WO2019062411A1

WO2019062411A1 - Method for managing and controlling background application program, storage medium, and electronic device

Info

Publication number: WO2019062411A1
Application number: PCT/CN2018/102205
Authority: WO
Inventors: 曾元清
Original assignee: Oppo广东移动通信有限公司
Priority date: 2017-09-30
Filing date: 2018-08-24
Publication date: 2019-04-04
Also published as: CN107729078B; CN107729078A

Abstract

Disclosed are a method for managing and controlling a background application program, a storage medium, and an electronic device. The method comprises: inputting a first sample image and a second sample image of a pre-set background application program and an electronic device into a reference model for learning, so as to obtain optimized parameters; and acquiring a first feature image and a second feature image of the pre-set background application program and the electronic device, and generating a prediction result according to the reference model, the optimized parameter, the first feature image and the second feature image, so as to manage and control the pre-set background application program.

Description

Background application management method, storage medium and electronic device

This application claims the priority of the Chinese patent application filed on September 30, 2017, the Chinese Patent Office, the application number is 201710917794.1, and the application name is “background application control method, device, storage medium and electronic device”. The citations are incorporated herein by reference.

Technical field

The present application belongs to the field of communications technologies, and in particular, to a background application control method, a storage medium, and an electronic device.

Background technique

With the development of electronic technology, people often install many applications on electronic devices. When a user opens multiple applications in an electronic device, multiple applications run in the background of the electronic device. These background-running applications consume the memory of the electronic device and increase power consumption.

Summary of the invention

The application provides a background application control method, a storage medium and an electronic device, which can improve the intelligence and accuracy of the application control.

In a first aspect, the embodiment of the present application provides a background application control method, which is applied to an electronic device, and the method includes:

Obtaining a first sample set of the preset background application, acquiring a second sample set of the electronic device, where the samples in the first sample set and the second sample set respectively include the preset background application and the Denoting a plurality of characteristic information of the electronic device;

Constructing the first sample set and the second sample set into a two-dimensional first sample map and a second sample map, respectively;

Obtaining a reference model, and inputting the first sample map and the second sample map as training data into the reference model, performing learning, and obtaining optimized parameters of the reference model after training;

Acquiring the preset background application and the current plurality of feature information of the electronic device, and forming a first feature map and a second feature map, according to the reference model, the optimization parameter, the first feature map, and The second feature map generates a prediction result, and controls the preset background application according to the prediction result.

In a second aspect, an embodiment of the present application provides a storage medium on which a computer program is stored, and when the computer program runs on a computer, causes the computer to execute the background application control method described above.

In a third aspect, an embodiment of the present application provides an electronic device, including a processor and a memory, where the memory has a computer program, and the processor invokes the computer program, wherein the processor further performs:

DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings used in the description of the embodiments will be briefly described below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to the drawings without any creative work.

1 is a schematic system diagram of a background application control device according to an embodiment of the present application;

FIG. 2 is a schematic diagram of an application scenario of a background application control device according to an embodiment of the present disclosure;

FIG. 3 is a schematic flowchart of a background application control method according to an embodiment of the present application;

4 is a schematic diagram of a sample diagram provided by an embodiment of the present application;

FIG. 5 is another schematic flowchart of a background application control method according to an embodiment of the present application;

FIG. 6 is a schematic diagram of another application scenario of a background application control device according to an embodiment of the present disclosure;

FIG. 7 is a schematic structural diagram of a background application control device according to an embodiment of the present application;

FIG. 8 is a schematic structural diagram of a part of a reference model according to an embodiment of the present disclosure;

FIG. 9 is another schematic structural diagram of a background application control device according to an embodiment of the present application;

FIG. 10 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure;

FIG. 11 is another schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed ways

Referring to the drawings, wherein like reference numerals represent the same components, the principles of the present application are illustrated by the implementation in a suitable computing environment. The following description is based on the specific embodiments of the present invention as illustrated, and should not be construed as limiting the specific embodiments that are not described herein.

The embodiment of the present application provides a background application control method, which is applied to an electronic device, and the method includes:

In the background application management method, the reference model includes two sub-reference models, and the two sub-reference models are convolutional neural network models.

In the background application management method, the reference model further includes a classifier;

And inputting the first sample map and the second sample map as the training data into the reference model, and learning, and obtaining the optimized parameters of the reference model after training includes:

And inputting the first sample map and the second sample map as training data into two of the sub-reference models respectively;

Combining output values of two of the sub-reference models into the classifier, and obtaining a probability corresponding to the plurality of prediction results;

Obtaining a loss value according to the plurality of prediction results and a plurality of the probabilities corresponding thereto;

Training is performed according to the loss value to obtain the optimization parameter.

In the background application management method, the training according to the loss value specifically includes:

Training is performed using a stochastic gradient descent method based on the loss value.

In the background application management method, the synthesizing the output values of the two sub-reference models into the classifier specifically includes:

The output values of the two sub-reference models are combined into the classifier by different weights.

In the background application management method, the synthesizing the output values of the two sub-reference models into the classifier and obtaining the probability corresponding to the plurality of prediction results specifically includes:

Generating an output value of two of the sub-reference models into the classifier based on a first preset formula, and obtaining a probability corresponding to the plurality of prediction results, wherein the first preset formula is:

Where Z _K is a composite value of the output values of the two sub-reference models, C is the number of categories of the prediction result, and Z _j is the j-th composite value.

