Resource scheduling method and device, electronic equipment and storage medium

Abstract

The invention discloses a resource scheduling method and device, electronic equipment and a storage medium, and belongs to the technical field of electronics. The method comprises the steps that under the condition that at least two application interfaces are displayed on the folding screen, if operation events on the application interfaces are detected, an interface data set is obtained, and the interface data set comprises operation information of the operation events and application information corresponding to the operation events; based on the interface data set, an interface prediction result is obtained, and the interface prediction result represents the attention degree of an application interface in the folding screen; and performing resource scheduling on the at least two application interfaces based on the interface prediction result.

Description

Technical Field

[0001] This application belongs to the field of electronic technology, and particularly relates to a resource scheduling method, apparatus, electronic device, and storage medium. Background Art

[0002] With the development of technology, the display technology of electronic devices has made remarkable progress. In particular, the emergence of foldable screen technology enables electronic devices to display two different application interfaces on the screen, such as simultaneously displaying the interfaces of a game application and a social application. This display method of multiple application interfaces improves the efficiency of users' operations on multiple applications and meets the needs of users for high efficiency and convenience during the use of electronic devices.

[0003] Generally, when an electronic device simultaneously displays two application interfaces through a foldable screen, the electronic device needs to allocate resources to these two application interfaces respectively, such as computing resources, memory, network bandwidth, etc., to ensure the normal use of these two application interfaces.

[0004] However, when an electronic device simultaneously displays multiple application interfaces, it needs to allocate a large amount of resources to these application interfaces at the same time, resulting in insufficient resources allocated to each application interface, and poor fluency of application operation, such as running stuttering, slow loading, etc. Thus, when simultaneously displaying multiple application interfaces, how to allocate resources to these multiple application interfaces is an urgent problem to be solved. Summary of the Invention

[0005] The purpose of the embodiments of this application is to provide a resource scheduling method, apparatus, electronic device, and storage medium, which can improve the flexibility of resource scheduling for application interfaces, thereby ensuring the fluency of application operation.

[0006] In a first aspect, the embodiments of this application provide a resource scheduling method, which includes: when at least two application interfaces are displayed on a foldable screen, if an operation event for an application interface is detected, obtain an interface data set, where the interface data set includes operation information of the operation event and application information corresponding to the operation event; based on the interface data set, obtain an interface prediction result, where the interface prediction result represents the attention degree to the application interfaces in the foldable screen; based on the interface prediction result, perform resource scheduling for at least two application interfaces.

[0007] Second aspect, an embodiment of the present application provides a resource scheduling device, which includes an acquisition module and an execution module. The acquisition module is configured to, when at least two application interfaces are displayed on a foldable screen, if an operation event on an application interface is detected, acquire an interface data set, where the interface data set includes operation information of the operation event and application information corresponding to the operation event. The acquisition module is configured to, based on the interface data set, acquire an interface prediction result, where the interface prediction result represents the attention degree to the application interfaces in the foldable screen. The execution module is configured to, based on the interface prediction result, perform resource scheduling on at least two application interfaces.

[0008] Third aspect, an embodiment of the present application provides an electronic device, which includes a processor and a memory. The memory stores a program or instruction that can run on the processor. When the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.

[0009] Fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0010] Fifth aspect, an embodiment of the present application provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor, and the processor is configured to run a program or instruction to implement the method described in the first aspect.

[0011] Sixth aspect, an embodiment of the present application provides a computer program product, which is stored in a storage medium and is executed by at least one processor to implement the method described in the first aspect.

[0012] In an embodiment of the present application, when at least two application interfaces are displayed on a folding screen, if an operation event on an application interface is detected, an interface data set is obtained. The interface data set includes the operation information of the operation event and the application information corresponding to the operation event. Then, based on the interface data set, an interface prediction result can be obtained, which characterizes the attention degree to the application interfaces in the folding screen. Then, based on the interface prediction result, resource scheduling can be performed on at least two application interfaces. In this solution, since the operation events of the user on the application interfaces in the folding screen can be detected to obtain the operation conditions of the application interfaces and the information of the applications corresponding to the operations, and then these information can be analyzed and processed to obtain the attention degree of the user to the application interfaces in the folding screen, so that resource scheduling can be performed on the multiple displayed application interfaces based on the interface prediction result. That is, this solution can dynamically adjust resource allocation according to the attention degree of the user to the application interfaces. Therefore, sufficient resources can be allocated to the application interfaces that the user pays attention to, and resource waste caused by allocating too many resources to the application interfaces that the user does not pay attention to can be avoided, improving the flexibility of resource scheduling for application interfaces, and thus ensuring the smoothness of application operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 is one of the flowcharts of the resource scheduling method provided by an embodiment of the present application;

