model) in which output data output from the target learning model and output data of an arbitrary stored learning model match.

Here, the meaning of matching may mean that the type of input data and the purpose for which the data is analyzed are at least partially identical.

In addition, the simulation module 150 is a stored-learning model (intermediate storage) in which data matching the output data of the starting stored-learning model is used as input data and data matching the input data of the final stored-learning model is used as output data of the stored-learning model. learning model) can be selected.

Each stored learning model may not exist, may be one, or may be plural. When there are multiple intermediate stored learning models, the output data of the preceding intermediate stored learning model and the input data of the succeeding intermediate stored learning model can be aligned so that they match each other, and the input data of the most preceding intermediate stored learning model starts storage. The simulation module 150 may select each stored learning model so that it matches the output data of the learning model, and the output data of the intermediate stored learning model that follows most closely matches the input data of the final stored learning model.

The simulation module 150 may generate a simulation model through machine learning/deep learning that calculates a state of a final learning model when a plurality of learning models are input.

Machine learning may be performed using a back propagation algorithm, which is an algorithm for updating weights of a neural network using labeled data of an output layer, but is not limited thereto.

In addition, since a deep neural network and a back propagation algorithm are conventionally known, a detailed description thereof may be omitted.

A virtual learning model may be generated by inputting the initial stored learning model, the final stored learning model, and/or the intermediate stored learning model selected by the simulation module 150 into the simulation model.

In addition, the simulation module 150 may calculate the calculation conditions of the virtual learning model based on the stored learning models input to the simulation model.

For example, the learning time may be calculated as the sum of learning times of stored learning models input to the simulation model.

For example, the amount of learning data may be calculated as the sum of stored learning models input to the simulation model.

The simulation module 150 may not be able to calculate calculation conditions for virtual learning models that cannot be estimated as stored learning models.

The similarity determination module 130 may determine the similarity based only on the calculation conditions calculated when determining the similarity between the virtual learning model and the target learning model.

When a predetermined learning condition is satisfied, the learning suggestion module 160 may propose to the manager to create a learning model.

The predetermined learning condition may be a condition in which the predetermined recommendation condition is not satisfied.

As a specific example, the predetermined learning condition may be a condition in which a predetermined number or more of similar target learning models are present in which the predetermined recommendation condition, the predetermined first additional condition, and the predetermined second additional condition are not satisfied.

The learning suggestion module 160 is a target learning model (hereinafter referred to as a rejection learning model) in which the predetermined recommendation condition, the predetermined first additional condition, and the predetermined second additional condition from the recommendation module 140 are not satisfied. ) can be delivered and stored.

The learning suggestion module 160 may vectorize and cluster the calculation conditions of the rejection learning model.

Since a method of clustering (grouping) data uses a known technique, a detailed description thereof may be omitted.

If one clustered group consists of a predetermined number or more rejection learning models, the learning restriction module may propose to the manager to create a learning model based on the calculation condition calculated by rebuilding the vector value of the average of the group.

To this end, the interface module 180 may display an interface in which a calculation condition for proposing learning is displayed to the administrator and an inquiry is made on whether or not to accept it.

The price calculation module 170 may calculate a cost according to using, selling, or licensing the stored learning model in a cloud environment.

The price calculation module 170 may set a cost according to the difficulty of manufacturing the stored learning model.

For example, the price calculation module 170 may set a higher cost as the manufacturing difficulty increases. Conversely, the price calculation module 170 may set a lower cost as the manufacturing difficulty decreases.

The price calculation module 170 may calculate the cost to be charged based on the degree of similarity between the target learning model and the recommended learning model to be recommended.

Here, the degree of similarity may be calculated based on the distance between the two models.

Due to this, the consumer can be compensated for the disadvantages caused by not being able to use the desired learning model.

The interface module 180 may calculate an interface capable of displaying or collecting necessary information for a manager or consumer while the learning model recommendation method is implemented, and provide the calculated interface to the manager or consumer's computing device.

The interface module 180 may calculate an interface for a learning model mediation platform capable of buying and selling a learning model, and provide the interface to a buyer or seller.