In the background application management method, the obtaining the loss value according to the plurality of the prediction results and the plurality of the probabilities corresponding thereto specifically includes:

And obtaining, according to the second preset formula, a loss value according to the plurality of the prediction results and a plurality of the probabilities corresponding thereto, wherein the second preset formula is:

Where C is the number of categories of predictions and y _k is the true value.

A plurality of the loss values are obtained, and training is performed based on an average of the plurality of the loss values.

And obtaining, according to a third preset formula, a loss value according to the plurality of the prediction results and a plurality of the probabilities corresponding thereto, wherein the third preset formula is:

Where C is the number of categories of prediction results, y _k is the true value, and E is the average.

In the background application management method, the acquiring the preset background application and the current plurality of feature information of the electronic device specifically includes:

Get the preset time;

If the current system time reaches the preset time, the preset background application and the plurality of feature information of the electronic device are obtained.

Detecting whether the default background application enters the background;

If the background is entered, the preset background application and the plurality of feature information of the electronic device are obtained.

In the related art, when the application in the background is controlled, the application in the background is usually cleaned according to the memory usage of the electronic device and the priority of each application to release the memory. However, some applications are important to the user, or the user needs to use some applications again in a short period of time. If these applications are cleaned up after cleaning, users need to reload the electronic devices when they use them again. These application processes take a lot of time and memory resources. The electronic device may be a smart phone, a tablet computer, a desktop computer, a notebook computer, or a handheld computer.

Please refer to FIG. 1. FIG. 1 is a schematic diagram of a system of a background application control device according to an embodiment of the present application. The background application management device is mainly configured to: obtain a first sample set of the preset background application, and acquire a second sample set of the electronic device, where the samples in the first sample set and the second sample set respectively include a preset background a plurality of feature information of the application and the electronic device; respectively constructing the first sample set and the second sample set into a two-dimensional first sample map and a second sample map; acquiring the reference model, and the first sample map And the second sample map is used as the training data input reference model to learn, obtain the optimized parameters of the trained reference model; obtain the preset background application and the current plurality of feature information of the electronic device, and form the first feature map and the second The feature map generates a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map, and then determines, according to the prediction result, whether the preset background application needs to be used, to the preset background application. The program is controlled, such as shutting down, or freezing.

Specifically, please refer to FIG. 2 , which is a schematic diagram of an application scenario of a background application control device according to an embodiment of the present application. For example, when receiving the control request, the background application management device detects that the application running in the background of the electronic device includes a preset background application a, a preset background application b, and a preset background application c; Presetting the reference model A corresponding to the background application a, the reference model B corresponding to the preset background application b, and the reference model C corresponding to the preset background application c; whether the preset background application a needs to be The probability of use is predicted to obtain the probability a', and the probability of whether the preset background application b needs to be used is predicted by the reference model B, and the probability b' is obtained, and the reference model C needs to be used for the preset background application c. The probability is predicted to obtain the probability c′; according to the probability a′, the probability b′ and the probability c′, the preset background application a running in the background, the preset background application b and the preset background application c are controlled, for example The default background application b with the lowest probability is turned off.

The embodiment of the present application provides a background application control method, and the execution entity of the background application control method may be a background application control device provided by the embodiment of the present application, or an electronic device integrated with the background application control device, wherein The background application management device can be implemented in hardware or software.

The embodiment of the present application will be described from the perspective of a background application control device, and the background application control device may be specifically integrated in an electronic device. The background application management method includes: acquiring a first sample set of the preset background application, and acquiring a second sample set of the electronic device, wherein the samples in the first sample set and the second sample set respectively include a preset background application And a plurality of feature information of the electronic device; respectively constructing the first sample set and the second sample set into a two-dimensional first sample map and a second sample map; acquiring the reference model, and the first sample map and the first sample The two sample map is used as the training data input reference model to learn, obtain the optimized parameters of the trained reference model, obtain the preset background application and the current plurality of feature information of the electronic device, and form the first feature map and the second feature map. And generating a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map, and controlling the preset background application according to the prediction result.

Referring to FIG. 3, FIG. 3 is a schematic flowchart of a background application control method according to an embodiment of the present application. The background application management and control method provided by the embodiment of the present application is applied to an electronic device, and the specific process may be as follows:

101. Acquire a first sample set of the preset background application, and acquire a second sample set of the electronic device, where the samples in the first sample set and the second sample set respectively include multiple preset background applications and multiple electronic devices. Feature information.

Specifically, the sample of the first sample set may include usage information of the preset application, and the sample of the second sample set may include at least one of status information, time information, and location information of the electronic device.

The usage information of the application may include, for example, usage time, background time, application type, application association information, and the like. The status information of the electronic device may include, for example, screen brightness, state of charge, remaining power, WIFI status, and the like. The time information may include, for example, a current time point, a work day, and the like. The location information may include, for example, GPS positioning, base station positioning, WIFI positioning, and the like.

A plurality of characteristic information is collected as a sample, and then a first sample set of the preset application and a second sample set of the electronic device are formed.

The preset application may be any application installed in the electronic device, such as a communication application, a multimedia application, a game application, a news application, or a shopping application.

The sample set may include: a plurality of samples collected according to a preset frequency during a historical time period. The historical time period can be, for example, within the past 15 days, within 7 days, and the like. The preset frequency can be, for example, every 10 minutes, 30 minutes, and the like.

The terms "first", "second", and the like in this application are used to distinguish different objects, and are not intended to describe a particular order.

102. The first sample set and the second sample set are respectively constructed into a two-dimensional first sample map and a second sample map.