[0014] Figure 2 is another flowchart of the resource scheduling method provided by an embodiment of the present application;

[0015] Figure 3 is yet another flowchart of the resource scheduling method provided by an embodiment of the present application;

[0016] Figure 4 is still another flowchart of the resource scheduling method provided by an embodiment of the present application;

[0017] Figure 5 is one of the flowcharts of the resource scheduling method provided by an embodiment of the present application;

[0018] Figure 6 is another flowchart of the resource scheduling method provided by an embodiment of the present application;

[0019] Figure 7 is yet another flowchart of the resource scheduling method provided by an embodiment of the present application;

[0020] Figure 8 is still another flowchart of the resource scheduling method provided by an embodiment of the present application;

[0021] Figure 9 is one of the schematic diagrams of the application interface provided by an embodiment of the present application;

[0022] Figure 10It is a schematic diagram of the included angle between the main screen and the secondary screen provided by an embodiment of the present application;

[0023] Figure 11 It is the second schematic diagram of the application interface provided by an embodiment of the present application;

[0024] Figure 12 It is a schematic diagram of the execution process provided by an embodiment of the present application;

[0025] Figure 13 It is a schematic diagram of the resource scheduling device provided by an embodiment of the present application;

[0026] Figure 14 It is a schematic diagram of the structure of the electronic device provided by an embodiment of the present application;

[0027] Figure 15 It is a schematic diagram of the hardware structure of the electronic device provided by an embodiment of the present application. Detailed implementation manners

[0028] Next, the technical solutions in the embodiments of the present application will be clearly described with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application belong to the scope protected by the present application.

[0029] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order different from those illustrated or described herein, and the objects distinguished by "first", "second", etc. generally belong to the same category, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / " generally means an "or" relationship between the associated objects before and after.

[0030] The terms "at least one (item)", "at least one of", etc. in the present application refer to any one, any two or more combinations of the objects it contains. For example, at least one (item) of a, b, and c can represent: "a", "b", "c", "a and b", "a and c", "b and c", and "a, b, and c", where a, b, and c can be single or multiple. Similarly, "at least two (items)" means two or more, and its meaning is similar to that of "at least one (item)".

[0031] The following will, in conjunction with the accompanying drawings, elaborate in detail on the resource scheduling method, apparatus, electronic device, and storage medium provided by the embodiments of the present application through specific embodiments and their application scenarios.

[0032] The embodiments of the present application can be applied to scenarios where resource scheduling is performed for multiple application interfaces displayed on an electronic device.

[0033] The following will take some specific scenarios of the embodiments of the present application as examples to exemplarily illustrate the resource scheduling method provided by the embodiments of the present application.

[0034] Scenario 1: Game and Social

[0035] Suppose the user uses the main screen of the electronic device to display the application interface of a game, and at the same time uses the secondary screen of the electronic device to display the social application interface. The user may, during the gaps in the game, such as when loading the game or waiting to revive, view the messages in the social application on the secondary screen, and then quickly switch their attention back to the game. In this scenario, the user's attention to the two application interfaces will change dynamically, and it is necessary to reasonably schedule resources for the applications, allocate resources to the game application when the user is focused on the game, and quickly load the messages of the social application when the user is not focused on the game.

[0036] Scenario 2: Video Playback and Document Browsing

[0037] Suppose the user uses the main screen of the electronic device to watch a high-definition video, and at the same time uses the secondary screen of the electronic device to browse documents. The user may occasionally view the documents during the video playback, or pause the video when searching for information in the documents. The user may be focused on the exciting parts of the movie, and at this time the electronic device needs to use more resources for movie playback, and the refresh rate of the documents displayed on the secondary screen can be appropriately reduced to save power; when the movie plot is a bit dull, the user may read the content in the documents, and at this time the secondary screen needs to respond quickly and preload more document content in advance. Therefore, the electronic device needs to reasonably allocate resources to balance power consumption and user experience.