A detailed description of the learning model mediation platform will be described later.

The interface module 180 may transmit information required to operate the learning model recommendation method and the learning model mediation platform.

The interface module 180 may transmit to the seller of the learning model a calculation condition that has been requested to be learned by the manager but has been rejected.

The seller list of the learning model may consist of sellers using the learning mediation platform.

Through this, it is possible to activate the learning model platform and the learning model sales market, and to meet the needs of consumers.

The learning model recommendation device not only recommends a learning model required inside the system to other devices inside the system, but also can recommend an appropriate learning model to a cloud service user or individual Internet user.

The management device 200 according to an embodiment of the present invention includes an operating module 210 for operating a virtual machine of a physical server, a repair module 220 for correcting an error when the virtual machine operates abnormally, and a virtual machine on the physical server. A migration module 230 for calculating and migrating an optimal arrangement of machines and a mediation module 240 for mediating between the learning model recommendation device and the monitoring device 300 may be provided.

The operating module 210 may perform operations such as creating and deleting a virtual machine according to a user's request.

The repair module 220 may repair the virtual machine when an error occurs in the virtual machine according to a predetermined repair method.

Corresponding methods according to errors may be databased and stored in the repair module 220 . The repair module 220 may receive log and metadata of a virtual machine generating an error from the monitoring device 300 and correct the error of the virtual machine according to a pre-stored corresponding method.

However, it is not limited thereto, and the repair method of the repair module 220 can be variously modified at a level obvious to those skilled in the art.

For example, the repair module 220 may request the model operation module 330 to calculate an appropriate repair method according to an error. Accordingly, one monitoring learning model stored in the model operation module 330 is operated, and an appropriate repair method according to the error can be calculated, and the calculated repair method is transmitted from the model operation module 330 to the repair module 220. It can be.

The migration module 230 may migrate the virtual common people in order to place the virtual machine on the physical machine.

As an example, the migration module 230 may request the model operation module 330 to predict the workload of the virtual machine, and one monitoring learning model stored in the model operation module 330 is operated and determined to be generated in the cloud service. A predicted workload may be calculated. The workload of the virtual machine predicted by the model operation module 330 may be transferred to the migration module 230, and the migration module 230 is a virtual machine in which a predetermined objective function is minimized based on the predicted workloads of the virtual machine. An optimized placement of machines can be calculated, and virtual machines can be placed on physical servers accordingly.

Accordingly, a detailed description thereof may be omitted within the known technical scope.

The mediation module 240 may mediate between the monitoring device 300 and the learning model recommendation device so that the monitoring device 300 can appropriately recommend a required learning model from the learning model recommendation device.

The mediation module 240 may receive information necessary for the operation module 210 , the repair module 220 , and the migration module 230 from the monitoring device 300 in association with the monitoring device 300 .

However, it is not limited thereto, and the monitoring device 300 may directly communicate with the learning model recommendation device and exchange information necessary for recommending a learning model.

In the monitoring device 300 according to an embodiment of the present invention, in the monitoring device 300 for monitoring the operation of at least one virtual machine running in the physical server within the allocated resources of the cloud server integrated management system, A transmission/reception module 310 for collecting operation information generated from the virtual machine, a monitoring module 320 for monitoring whether the virtual machine operates abnormally based on the operation information collected by the transmission/reception module 310, Based on the request of the monitoring module 320, the model operation module 330 that activates the monitoring learning model that helps the monitoring function of the monitoring module 320 and the monitoring learning model can be operated within the allocated resources. , may include a schedule module 340 that determines the operation timing of the monitoring learning model through a predetermined timing determination method.

In addition, the cloud server integrated management system may further include a model selection module 350 for selecting the monitoring learning model satisfying a predetermined operation condition.

In addition, the cloud server integrated management system may further include a second storage module 360 for storing all information required to implement the monitoring method.

The transmission/reception module 310 may collect operation information generated in the virtual machine.

In addition, the transmission/reception module 310 may transmit/receive information necessary to implement the monitoring method with internal components of an external server, a client, and a cloud server integrated management system.