The samples in the first sample set and the second sample set are represented by numerical values, such as the state of charge, which can be represented by 0 or 1 as being uncharged and being charged. For example, if the remaining power can be used to indicate the remaining power with 00-100, or divide the power into 5 levels, use 0-5 to indicate the remaining power of different levels. The first sample map and the second sample map may take a map such as 12×12 pixel points, and each pixel point corresponds to one sample, that is, one feature information. Of course, the sample map can adjust the pixels it includes as needed, such as 10×10, 16×16, 12×16, and the like. The larger the amount of data, the more accurate the subsequent predictions. It should be noted that the pixel point specific performance may be 1 or (0, 1). The acquired plurality of feature information is stored in a two-dimensional mathematical image, similar to a grayscale image, that is, different feature values are recorded at pixel points (x, y).

Please refer to FIG. 4. FIG. 4 is a schematic diagram of a second sample diagram provided by an embodiment of the present application. According to the type of the feature information of the electronic device, the information may be divided into several types of feature information, for example, the feature information of the electronic device is divided into four types: state information, time information, network information, and location information of the electronic device, and then the four types of feature information are respectively formed into one sub-category. Sample map 3061, then four subsample map 3061 matrices are arranged to form a large second sample map 306. The subsample image 3061 can adopt, for example, 6×6 pixels. If the feature information is insufficient to fill the subsample graph, the position is zero-padded. The four subsample views 3061 form a 12 x 12 large second sample map 306. Both the second sample map 306 and the subsample map 3061 are two-dimensional maps.

It should be noted that the first sample image of the background application may also be set by a method similar to the second sample image.

103. Acquire a reference model, and input the first sample map and the second sample map as training data into the reference model, and learn to obtain optimized parameters of the trained reference model.

The reference model includes two sub-reference models, and the two sub-reference models are convolutional neural network models. Of course, the sub-reference model can be a hybrid neural network model, a Gaussian mixture model, or the like. The two sub-reference models are weak learners, and the two weak learners combine to form a strong learner. There are certain rules in the application of electronic devices corresponding to the user's usage habits and preferences. Learning these rules is of great significance for cleaning up applications that are not used in the background. Integrated learning accomplishes learning tasks by building and combining multiple weak learners. Applying integrated learning to the classification of user behavior characteristics, by combining two independent convolutional neural networks, a strong learner can be built to dig into the user's usage habits. As users use the device for a longer period of time, the training will become more and more complete, and the system prediction will become more accurate.

The sub-reference model includes a convolutional layer and a fully connected layer that are sequentially connected, and the reference model also includes a classifier. Specifically, the reference model mainly includes a network structure part and a network training part, wherein the network structure part includes a convolution layer and a full connection layer connected in sequence. A pooling layer may also be included between the convolutional layer and the fully connected layer.

Optionally, the network structure part of the convolutional neural network model reference model may include a seven-layer network, the first five layers are convolution layers, the convolution kernel size is unified to 3×3, and the sliding step length is unified to 1, due to a small dimension. The pooling layer may not be used, and the latter two layers are fully connected layers, which are 20 neurons and 2 neurons, respectively.

It should be noted that the network structure part may further include other layers of convolution layers, such as a 3-layer convolution layer, a 5-layer convolution layer, a 9-layer convolution layer, etc., and may also include a full-connection layer of other layers. Such as a 1-layer fully connected layer, a 3-layer fully connected layer, and the like. It is also possible to increase the pooling layer or not to use the pooling layer. The convolution kernel size can be other sizes, such as 2 x 2. Convolution kernels of different sizes can also be used for different convolutional layers. For example, the first layer convolution layer uses a 3×3 convolution kernel, and the other layer convolution layer uses a 2×2 convolution kernel. The sliding step size can be unified to 2 or other values, or a different sliding step size can be used, such as a sliding step of 2 for the first layer and a sliding step of 1 for the other layers.

The network training part includes a classifier, and the classifier can be a Softmax classifier.

Referring to FIG. 5, FIG. 5 is another schematic flowchart of a background application control method according to an embodiment of the present application. The training methods specifically include:

1031. The first sample map and the second sample map are respectively input as training data into two sub-reference models.

1032. Synthesize the output values of the two sub-reference models into the input classifier, and obtain a probability corresponding to the plurality of prediction results.

It should be noted that the output values of the two sub-reference models are combined into the input classifier, and the output values of the two sub-reference models can be combined into the input classifier according to different weights. By weighting the output values of the two sub-reference models, two shallower convolutional neural networks can be used as weak classifiers and then merged into strong classifiers. The specific formula is as follows:

Z _K =Z _K ^APP +λ*Z _K ^Device ,

Where λ is the weight, Z _K ^APP is the output value of the first sub-reference model, and Z _K ^Device is the output value of the second sub-reference model.

In some embodiments, the probability of obtaining the prediction result may be based on the first preset formula to synthesize the output values of the two sub-reference models into the input classifier, and obtain a probability corresponding to the plurality of prediction results, wherein the first preset formula is:

Where Z _K is the composite value of the output values of the two sub-reference models, and C is the number of categories of the prediction results,

Is the jth composite value.

1033. Obtain a loss value according to multiple prediction results and multiple probabilities corresponding thereto.

In some embodiments, obtaining the loss value may obtain a loss value according to the plurality of prediction results and the plurality of probabilities corresponding thereto according to the second preset formula, where the second preset formula is:

Where C is the number of categories of predictions and y _k is the true value.

1034, training according to the loss value, and obtaining optimized parameters.

Training can be performed using a stochastic gradient descent method based on the loss value. Training can also be performed according to the batch gradient descent method or the gradient descent method.