[0038] Scenario 3: Multitasking Office

[0039] Suppose the user simultaneously opens multiple office applications on a foldable screen device, such as performing table editing on the main screen and viewing emails on the secondary screen. The user may need to frequently switch tasks, such as viewing the data in the emails when editing the table, or returning to the table to make modifications after searching for information in the emails. In this scenario, the loading priorities and interface switching of different applications are difficult to predict, and the electronic device needs to reasonably adjust resource scheduling to balance power consumption and user experience.

[0040] It should be noted that the above Scenarios 1 to 3 only exemplarily list some scenarios where the embodiments of the present application may be applied. In actual implementation, the embodiments of the present application can also be applied to any possible scenarios of more demand resource scheduling, and the embodiments of the present application are not limited thereto.

[0041] The embodiments of the present application provide a resource scheduling method, apparatus, electronic device, and storage medium. Since the electronic device can detect the operation events of the user on the application interface in the folding screen to obtain the operation situation of the application interface and the information of the application corresponding to the operation, and then the electronic device can analyze and process this information to obtain the attention degree of the user to the application interface in the folding screen. Based on this interface prediction result, resources are allocated to multiple displayed application interfaces, that is, this solution can dynamically adjust resource scheduling according to the attention degree of the user to the application interface. Therefore, sufficient resources can be allocated to the application interface that the user pays attention to, avoiding wasting resources by allocating too many resources to the application interface that the user does not pay attention to, improving the flexibility of allocating resources to the application interface, and thus ensuring the smoothness of the application operation.

[0042] The execution subject of the resource scheduling method provided by the embodiments of the present application can be a resource scheduling apparatus, and this resource scheduling apparatus can be an electronic device, or a functional module or functional entity in the electronic device. Hereinafter, taking an electronic device as an example, the technical solution provided by the embodiments of the present application will be described.

[0043] Figure 1 The flowchart of a resource scheduling method provided by the embodiments of the present application is shown. As Figure 1 shown, the resource scheduling method provided by the embodiments of the present application may include the following steps 201 to 203.

[0044] Step 201: When at least two application interfaces are displayed on the folding screen, if an operation event on the application interface is detected, the electronic device obtains an interface data set.

[0045] In the embodiments of the present application, the above interface data set includes the operation information of the operation event and the application information corresponding to the operation event.

[0046] In the embodiments of the present application, the operation information of the above operation event includes the operation type of the operation event and the operation time of the operation event.

[0047] In the embodiments of the present application, the above operation type is the type of the specific interaction behavior executed by the user, and may include types such as click, slide, long press, double click, drag, etc.

[0048] In the embodiments of the present application, the above operation time is the specific time point when the operation event occurs, and is recorded in the form of a time stamp.

[0049] Optionally, in the embodiments of the present application, the electronic device may collect events based on a unified timestamp set T = {t1, t2, …, t n}, where t k represents the kth sampling moment, so that the events of the electronic device operating on each application interface can be processed under a unified time reference system.

[0050] In the embodiments of the present application, the application information corresponding to the above operation events includes the identifier of the application interface corresponding to the operation event, the running state of the application corresponding to the operation event, and the visibility of the application interface corresponding to the operation event.

[0051] In the embodiments of the present application, the identifier of the above application interface is a label or identifier used to uniquely identify a certain application interface. The identifier of the application interface can be the name, icon, ID, etc. of the application, which can help the device distinguish different application interfaces and ensure that subsequent resource scheduling and processing can accurately target specific applications.

[0052] In the embodiments of the present application, the running state of the above application represents the execution situation of the application on the current device, which is mainly divided into the foreground state and the background state. Among them:

[0053] (1) The foreground state is the state where the application is running in the foreground, that is, the application interface that the current user is directly interacting with. The electronic device needs to allocate more system resources to ensure its smooth operation to ensure the user experience.

[0054] (2) The background state is the state where the application is running in the background. The application can perform some auxiliary tasks in a silent state or wait for the user to switch back to the foreground. The electronic device can allocate fewer system resources to background applications, such as restricting their network access and reducing their computing priority, to avoid negatively affecting the performance of foreground applications.