As an example, operation information may include both metric data and log data generated in a virtual machine.

The monitoring module 320 may monitor in real time whether or not the virtual machine operates abnormally based on the operation information.

To this end, the monitoring module 320 may store error patterns of operation information that may be determined to be abnormal, and when a similar error pattern is detected, it may be determined that an error has occurred in the corresponding virtual machine.

The monitoring module 320 may transmit information about a virtual machine in which an error occurs to the management device 200 .

The monitoring module 320 may calculate an interface capable of displaying information on load generated in the virtual machine, a load state of the physical server, a list of operating physical servers, and the like to the cloud server integrated management system manager.

In addition, the monitoring module 320 may calculate an interface through which the user can check the usage state of the virtual machine, the incurred cloud usage cost, and the like.

Here, the interface may be expressed through a visual image such as a graph.

The monitoring function in the present invention calculates all information generated in the cloud service, such as detection of abnormal symptoms of virtual machines, prediction of virtual machine load, load prediction of cloud integrated operating system, amount of power generated from physical servers, detection of abnormal symptoms of physical machines, It may include functions for detection, monitoring, and prediction.

The model operating module 330 may operate a monitoring learning model that is a learning model that helps the management device 200 operate or helps the monitoring function of the monitoring module 320 .

For example, the monitoring learning model may include a first monitoring learning model that predicts a load generated in the cloud server integrated management system for a predetermined future period.

For example, the monitoring learning model may include a second monitoring learning model that predicts the workload generated in the virtual machine on the physical server for a predetermined future period.

In one example, the predetermined future period may be 30 days.

However, it is not limited thereto, and the predetermined future period can be variously modified at a level obvious to those skilled in the art.

As an example, the monitoring learning model may include a third monitoring learning model for determining whether or not there is an error in the virtual machine through operation information and calculating a solution to solve the error.

A monitoring learning model is stored in the model operating module 330 .

The model selection module 350 may select the monitoring learning model that satisfies a predetermined operating condition, and transmit the selected monitoring learning model to the model operating module 330 .

The predetermined operating condition may be a condition in which a load lower than a predetermined threshold value is applied.

The model selection module 350 may change the predetermined threshold based on the load predicted by the first monitoring learning model.

The model selection module 350 may receive information about used resources from among all resources from the integrated cloud operating system for a predetermined period stored in the storage module to calculate a predetermined threshold.

Unlike this, the model selection module 350 may receive information about spare resources of the cloud integrated operating system for a predetermined future period through the first monitoring learning model of the model operation module 330 .

Here, the spare resource may mean a value obtained by subtracting used resources from all resources.

For example, the predetermined threshold may mean 70% of an average spare resource for a future period or a predetermined period.

However, it is not limited thereto, and the exact value of the predetermined threshold value can be variously modified at a level obvious to those skilled in the art.

The model selection module 350 may request a learning model recommendation device to recommend a monitoring learning model that satisfies the predetermined operation condition.

When the model selection module 350 requests a monitoring learning model necessary for the learning model recommendation device, it may transmit calculation conditions of the monitoring learning model together.

The model selection module 350 receives a monitoring learning model (recommended learning model) from the learning model recommendation device and transfers it to the model operating module 330, and the model operating module 330 may store the monitoring learning model.

The schedule module 340 may determine the operation time based on the load predicted by the first monitoring learning model.

Here, the operation time may be a term including a start time as well as an ongoing time.

As an example, the predetermined timing determination method may be a method of determining the operation timing at a time when a load is generated at a predetermined ratio or less based on resources allocated to the cloud integrated operating system.

As a specific example, the schedule module 340 may divide a predetermined future period into predetermined division periods.

Here, the division period may be a unit of time, a unit of a day, or a unit of a week.

Here, the schedule module 340 may calculate average used resources predicted to be generated in the integrated cloud operating system for each partition period.

The schedule module 340 may determine, as an operation period, a period when used resources calculated for each partition period is less than or equal to a predetermined ratio based on all resources.

Here, the predetermined ratio (first ratio) may be 10%.