Training is performed using the stochastic gradient descent method, and the training can be completed when the loss value is equal to or less than the preset loss value. It is also possible to complete the training when there are no changes in the two or more loss values continuously acquired. Of course, it is also possible to directly set the number of iterations of the random gradient descent method according to the loss value. After the number of iterations is completed, the training is completed. After the training is completed, each parameter of the reference model at this time is obtained, and the each parameter is saved as an optimization parameter, and when the prediction is needed later, the optimization parameter is used for prediction.

Further, the obtained loss value may be obtained according to the third preset formula according to the plurality of sets of parameters, and each set of parameters includes a plurality of prediction results and a plurality of probability corresponding to the obtained loss values, wherein the third preset formula is:

Among them, the optimal parameters can be trained in a small batch manner. If the batch size is 128, E in the third preset formula is expressed as the average of 128 loss values.

Specifically, multiple sample sets may be acquired first, and multiple sample sets are constructed into multiple two-dimensional sample images, and then multiple sample images are input as training data into the reference model to obtain multiple loss values, and then multiple loss values are obtained. average of.

104. Acquire a preset background application and a plurality of feature information of the electronic device, and form a first feature map and a second feature map, and generate a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map. And control the default background application based on the predicted results.

If it is necessary to determine whether the current background application is cleanable, obtain a preset background application and a plurality of current feature information of the electronic device, and form a two-dimensional first feature map and a second feature map, and the first feature map and the second feature are obtained. The graph is input to the reference model, and the reference model is calculated based on the optimized parameters to obtain the predicted value. Determine if the default background application needs to be cleaned up.

It should be noted that the training process of the reference model can be completed on the server side or on the electronic device side. When the training process and the actual prediction process of the reference model are completed on the server side, when the optimized reference model needs to be used, the preset background application and the current multiple feature information of the electronic device may be formed into a feature map and input to the server. After the actual prediction of the server is completed, the prediction result is sent to the electronic device end, and the electronic device controls the preset background application according to the prediction result.

When the training process and the actual prediction process of the reference model are completed on the electronic device end, when the optimized reference model needs to be used, the preset background application and the current plurality of feature information of the electronic device may be formed into the first feature map and the first feature map. The second feature map is input to the electronic device, and after the actual prediction of the electronic device is completed, the electronic device controls the preset background application according to the predicted result.

Please refer to FIG. 6. FIG. 6 is a schematic diagram of another application scenario of a background application control device according to an embodiment of the present disclosure. When the training process of the reference model is completed on the server side, when the actual prediction process of the reference model is completed on the electronic device side, when the optimized reference model needs to be used, the preset background application and the plurality of feature information of the electronic device can be formed. The first feature map and the second feature map are input to the electronic device, and after the actual prediction of the electronic device is completed, the electronic device controls the preset background application according to the predicted result. Optionally, the trained reference model file (model file) can be transplanted to the smart device. If it is necessary to determine whether the current background application can be cleaned up, the current sample image is updated, and the trained reference model file (model file) is input. , the calculation can get the predicted value.

In some embodiments, before acquiring the preset background application and the current plurality of feature information of the electronic device, the method includes:

A preset background application is detected to enter the background. If the background is entered, the preset background application and the current plurality of feature information of the electronic device are acquired, and the first feature map and the second feature map are formed. Then, the prediction is performed according to the reference model, the optimization parameter, the first feature map, and the second feature map, and the prediction result is generated, and the preset background application is controlled according to the prediction result.

The preset time is obtained. If the current system time reaches the preset time, the preset background application and the current plurality of feature information of the electronic device are acquired, and the first feature map and the second feature map are formed. The preset time can be a time point in the day, such as 9 am, or several time points in the day, such as 9 am, 6 pm, and the like. It can also be one or several time points in multiple days. Then, the prediction result is generated according to the reference model, the optimization parameter, the first feature map, and the second feature map, and the preset background application is controlled according to the prediction result.

It should be noted that the reference model may also include multiple sub-reference models, such as three, five, and the like. Each sub-reference model can input different sample maps, or two or more of them can input the same sample map.

All of the above technical solutions may be combined to form an optional embodiment of the present application, and will not be further described herein.

Many applications are open at the same time on the electronic device. Many applications in the background will not be used for a period of time. If these applications are not cleaned up, they consume a lot of memory and consume a lot of power. Therefore, accurately determining the applications that can be cleaned is of great significance for improving the user experience. The traditional method of judging whether an application can be cleaned based on user behavior habits has the problem of insufficient prediction accuracy.

It can be seen from the above that the background application control method provided by the embodiment of the present application acquires the second sample set of the electronic device by acquiring the first sample set of the preset background application; and the first sample set and the second sample set are obtained. Constructing a two-dimensional first sample map and a second sample map respectively; and inputting the first sample map and the second sample map as training data into the reference model, and learning, obtaining the optimized parameters of the trained reference model; Setting a plurality of feature information of the background application and the electronic device, and forming a first feature map and a second feature map, and generating a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map, and according to the prediction As a result, the default background application is managed. It can improve the accuracy of forecasting the default background application, thus improving the intelligence and accuracy of controlling the application entering the background. There is no need to do a lot of feature engineering for user behavior, that is, there is no need to select appropriate user behaviors and appropriate pre-processing, and the quality of the feature engineering has a great impact on the final result. Simply enter the property information into the reference model.

Please refer to FIG. 7. FIG. 7 is a schematic structural diagram of a background application control device according to an embodiment of the present disclosure. The background application management device 300 is applied to an electronic device, and the background application management device 300 includes an obtaining unit 301, a sample map generating unit 302, a training unit 303, and a control unit 304.