[0055] In the embodiments of the present application, the visibility of the above application interface is the display situation of the application interface on the screen, which can be specifically divided into the following three states:

[0056] (1) Completely visible: The application interface is completely displayed on the screen, and the user can clearly see all its content and interface elements. At this time, the electronic device can allocate normal display resources and computing resources to ensure its normal operation and rendering.

[0057] (2) Partially visible: Only a part of the application interface is displayed on the screen, and it may be blocked by other application interfaces or in a partial display area of a foldable screen. In this case, the electronic device may dynamically adjust the resources allocated to the application according to the size and importance of the visible area to balance performance and power consumption.

[0058] (3) Invisible: The application interface is not displayed on the screen at all, and may be covered by other application interfaces, in the background, or hidden by the user. At this time, the electronic device can significantly reduce the resource scheduling for this application and can pause some of its functions,

[0059] to save system resources and extend battery life.

[0060] In the embodiments of the present application, the above-mentioned foldable screen is a screen that can be folded. This kind of screen uses flexible screen materials and special mechanical mechanisms to achieve the folding function, so as to allow the electronic device to flexibly adjust the display area in different usage scenarios.

[0061] Optionally, in the embodiments of the present application, the above-mentioned application interface can be the application interface of any application, such as a game application, a video application, a social application, a shopping application, etc.

[0062] Optionally, in the embodiments of the present application, the above-mentioned operation event is any interaction behavior of the user on the application interface, which can be an operation directly on the application interface in the touch screen, such as touch, swipe, click and other inputs, or an operation on the application interface through an input device, such as being triggered by an input device such as a mouse, keyboard, or gamepad. The electronic device can respond to the user's operation by detecting and processing these operation events.

[0063] Optionally, in the embodiments of the present application, the electronic device can perform multi-source data collection and preliminary marking for the user operation behavior, application foreground / background state, and interface visibility in a multi-application interface scenario, that is, a dual-screen separation scenario.

[0064] Optionally, in the embodiments of the present application, when the electronic device is displaying two application interfaces, it can perform real-time collection for each user operation event, and the collection content can include: touch events, input events, swipe events, and screen switching events.

[0065] Optionally, in the embodiments of the present application, when recording an event, the electronic device can record the application interface identifier i and the operation type op k , and combine the occurrence time t of this event k , to generate an initial event sequence E, and its expression form is as follows:

[0066] E = {e1, e2,..., e m}, e j = (i j , op j , t j )

[0067] where j is the event index, i j is the application interface identifier, op j is the operation type symbol, tj is the timestamp of the event occurrence. In this way, the electronic device can form an initial chronological record, that is, an initial event sequence, by integrating all interaction events in the dual-screen scenario of the folding screen.

[0068] Optionally, in the embodiments of the present application, the electronic device can detect the foreground and background states of the application through the operating system, so as to add a foreground or background identifier to each event in the initial event sequence to distinguish whether the user is actually operating on the current application on the screen. Specifically, the electronic device can use the binary scalar f j to represent the foreground and background state of the event. When the application is in the foreground, it is recorded as f j = 1, and when the application is in the background, it is recorded as f j = 0. The electronic device can establish a switching label sequence F = {f1, f2, …, f m} based on this scalar.

[0069] Optionally, in the embodiments of the present application, the electronic device can use the visibility metric function V i (t k ) to characterize the interface visibility of the application interface i at time t k . The value range of V i (t k ) can be [0, 1]. This interface visibility is the integrity of the display of the application interface. For example, if an application interface is completely displayed, its interface visibility is 1. When the application interface is blocked by other interfaces, the interface visibility can take a value between 0 and 1 according to the size of the application interface blocked. For example, if the application interface is blocked by half, the interface visibility of the application interface can be 0.5, and when the application interface is completely blocked, the interface visibility of the application interface is 0.

[0070] Optionally, in the embodiments of the present application, in order to construct a visibility time series model, the electronic device can use the operation interval duration Δt k = t k - t k-1 to measure the time distribution of the operation. Δt k can be obtained by differential calculation of the timestamps of adjacent events. Combining with the visibility metric function V i (t k ), a visibility time series dataset V can be formed in the same time reference system, and its expression form is as follows:

[0071] v = {V1(t1), V1(t2), …, V2(t1), V2(t2), …}

[0072] Optionally, in the embodiments of the present application, the electronic device may fuse the obtained event sequence E, the label sequence F, the visibility time series dataset V, and the operation interval duration sequence ΔT to form a multi-source fusion dataset D.