However, it is not limited thereto, and the predetermined ratio can be variously modified at a level obvious to those skilled in the art.

For example, the predetermined time determination method may be a method of determining the operation timing at a time when the least load is generated among the predetermined future periods.

As a specific example, the schedule module 340 may divide a predetermined future period into predetermined division periods.

Here, the division period may be in units of minutes, units of hours, units of days, or units of weeks.

Here, the schedule module 340 may calculate average used resources generated in the cloud integrated operating system for each partition period.

The schedule module 340 may determine, as an operation time, a partition period having the lowest average used resources calculated for each partition period.

The schedule module 340 selects and determines the operation time when the load of the cloud integrated operating system is low, thereby effectively solving the problem that the operation rate of the cloud integrated operating system is hindered by the operation of the operation learning model.

The second storage module 360 may store all information necessary for implementing the monitoring method.

The second storage module 360 may also store overall resource information and past used resource information of the cloud integrated operating system in the past.

The second storage module may include a built-in memory and/or an external memory.

By way of example, built-in memory may include volatile memory (eg, DRAM, SRAM, or SDRAM, etc.), non-volatile memory (eg, one time programmable ROM (OTPROM), PROM, EPROM, EEPROM, mask ROM, flash ROM, flash memory, hard drive memory, etc.) drive, or at least one of a solid state drive (SSD).

External memory is a flash drive, such as a compact flash (CF), secure digital (SD), Micro-SD, Mini-SD, extreme digital (xD), multi-media card (MMC), or memory stick. etc. may be included.

A storage module included in the learning model recommendation device may be defined as the first storage module 120 , and a storage module included in the monitoring device 300 may be defined as the second storage module 360 .

Hereinafter, the learning model recommendation method and monitoring method implemented by each module will be described in detail.

4 is a flowchart of a learning model recommendation method implemented by an apparatus for recommending a learning model of a cloud integrated operating system according to an embodiment of the present invention.

Hereinafter, a detailed description may be omitted to the extent that it overlaps with the above description.

Referring to FIG. 4 , a learning model recommendation method according to an embodiment of the present invention is implemented by a learning model recommendation device and recommends a learning model calculated by machine learning or deep learning according to a request of a consumer. In the recommendation method, a request receiving step in which the request of the consumer and the calculation conditions of the target learning model, which is the learning model requested by the consumer, are received by the receiving module, and the stored learning model, which is a learning model created in the past, by the storage module The storage stored in the storage module is performed by a condition storage step in which calculation conditions of the stored learning model are stored, a similarity determination step in which the similarity between the target learning model and the stored learning model is determined by a similarity determination module, and a recommendation module. It may include a recommendation step in which a learning model satisfying a predetermined recommendation condition among learning models is recommended to a consumer.

Here, even if the predetermined recommendation condition is not satisfied, the recommendation step combines a predetermined first additional condition for determining whether the condition is satisfied based on a second similarity lower than the first similarity and a plurality of the stored learning models. It may be a step of selecting or calculating a learning model to be recommended to the consumer by determining whether a predetermined second additional condition for determining whether the condition is satisfied is satisfied based on the generated virtual learning model.

The consumer may input calculation conditions of the target learning model through the interface received from the interface module using a computing device.

The collection module may receive the calculation conditions of the target learning model from the consumer's computing device and transmit the calculation conditions to the similarity determination module.

The similarity determination module may determine the degree of similarity between the target learning model and the stored learning model stored in the storage module.

The similarity judgment module may deliver the similarity judgment result between the target learning model and the stored learning model to the recommendation module.

5 is a diagram for explaining a predetermined recommendation condition and a predetermined first additional condition in a learning model recommendation method according to an embodiment of the present invention.

Referring to FIGS. 4 and 5(a) , the recommendation module may determine whether a stored learning model satisfying a predetermined recommendation condition exists.

The pre-determined recommendation condition may be a condition similar to the target learning model with a first similarity or higher.

Similar to the first degree of similarity or more may mean that they exist within the first distance X10 from the target learning model A10.

If there are a plurality of stored learning models satisfying a predetermined recommendation condition, the recommendation module may select a stored learning model most similar to the target learning model as the recommended learning model C10.