The obtaining unit 301 is configured to acquire a first sample set of the preset background application, and acquire a second sample set of the electronic device, where the samples in the first sample set and the second sample set respectively include a preset background application. And multiple feature information of the electronic device.

Specifically, the sample of the first sample set may include usage information of the preset background application, and the sample of the second sample set may include at least one of status information, time information, and location information of the electronic device.

The usage information of the application may include, for example, usage time, background time, application type, application association information, and the like. The status information of the electronic device may include, for example, screen brightness, state of charge, remaining power, WIFI status, and the like. The time information may include, for example, a current time period, a work day, and the like. The location information may include, for example, GPS positioning, base station positioning, WIFI positioning, and the like.

A plurality of feature information is collected as a sample, and then a first sample set of the preset background application and a second sample set of the electronic device are formed.

The default background application may be any application installed in the electronic device, such as a communication application, a multimedia application, a game application, a news application, or a shopping application.

The sample set may include a plurality of samples acquired at a preset frequency during a historical time period. The historical time period can be, for example, within the past 15 days, within 7 days, and the like. The preset frequency can be, for example, every 10 minutes, 30 minutes, and the like.

The sample map generating unit 302 is configured to respectively construct the first sample set and the second sample set into a two-dimensional first sample map and a second sample map.

According to the type of the feature information of the sub-device, the information may be divided into several types of feature information, for example, the feature information of the electronic device is divided into four categories: state information, time information, network information, and location information of the electronic device, and then the four types of feature information are respectively formed into one sub-category. The sample map is then set up to form a large second sample map. The subsample image can be used as 6×6 pixels. If the feature information is not enough to fill the subsample graph, the position is zero-filled. The four subsample plots form a 12 x 12 large second sample plot. Both the second sample map and the subsample map are two-dimensional maps.

The training unit 303 is configured to acquire a reference model, and input the first sample map and the second sample map as training data into the reference model, and learn to obtain optimized parameters of the trained reference model.

The reference model includes two sub-reference models, and the two sub-reference models are convolutional neural network models. Of course, the sub-reference model can be a hybrid neural network model, a Gaussian mixture model, or the like.

Please refer to FIG. 8. FIG. 8 is a partial structural diagram of a reference model according to an embodiment of the present application. Specifically, the reference model includes two sub-reference models, and the two sub-reference models are convolutional neural network models. Of course, the sub-reference models may be hybrid neural network models, Gaussian mixture models, and the like. The convolutional neural network model includes a convolutional layer 3031 and a fully connected layer 3032 that are sequentially connected, and the reference model further includes a classifier 3033. Specifically, the reference model mainly includes a network structure part and a network training part, wherein the network structure part includes a convolution layer 3031 and a full connection layer 3032 which are sequentially connected. A pooling layer (not shown) may also be included between the convoluted layer 3031 and the fully connected layer 3032. The first sample map 305 and the second sample map 306 are respectively input as training data into the convolution layer 3031 of the two sub-reference models.

Optionally, the network structure part of the convolutional neural network model reference model may include a seven-layer network, the first five layers are convolutional layers 3031, the convolution kernel size is uniformly 3×3, and the sliding step length is unified to 1, due to the dimension Small, you can not use the pooling layer, the latter two layers are fully connected layers 3032, respectively 20 neurons, 2 neurons.

It should be noted that the network structure part may further include other layers of convolution layers, such as a 3-layer convolution layer, a 7-layer convolution layer, a 9-layer convolution layer, etc., and may also include a full-connection layer of other layers. Such as a 1-layer fully connected layer, a 3-layer fully connected layer, and the like. It is also possible to increase the pooling layer or not to use the pooling layer. The convolution kernel size can be other sizes, such as 2 x 2. Convolution kernels of different sizes can also be used for different convolutional layers. For example, the first layer convolution layer uses a 3×3 convolution kernel, and the other layer convolution layer uses a 2×2 convolution kernel. The sliding step size can be unified to 2 or other values, or a different sliding step size can be used, such as a sliding step of 2 for the first layer and a sliding step of 1 for the other layers.

The network training portion includes a classifier 3033, and the classifier can be a Softmax classifier.

Referring to FIG. 9 , FIG. 9 is another schematic structural diagram of a background application control device according to an embodiment of the present application. In some embodiments, the training unit 303 includes two network structure portions, each network structure portion including a convolution layer 3031, a full connection layer 3032, and the training unit further includes a classifier 3033, a loss calculator 3034, and a training subunit 3035.

The convolution layer 3031 can be configured to input the first sample map and the second sample map as training data into the convolution layers of the two sub-reference models respectively to obtain a first intermediate value and a second intermediate value.

The fully connected layer 3032 can be configured to process the first intermediate value and the second intermediate value to obtain a third intermediate value and a fourth intermediate value.

The classifier 3033 can be configured to synthesize the output values of the two sub-reference models into the input classifier and obtain the probability corresponding to the plurality of prediction results. The third intermediate value and the fourth intermediate value are combined into the input classifier to obtain a probability corresponding to the plurality of prediction results.

Z _K =Z _K ^APP +λ*Z _K ^Device ,

Where Z _K is a composite value of the output values of the two sub-reference models, that is, a composite value of the third intermediate value and the fourth intermediate value, C is the number of categories of the prediction result, and Z _j is the j-th composite value.

The loss value calculator 3034 can be used to obtain a loss value based on a plurality of prediction results and a plurality of probabilities corresponding thereto.