[0073] Step 202: The electronic device obtains an interface prediction result based on the interface dataset.

[0074] In the embodiments of the present application, the above interface prediction result characterizes the attention degree to the application interface in the folding screen.

[0075] In the embodiments of the present application, the above interface prediction result is a quantitative representation of the user's attention to each application interface in the folding screen, reflecting the degree of attention that the user may have to different application interfaces at the current moment and in a period of time in the future. For example, if the interface prediction result shows that the user has a high degree of attention to a certain game application interface, then the electronic device can allocate more resources to this application accordingly to ensure its smooth operation. The specific degree of attention can be reflected in the form of a numerical value or a probability. For example, the degree of attention can be divided into three levels: high, medium, and low, corresponding to different numerical value ranges. Or, directly use a probability value from 0 to 1 to represent the possibility that the user pays attention to a certain application interface. The closer the probability value is to 1, the higher the degree of attention, and the closer it is to 0, the lower the degree of attention.

[0076] Step 203: The electronic device performs resource scheduling on at least two application interfaces based on the interface prediction result.

[0077] Optionally, in the embodiments of the present application, the electronic device may interpret the prediction result, determine the attention level of each application interface, then formulate a corresponding resource scheduling strategy according to the interpreted prediction result, and finally, according to the formulated resource scheduling strategy, specifically perform the operation of resource scheduling.

[0078] Optionally, in the embodiments of the present application, during the execution of resource scheduling, the electronic device may continuously detect the user's behavior, continuously predict the user's behavior, and timely adjust the resource scheduling strategy according to the new prediction result, and then re-perform resource scheduling to form a continuous resource management process.

[0079] An embodiment of the present application provides a resource scheduling method, device, electronic device, and storage medium. Since the electronic device can detect the operation events of the user on the application interface in the folding screen to obtain the operation situation of the application interface and the information of the application corresponding to the operation, and then the electronic device can analyze and process this information to obtain the attention degree of the user to the application interface in the folding screen, so as to allocate resources to multiple displayed application interfaces based on the interface prediction result. That is, this solution can dynamically adjust resource scheduling according to the attention degree of the user to the application interface. Therefore, sufficient resources can be allocated to the application interface that the user pays attention to, avoiding wasting resources by allocating too many resources to the application interface that the user does not pay attention to, improving the flexibility of allocating resources to the application interface, and thus ensuring the smoothness of the application operation.

[0080] Optionally, in the embodiment of the present application, in combination with Figure 1 , as Figure 2 shown, the above step 202 can be specifically implemented by the following steps 202a to 202e.

[0081] Step 202a: Extract scene features and perform lightweight encoding on the interface data set to obtain a sequence of scene lightweight encoding vectors.

[0082] In the embodiment of the present application, the above-mentioned scene feature extraction refers to extracting features related to specific usage scenarios from the interface data set. These features can reflect the user's behavior patterns and application usage habits in different scenarios. For example, the user may pay more attention to the operation response speed in the game scenario, while in the document editing scenario, the user may pay more attention to the content loading and saving speed.

[0083] In the embodiment of the present application, the above-mentioned lightweight encoding refers to compressing and encoding the extracted features to reduce the data dimension and storage space, while retaining key information, which can improve the processing speed and efficiency of the subsequent electronic device for the features.

[0084] Optionally, in the embodiment of the present application, the electronic device can refine scene features with strong pertinence around the special requirements of the multi-application interface scenario, and effectively compress the data scale and retain the key timing information through a lightweight encoding operator.

[0085] Optionally, in the embodiment of the present application, in combination with Figure 2 , as Figure 3 shown, the above step 202a can be specifically implemented by the following steps 202a1 to 202a3.

[0086] Step 202a1: The electronic device extracts scene features from the interface data set through a scene mapping function to obtain a scene feature matrix.

[0087] In the embodiments of the present application, the above-mentioned scenario mapping function is a mathematical function or algorithm used to map the input data to a specific scenario feature space.

[0088] In the embodiments of the present application, the above-mentioned scenario feature extraction refers to extracting features related to specific usage scenarios from the interface dataset, and these features can reflect the user's behavior patterns and application usage habits in different scenarios.