Even if there is no storage learning model satisfying a predetermined recommendation condition, the recommendation module may calculate the recommendation learning model according to a predetermined calculation method when a predetermined first additional condition is satisfied.

A first additional pre-determined condition is a cluster of the stored learning models that is similar to the target learning model to a second degree of similarity or more, is similar to the target learning model to a degree of less than the first similarity, and exists to a degree of less than the second degree of similarity based on the target learning model. It may be a condition that exists in less than a predetermined number.

This may be for complementing and correcting when the calculation conditions of the target learning model cannot be accurately specified due to the immaturity of the consumer.

Referring to FIGS. 5(b) and 5(c) , a first additional condition determined in advance may be satisfied as there are 1 and 3 clusters of stored learning models.

Here, a cluster may be defined as being formed when the number of stored learning models within a predetermined distance from the center point of the cluster is greater than or equal to a predetermined number.

Here, the predetermined distance and the predetermined number can be variously modified at a level obvious to those skilled in the art.

The predetermined calculation method is a method of selecting the stored learning model closest to the mean of the cluster as the recommended learning model when there is only one cluster of the stored learning models existing at the second similarity or higher based on the target learning model. can be

Similarity greater than or equal to the second degree of similarity may mean that the object learning model is less than a second distance X20 from the target learning model.

Referring to FIG. 5( b ), one cluster may be formed within the first distance X10 and the second distance X20 or less.

The recommendation module calculates the calculation condition by inversely calculating the vector of the average point (M10) of one cluster within the first distance (X10) and the second distance (X20) or less based on the target learning model (A10), and recommending learning can be selected as a model.

The predetermined calculation method is a combination of the stored learning models closest to the average of each cluster when the number of clusters of the stored learning models that exist with the second similarity or higher based on the target learning model exceeds one and is less than or equal to a predetermined number. It may be a method of selecting a recommendation learning model.

Referring to FIG. 5(c), three clusters may be formed within the first distance X10 and the second distance X20 or less based on the target learning model A10.

For example, the average point (M11) of the calculation conditions of the storage learning model constituting the first cluster, the average point (M12) of the calculation condition calculation conditions of the storage learning model constituting the second cluster, and the average calculation condition of the storage learning model constituting the third cluster The calculation condition may be calculated by taking a weighted average according to the distance to the points M13.

Here, the closer the distance to the target learning model is, the higher the weight may be.

For example, the distance between the calculation condition average point (M11) of the storage learning model constituting the first cluster and the target learning model is '10', and the calculation condition average point (M12) of the storage learning model constituting the second cluster and the target learning model Assuming that the distance to the model is '20' and the distance between the calculation condition average point (M13) of the stored learning model constituting the first cluster and the target learning model is '30', the first cluster, the second cluster and the second cluster The weight of the 3 clusters is set as 3:2:1, so that the above calculation conditions can be calculated.

In this way, the recommendation module calculates the calculation condition by inversely calculating the vector with respect to the calculated average value, and can select it as a recommendation learning model.

In FIG. 5 , vectorization is illustrated in two axes, but the present invention is not limited thereto, and calculation conditions are vectorized in three axes, and similarity and whether or not the above conditions are satisfied can be determined.

The recommendation module may select the recommended learning model based on the stored learning models satisfying a second additional condition, even if the predetermined recommendation condition and the first additional condition are not satisfied.

As a specific example, the recommendation module may select a virtual learning model satisfying a predetermined second additional condition as the recommended learning model based on the virtual learning model transmitted from the simulation module.

If there is no stored learning model that satisfies the predetermined recommendation condition, the first predetermined additional condition, and the second predetermined additional condition, the learning suggestion module may store the calculation condition of the target learning model.

In addition, the interface model may calculate an interface for displaying a message that there is no stored learning model similar to the target learning model to the consumer, and transmit the message to the consumer's computing device.

6 is a diagram illustrating a learning model mediation interface screen provided by an interface module of a learning model recommendation device according to an embodiment of the present invention.