The obtained loss value may be based on the second preset formula, and the loss value is obtained according to the multiple prediction results and the multiple probabilities corresponding thereto, where the second preset formula is:

Where C is the number of categories of predictions and y _k is the true value.

The training sub-unit 3035 can be used to train according to the loss value to obtain optimized parameters.

Specifically, the random gradient descent method can be used for training according to the loss value. Training can also be performed according to the gradient descent method or the batch gradient descent method.

The control unit 304 is configured to acquire a plurality of feature information of the preset background application and the electronic device, and form a first feature map and a second feature map according to the reference model, the optimization parameter, the first feature map, and the second feature map. , generate prediction results, and control the default background application based on the prediction results.

In some embodiments, the control unit 304 is further configured to detect whether the preset background application enters the background, and if the background is entered, acquire the preset background application and the plurality of feature information of the electronic device, and form the first feature. Figure and second feature map. Then, the prediction is performed according to the reference model, the optimization parameter, the first feature map, and the second feature map, and the prediction result is generated, and the preset background application is controlled according to the prediction result.

In some embodiments, the control unit 304 is further configured to acquire a preset time, and if the current system time reaches the preset time, acquire the preset background application and the plurality of feature information of the electronic device, and form the first feature. Figure and second feature map. The preset time can be a time point in the day, such as 9 am, or several time points in the day, such as 9 am, 6 pm, and the like. It can also be one or several time points in multiple days. Then, the prediction result is generated according to the reference model, the optimization parameter, the first feature map, and the second feature map, and the preset background application is controlled according to the prediction result.

It can be seen from the above that the background application control device of the embodiment of the present application is applied to an electronic device, and acquires a second sample set of the electronic device by acquiring a first sample set of the preset background application; The second sample set is respectively constructed into a two-dimensional first sample map and a second sample map; the first sample map and the second sample map are input as reference data for training data, and learning is performed to obtain an optimized reference model after training. Obtaining a plurality of feature information of the preset background application and the electronic device, and forming a first feature map and a second feature map, and generating a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map And control the default background application based on the predicted results. It can improve the accuracy of forecasting the default background application, thus improving the intelligence and accuracy of controlling the application entering the background.

In the embodiment of the present application, the background application control device and the background application control method in the above embodiment are in the same concept, and any method provided in the embodiment of the background application management and control method may be run on the background application control device. For details of the implementation process, refer to the embodiment of the background application management and control method, which is not described here.

An embodiment of the present application further provides an electronic device. Referring to FIG. 10, the electronic device 400 includes a processor 401 and a memory 402. The processor 401 is electrically connected to the memory 402.

The processor 400 is a control center of the electronic device 400 that connects various portions of the entire electronic device using various interfaces and lines, executes the electronic by running or loading a computer program stored in the memory 402, and recalling data stored in the memory 402. The various functions of device 400 and processing data to provide overall monitoring of electronic device 400.

The memory 402 can be used to store software programs and modules, and the processor 401 executes various functional applications and data processing by running computer programs and modules stored in the memory 402. The memory 402 can mainly include a storage program area and a storage data area, wherein the storage program area can store an operating system, a computer program required for at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area can be stored according to Data created by the use of electronic devices, etc. Moreover, memory 402 can include high speed random access memory, and can also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, memory 402 can also include a memory controller to provide processor 401 access to memory 402.

In the embodiment of the present application, the processor 401 in the electronic device 400 loads the instructions corresponding to the process of one or more computer programs into the memory 402 according to the following process, and is stored in the memory 402 by the processor 401. The computer program thus implements various functions as follows:

Obtaining a first sample set of the preset background application, acquiring a second sample set of the electronic device; and constructing the first sample set and the second sample set into a two-dimensional first sample map and a second sample map respectively; The first sample map and the second sample map are input as reference data into the reference model, and the optimized parameters of the reference model are obtained after the training; the preset background application and the current plurality of characteristic information of the electronic device are obtained, and the first A feature map and a second feature map generate a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map, and control the preset background application according to the prediction result. It can improve the accuracy of forecasting the default background application, thus improving the intelligence and accuracy of controlling the application entering the background.

In some embodiments, the processor 401 is further configured to:

The reference model includes two sub-reference models, and the two sub-reference models are convolutional neural network models.

In some embodiments, the processor 401 is further configured to:

The reference model also includes a classifier;

The first sample map and the second sample map are respectively input as training data into two sub-reference models;

Combining the output values of the two sub-reference models into the input classifier, and obtaining the probability corresponding to the plurality of prediction results;

Obtaining a loss value according to a plurality of prediction results and a plurality of probabilities corresponding thereto;

Training is performed according to the loss value to obtain optimized parameters.

In some embodiments, the processor 401 is further configured to:

The output values of the two sub-reference models are combined into different input weights.

In some embodiments, the processor 401 is further configured to:

The output values of the two sub-reference models are combined into the input classifier based on the first preset formula, and the probability corresponding to the plurality of prediction results is obtained, wherein the first preset formula is:

Where Z _K is the composite value of the output values of the two sub-reference models, C is the number of categories of the prediction results, and Z _j is the j-th composite value.

In some embodiments, the processor 401 is further configured to:

And obtaining a loss value according to the plurality of prediction results and the plurality of probabilities corresponding thereto according to the second preset formula, wherein the second preset formula is:

Where C is the number of categories of predictions and y _k is the true value.

In some embodiments, the processor 401 is further configured to:

Obtain multiple loss values and train based on the average of multiple loss values.