Referring to FIG. 6 , the interface module may calculate an interface for a learning model mediation platform that mediates between a seller who creates and sells a learning model and a buyer who purchases the learning model.

For example, the specifications (calculation conditions) of the learning model created by the seller may be listed, and a price may be set and displayed on the interface.

Here, the consumer can select, purchase, rent, and use a learning model required for his or her program, just like shopping on the Internet.

7 is a flowchart illustrating a process of selecting a learning model to be recommended to a monitoring device by a learning model recommendation device according to an embodiment of the present invention, and FIG. 8 is a model selection of a monitoring device according to an embodiment of the present invention. It is a diagram for explaining how the module calculates the first operating time and the second operating time.

Referring to FIGS. 7 and 8 , when the model selection module transmits the calculation condition to the learning model recommendation device, the first operating time and the second operating time may be transmitted together.

The model selection module may calculate the first operation time and the second operation time while determining the operation time.

The first uptime may refer to the total time of consecutive division periods in which the average resource use of the cloud integrated operating system is generated at a first ratio or less.

The second uptime may refer to a sum of the first uptime and the total continuous time adjacent to the first uptime, in which the average used resource of the cloud integrated operating system is generated at a second ratio or less.

Referring to FIG. 8 (a), as an example, when the division period is a unit of time, the period during which the average used resources of the cloud integrated operating system will be calculated below the first ratio (W10) is expected to be operational (T12, T13). , T14), and the first operation time may be 3 hours.

Here, the second operation time may not exist because the average used resource of the operation time and the neighboring partition periods T11 and T15 is higher than the second ratio W20.

Referring to FIG. 8 (b), as an example, when the division period is a unit of time, the period during which the average used resource of the integrated cloud operating system is calculated at or below the first ratio (W10) is the operation period (T12, T13). , T14), and the first operation time may be 3 hours.

In addition, since the average used resource of the partition periods T11 and T15 adjacent to the operation time is lower than the second ratio W20, the second operation time may be 5 hours, which is 2 hours plus the first operation time.

The second uptime may mean a period during which the learning model can be operated within an allowable range, although the cloud integrated operating system is somewhat burdened.

The learning model recommendation device may receive a request for recommending a learning model, a calculation condition, a first operating time, and a second operating time from the monitoring device directly or from an intermediary module.

Among the above, the target learning model can be understood by changing it to a monitoring learning model that is desired to be recommended.

The learning model recommendation device can calculate a storage learning model similar to the monitoring learning model requested for recommendation as described above, and a detailed description thereof may be omitted to the extent that it overlaps with the above description.

However, unlike the above-mentioned recommendation method, when a learning model is recommended internally in the system, it is possible to determine whether the recommended learning model exists by further considering the operation time.

The simulation module may simulate the time required when the stored learning model stored in the first storage module is operated.

Here, the amount of data expected to be input may be received together when the calculation condition is received from the monitoring device.

The simulation module may transmit the expected operation time of the stored learning model to the recommendation module.

The recommendation module may calculate whether there is a stored learning model that satisfies a predetermined recommendation condition, a predetermined first additional condition, or a predetermined second additional condition in the storage learning module for which the first operating time is satisfied. A detailed description thereof may be omitted to the extent that it overlaps with the above description.

Here, the expected operating time of the stored learning module that satisfies the predetermined second additional condition may be calculated by summing up the expected operating time of the stored learning model (eg, the start stored learning model and the final stored learning model) to be utilized.

If there is no stored learning model satisfying the pre-determined recommendation condition, the pre-determined first additional condition, or the pre-determined second additional condition within the stored-learning module for which the first uptime is satisfied, the recommendation module determines the second uptime-time It is possible to calculate whether there is a storage learning model that satisfies a predetermined recommendation condition, a predetermined first additional condition, or a predetermined second additional condition in the stored learning module that is satisfied. A detailed description thereof may be omitted to the extent that it overlaps with the above description.

Here, the expected operating time of the stored learning module that satisfies the predetermined second additional condition may be calculated by summing up the expected operating time of the stored learning model (eg, the start stored learning model and the final stored learning model) to be utilized.