In some embodiments, the processor 401 is further configured to:

The loss value is obtained according to the plurality of prediction results and the plurality of probabilities corresponding thereto according to the third preset formula, wherein the third preset formula is:

It can be seen from the above that the electronic device provided by the embodiment of the present application acquires the second sample set of the electronic device by acquiring the first sample set of the preset background application; and constructs the first sample set and the second sample set respectively. The first sample map and the second sample map are two-dimensional; the first sample map and the second sample map are input as training data into the reference model, and the optimized parameters of the reference model after training are obtained; and the preset background application is obtained. a plurality of feature information of the program and the electronic device, and forming a first feature map and a second feature map, generating a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map, and predicting the prediction according to the prediction result Set the background application to manage. It can improve the accuracy of forecasting the default background application, thus improving the intelligence and accuracy of controlling the application entering the background.

Referring to FIG. 11 together, in some embodiments, the electronic device 400 may further include: a display 403, a radio frequency circuit 404, an audio circuit 405, and a power source 406. The display 403, the radio frequency circuit 404, the audio circuit 405, and the power source 406 are electrically connected to the processor 401, respectively.

Display 403 can be used to display information entered by a user or information provided to a user, as well as various graphical user interfaces, which can be composed of graphics, text, icons, video, and any combination thereof. The display 403 can include a display panel. In some embodiments, the display panel can be configured in the form of a liquid crystal display (LCD), or an organic light-emitting diode (OLED).

The radio frequency circuit 404 can be used to transmit and receive radio frequency signals to establish wireless communication with network devices or other electronic devices through wireless communication, and to transmit and receive signals with network devices or other electronic devices.

The audio circuit 405 can be used to provide an audio interface between the user and the electronic device through the speaker and the microphone.

Power source 406 can be used to power various components of electronic device 400. In some embodiments, the power supply 406 can be logically coupled to the processor 401 through a power management system to enable functions such as managing charging, discharging, and power management through the power management system.

Although not shown in FIG. 11, the electronic device 400 may further include a camera, a Bluetooth module, and the like, and details are not described herein.

The embodiment of the present application further provides a storage medium, where the storage medium stores a computer program, and when the computer program runs on the computer, causes the computer to execute the application management and control method in any of the above embodiments, for example, acquiring a preset background application. a first sample set, acquiring a second sample set of the electronic device; respectively constructing the first sample set and the second sample set into a two-dimensional first sample map and a second sample map; And the second sample map is used as the training data input reference model to learn, obtain the optimized parameters of the trained reference model; obtain the preset background application and the current plurality of feature information of the electronic device, and form the first feature map and the second The feature map generates a prediction result according to the reference model, the optimization parameter, the first feature map, and the second feature map, and controls the preset background application according to the prediction result.

In the embodiment of the present application, the storage medium may be a magnetic disk, an optical disk, a read only memory (ROM), or a random access memory (RAM).

In the above embodiments, the descriptions of the various embodiments are different, and the details that are not detailed in a certain embodiment can be referred to the related descriptions of other embodiments.

It should be noted that, in the background application management and control method of the embodiment of the present application, a general tester in the field can understand all or part of the process of implementing the background application control method of the embodiment of the present application, and the related program can be controlled by a computer program. The computer program can be stored in a computer readable storage medium, such as in a memory of the electronic device, and executed by at least one processor in the electronic device, and can include, for example, a background application during execution. The flow of an embodiment of a management method. The storage medium may be a magnetic disk, an optical disk, a read only memory, a random access memory, or the like.

For the background application management device of the embodiment of the present application, each functional module may be integrated into one processing chip, or each module may exist physically separately, or two or more modules may be integrated into one module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules. An integrated module, if implemented in the form of a software functional module and sold or used as a standalone product, may also be stored in a computer readable storage medium such as a read only memory, a magnetic disk or an optical disk.

The background application control method, the storage medium, and the electronic device provided by the embodiments of the present application are described in detail. The principles and implementation manners of the application are described in the specific examples. The description of the above embodiments is only The method for understanding the present application and its core idea; at the same time, those skilled in the art, according to the idea of the present application, there will be changes in the specific embodiment and the scope of application, in summary, the present specification The content should not be construed as limiting the application.

Claims

A background application management method for an electronic device, the method comprising:

Obtaining a first sample set of the preset background application, acquiring a second sample set of the electronic device, where the samples in the first sample set and the second sample set respectively include the preset background application and the Denoting a plurality of characteristic information of the electronic device;

Constructing the first sample set and the second sample set into a two-dimensional first sample map and a second sample map, respectively;

Obtaining a reference model, and inputting the first sample map and the second sample map as training data into the reference model, performing learning, and obtaining optimized parameters of the reference model after training;

Acquiring the preset background application and the current plurality of feature information of the electronic device, and forming a first feature map and a second feature map, according to the reference model, the optimization parameter, the first feature map, and The second feature map generates a prediction result, and controls the preset background application according to the prediction result.
The background application management method according to claim 1, wherein the reference model comprises two sub-reference models, and the two sub-reference models are convolutional neural network models.
The background application management method according to claim 2, wherein the reference model further comprises a classifier;

And inputting the first sample map and the second sample map as the training data into the reference model, and learning, and obtaining the optimized parameters of the reference model after training includes:

And inputting the first sample map and the second sample map as training data into two of the sub-reference models respectively;

Combining output values of two of the sub-reference models into the classifier, and obtaining a probability corresponding to the plurality of prediction results;

Obtaining a loss value according to the plurality of prediction results and a plurality of the probabilities corresponding thereto;

Training is performed according to the loss value to obtain the optimization parameter.
The background application management method according to claim 3, wherein the training according to the loss value specifically includes:

Training is performed using a stochastic gradient descent method based on the loss value.
The background application management method according to claim 3, wherein the synthesizing the output values of the two sub-reference models into the classifier comprises:

The output values of the two sub-reference models are combined into the classifier by different weights.
The background application management method according to claim 3, wherein the synthesizing the output values of the two sub-reference models into the classifier and obtaining the probability corresponding to the plurality of prediction results specifically includes:

Generating an output value of two of the sub-reference models into the classifier based on a first preset formula, and obtaining a probability corresponding to the plurality of prediction results, wherein the first preset formula is:

Where Z K is a composite value of the output values of the two sub-reference models, C is the number of categories of the prediction result, and Z j is the j-th composite value.
The background application management method according to claim 3, wherein the obtaining the loss value according to the plurality of the prediction results and the plurality of the probabilities corresponding thereto specifically comprises:

And obtaining, according to the second preset formula, a loss value according to the plurality of the prediction results and a plurality of the probabilities corresponding thereto, wherein the second preset formula is:

Where C is the number of categories of predictions and y k is the true value.
The background application management method according to claim 3, wherein the training according to the loss value specifically includes:

A plurality of the loss values are obtained, and training is performed based on an average of the plurality of the loss values.
The background application management method according to claim 8, wherein the obtaining the loss value according to the plurality of the prediction results and the plurality of the probabilities corresponding thereto specifically comprises:

And obtaining, according to a third preset formula, a loss value according to the plurality of the prediction results and a plurality of the probabilities corresponding thereto, wherein the third preset formula is:

Where C is the number of categories of prediction results, y k is the true value, and E is the average.
The background application management method according to claim 1, wherein the obtaining the preset background application and the plurality of feature information of the electronic device specifically includes:

Get the preset time;

If the current system time reaches the preset time, the preset background application and the plurality of feature information of the electronic device are obtained.
The background application management method according to claim 1, wherein the obtaining the preset background application and the plurality of feature information of the electronic device specifically includes:

Detecting whether the default background application enters the background;

If the background is entered, the preset background application and the plurality of feature information of the electronic device are obtained.
A storage medium having stored thereon a computer program, wherein when the computer program is run on a computer, the computer is caused to execute the background application management method according to any one of claims 1 to 12.
An electronic device comprising a processor and a memory, the memory having a computer program, wherein the processor calls the computer program, wherein the processor further executes:

Obtaining a first sample set of the preset background application, acquiring a second sample set of the electronic device, where the samples in the first sample set and the second sample set respectively include the preset background application and the Denoting a plurality of characteristic information of the electronic device;

Constructing the first sample set and the second sample set into a two-dimensional first sample map and a second sample map, respectively;

Obtaining a reference model, and inputting the first sample map and the second sample map as training data into the reference model, performing learning, and obtaining optimized parameters of the reference model after training;

Acquiring the preset background application and the current plurality of feature information of the electronic device, and forming a first feature map and a second feature map, according to the reference model, the optimization parameter, the first feature map, and The second feature map generates a prediction result, and controls the preset background application according to the prediction result.
The electronic device of claim 13, wherein the reference model comprises two sub-reference models, two of which are convolutional neural network models.
The electronic device of claim 14, wherein the reference model further comprises a classifier;

And inputting, by the first sample map and the second sample map as the training data into the reference model, learning, and obtaining the optimized parameters of the reference model after training, the processor further performing:

And inputting the first sample map and the second sample map as training data into two of the sub-reference models respectively;

Combining output values of two of the sub-reference models into the classifier, and obtaining a probability corresponding to the plurality of prediction results;

Obtaining a loss value according to the plurality of prediction results and a plurality of the probabilities corresponding thereto;

Training is performed according to the loss value to obtain the optimization parameter.
The electronic device of claim 15, wherein the reference model further comprises a classifier;

In training based on the loss value, the processor also performs:

Training is performed using a stochastic gradient descent method based on the loss value.
The electronic device of claim 15, wherein the reference model further comprises a classifier;

In synthesizing the output values of the two sub-reference models into the classifier and obtaining a probability corresponding to the plurality of prediction results, the processor further performs:

Generating an output value of two of the sub-reference models into the classifier based on a first preset formula, and obtaining a probability corresponding to the plurality of prediction results, wherein the first preset formula is:

Where Z K is a composite value of the output values of the two sub-reference models, C is the number of categories of the prediction result, and Z j is the j-th composite value.
The electronic device of claim 15, wherein the reference model further comprises a classifier;

In the obtaining a loss value according to the plurality of the prediction results and the plurality of the probabilities corresponding thereto, the processor further performs: performing, according to the second preset formula, the plurality of the prediction results and the plurality of the corresponding ones The probability yields a loss value, wherein the second preset formula is:

Where C is the number of categories of predictions and y k is the true value.
The electronic device of claim 15, wherein the reference model further comprises a classifier;

In training based on the loss value, the processor also performs:

A plurality of the loss values are obtained, and training is performed based on an average of the plurality of the loss values.
The electronic device according to claim 19, wherein said processor further performs: in accordance with a plurality of said prediction results and a plurality of said probabilities corresponding to the loss values, said processor further performing:

And obtaining, according to a third preset formula, a loss value according to the plurality of the prediction results and a plurality of the probabilities corresponding thereto, wherein the third preset formula is:

Where C is the number of categories of prediction results, y k is the true value, and E is the average.