If, within the first uptime and the second uptime, there is no stored learning model that satisfies the predetermined recommendation condition, the first predetermined additional condition, or the predetermined second additional condition, the absence of the recommendation learning model is determined as a model. It can be transmitted to the selection module, and the calculation conditions of the monitoring learning model are stored in the learning suggestion module and can be used for the above-mentioned learning.

The model selection module may select the recommended learning model received from the learning model recommendation device as a monitoring learning model, and may transmit it to the model operation module.

The model selection module may select the recommended learning model received from the learning model recommendation device as a monitoring learning model, and may transmit it to the model operation module.

A monitoring method according to an embodiment of the present invention is a monitoring method for monitoring the operation of at least one virtual machine running in a physical server within an allocated resource of a cloud integrated operating system through a monitoring device, Collecting operation information generated in the virtual machine by a monitoring module, monitoring whether or not the virtual machine operates non-ideally based on the operation information by a monitoring module, by a model operation module, the monitoring module Based on the request of the monitoring module, the monitoring learning model assisting the monitoring function is operated, and the monitoring learning model is operated within the allocated resource by the schedule module, through a predetermined time point determination method. A step of determining when to activate the monitoring learning model may be included.

In addition, the monitoring method includes the operation timing based on the load (used resources) predicted by the first monitoring learning model that predicts the load generated in the cloud integrated operating system for a predetermined future period by the schedule module. It may further include the step of determining.

In addition, the monitoring method may further include selecting, by a model selection module, a monitoring learning model that satisfies a predetermined operating condition.

The monitoring method may further include changing the predetermined threshold based on the load predicted by the first monitoring learning model by a model selection module.

In the accompanying drawings, in order to more clearly express the technical idea of the present invention, components that are not related to or detached from the technical idea of the present invention are briefly expressed or omitted.

In the above, the configuration and characteristics of the present invention have been described based on the embodiments according to the present invention, but the present invention is not limited thereto, and various changes or modifications can be made within the spirit and scope of the present invention. It will be apparent to those skilled in the art, and thus such changes or modifications are intended to fall within the scope of the appended claims.

10: Cloud integrated operating system 100: Learning model recommendation device
200: management device 300: monitoring device

Claims (10)

In the learning model recommendation device for recommending a learning model calculated by machine learning or deep learning at the request of the consumer,
A receiving module for receiving a request of a consumer and calculation conditions of a target learning model, which is a learning model requested by the consumer;
a storage module in which a stored learning model, which is a learning model created in the past, and calculation conditions of the stored learning model are stored;
a similarity determination module for determining a degree of similarity between the target learning model and the stored learning model; and
A recommendation module for recommending a learning model satisfying a predetermined recommendation condition among the stored learning models stored in the storage module as a recommendation learning model to a consumer;
The recommendation module,
Even if the predetermined recommendation condition is not satisfied, when a predetermined first additional condition is satisfied, the recommendation learning model is calculated according to a predetermined calculation method,
The pre-determined recommendation conditions are,
It is a condition similar to the target learning model and more than the first similarity,
The predetermined first additional condition is,
A condition in which a predetermined number or less of clusters of the stored learning model, which are similar to the target learning model to the second similarity or more, are similar to the target learning model by less than the first similarity, and exist at the second similarity or more based on the target learning model person,
Learning model recommendation device.
delete According to claim 1,
The predetermined calculation method,
A method of selecting the stored learning model closest to the average of the cluster as the recommended learning model when there is one cluster of the stored learning models existing at the second similarity or higher based on the target learning model,
Learning model recommendation device.
According to claim 1,
The predetermined calculation method,
When the number of clusters of the stored learning models that exist with the second similarity or higher based on the target learning model exceeds one and is less than or equal to a predetermined number, the recommended learning model is selected as a combination of the stored learning models closest to the average of each cluster. how to do it,
Learning model recommendation device.
According to claim 1,
The recommendation module,
Even if the predetermined recommendation condition and the predetermined first additional condition are not satisfied, the recommended learning model is selected based on the stored learning models for which a predetermined second additional condition is satisfied,
The predetermined second additional condition is,
When the stored learning models that are similar to the target learning model and are less than the second similarity are connected, the similarity with the target learning model is equal to or greater than the first similarity,
Learning model recommendation device.
According to claim 5,
Further comprising a simulation module for simulating the state of the final learning model when connecting arbitrary learning models to each other,
The simulation module,
When a plurality of learning models are input, a simulation model that calculates the state of the final learning model is created through deep learning,
Learning model recommendation device.
In the learning model recommendation device for recommending a learning model calculated by machine learning or deep learning at the request of the consumer,
A receiving module for receiving a request of a consumer and calculation conditions of a target learning model, which is a learning model requested by the consumer;
a storage module in which a stored learning model, which is a learning model created in the past, and calculation conditions of the stored learning model are stored;
a similarity determination module for determining a degree of similarity between the target learning model and the stored learning model;
a recommendation module for recommending a learning model satisfying a predetermined recommendation condition among the stored learning models stored in the storage module as a recommendation learning model to a consumer; and
When a predetermined learning condition is satisfied, a learning suggestion module that proposes to the manager to create a learning model; further comprising,
The pre-determined recommendation conditions are,
It is a condition similar to the target learning model and more than the first similarity,
The predetermined learning conditions are,
A condition in which the predetermined recommendation condition is not satisfied,
Learning model recommendation device.
In the learning model recommendation device for recommending a learning model calculated by machine learning or deep learning at the request of the consumer,
A receiving module for receiving a request of a consumer and calculation conditions of a target learning model, which is a learning model requested by the consumer;
a storage module in which a stored learning model, which is a learning model created in the past, and calculation conditions of the stored learning model are stored;
a similarity determination module for determining a degree of similarity between the target learning model and the stored learning model;
a recommendation module for recommending a learning model satisfying a predetermined recommendation condition among the stored learning models stored in the storage module as a recommendation learning model to a consumer; and
Further comprising a price calculation module for calculating a cost to be charged while recommending the stored learning model;
The pre-determined recommendation conditions are,
It is a condition similar to the target learning model and more than the first similarity,
The price calculation module,
Calculating the cost to be charged based on the degree of similarity between the target learning model and the recommended learning model to be recommended,
Learning model recommendation device.
In the learning model recommendation method implemented by the learning model recommendation device and recommending a learning model calculated by machine learning or deep learning according to the request of the consumer,
a request receiving step in which a request of a consumer and a calculation condition of a target learning model, which is a learning model requested by the consumer, are received by the receiving module;
a condition storage step of storing a stored learning model, which is a learning model generated in the past, and a calculation condition of the stored learning model by a storage module;
a similarity determination step of determining a similarity between the target learning model and the stored learning model by a similarity determination module;
a recommendation step of recommending, by a recommendation module, a learning model satisfying a predetermined recommendation condition among the stored learning models stored in the storage module to a consumer; and
When a predetermined learning condition is satisfied by the learning suggestion module, proposing to the manager to create a learning model; includes,
The predetermined learning condition is,
A condition in which the predetermined recommendation condition is not satisfied,
Learning model recommendation device.
In the learning model recommendation method implemented by the learning model recommendation device and recommending a learning model calculated by machine learning or deep learning according to the request of the consumer,
a request receiving step in which a request of a consumer and a calculation condition of a target learning model, which is a learning model requested by the consumer, are received by the receiving module;
a condition storage step of storing a stored learning model, which is a learning model generated in the past, and a calculation condition of the stored learning model by a storage module;
a similarity determination step of determining a similarity between the target learning model and the stored learning model by a similarity determination module;
a recommendation step of recommending, by a recommendation module, a recommended learning model, which is a learning model that satisfies a predetermined recommendation condition, among the stored learning models stored in the storage module, to a consumer; and
Further comprising, by a price calculation module, calculating a cost to be charged based on a degree of similarity between the target learning model and the recommended learning model while recommending the stored learning model,
The pre-determined recommendation conditions are,
A condition similar to the target learning model and the first similarity or higher,
Learning model recommendation method.

Images