[0089] In the embodiments of the present application, the above-mentioned scenario feature matrix is the result of scenario feature extraction, and this matrix contains all scenario-related features extracted from the interface dataset.

[0090] Optionally, in the embodiments of the present application, the electronic device can combine the special scenario attributes of the folded screen with dual-screen separation to determine a list of scenario feature requirements. This list of scenario feature requirements is a requirement list formulated to accurately capture the key features of the user's behavior and application status in a specific usage scenario. The electronic device can clearly identify the need to focus on processing the screen position, application type, user attention duration, and switching period during the subsequent extraction process through the list of scenario feature requirements.

[0091] Optionally, in the embodiments of the present application, the electronic device can introduce a scenario mapping function to perform a preliminary mapping on the multi-source fusion dataset D based on the list of scenario feature requirements to obtain a multi-dimensional feature matrix. The form of this scenario mapping function is shown in the following formula (1):

[0092]

[0093] where d j is the j-th record in the multi-source fusion dataset, is the k-th dimensional scenario feature extraction operator, and p represents the total dimension of the required features after extraction. Specifically, is strongly related to the folded screen with dual-screen separation scenario, and focuses on processing the screen position identifier, the timing marker when the application switches between the foreground and background, the dynamic change of visibility, and the user attention level.

[0094] Optionally, in the embodiments of the present application, the electronic device can obtain a preprocessing feature matrix by substituting each record in D into the scenario mapping function Φ(·) one by one. The form of this preprocessing feature matrix is shown as follows:

[0095]

[0096] where m is the total number of records in the multi-source fusion dataset.

[0097] Step 202a2: The electronic device compresses the scenario feature matrix through a lightweight encoding function to obtain an encoding mapping result.

[0098] In the embodiments of the present application, the above lightweight encoding function is a function or algorithm for compressing and encoding data, which is used to reduce the dimension and storage space of data while retaining key information, thereby improving the efficiency of subsequent processing.

[0099] In the embodiments of the present application, the above encoding mapping result is the output result after the lightweight encoding function compresses and encodes the scene feature matrix, which is a low-dimensional vector or matrix that retains the key information in the original data.

[0100] Optionally, in the embodiments of the present application, the electronic device may input the preprocessed feature matrix M into the lightweight encoding operator to achieve data dimensionality reduction and retention of core timing and semantic information.

[0101] Optionally, in the embodiments of the present application, the electronic device may set a group of encoding weight vectors w = {w1, w2,..., w p}, and construct an encoding mapping function based on the matrix decomposition idea and the folding screen dual-screen separation interaction mode, that is, the lightweight encoding function, whose form is shown in the following formula (2):

[0102] Ψ(Φ(d j )) = σ(M j ·w T ) Formula (2)

[0103] Wherein, represents the j-th row of the feature vector, and σ(·) is an activation function used to enhance the recognition ability of attention switching features. The electronic device can execute this mapping function for all rows of the matrix M, so as to retain the folding screen dual-screen switching timing features in the lower-dimensional encoded vector space.

[0104] Optionally, in the embodiments of the present application, the electronic device may add the timing weighting factor λ j to the lightweight encoding function, so as to incorporate the interaction event trigger frequency and the visibility mutation amplitude into the encoding process, and can highlight the attention jump rule in the folding screen dual-screen separation scenario. After adding the weighting factor λ j , the form of the lightweight encoding function is shown in the following formula (3):

[0105]

[0106] Wherein, represents the comprehensive change rate of the j-th record in terms of visibility and switching label, and κ is a constant coefficient.

[0107] In this way, the electronic device can increase the sensitivity control of the large attention jump in the dual-screen separation scenario, so that the dynamic information of the jump behavior can also be retained during the dimensionality reduction and compression process.

[0108] Step 202a3: The electronic device integrates the coding mapping results to obtain a sequence of scenario-based lightweight coding vectors.

[0109] Optionally, in the embodiments of the present application, the above sequence of scenario-based lightweight coding vectors is the final output after integrating the coding mapping results, and is a vector sequence arranged in chronological order, in the form of Each vector therein represents the feature information at a time point.

[0110] Optionally, in the embodiments of the present application, the electronic device can integrate the coding mapping results into a sequence of scenario-based lightweight coding vectors C as the input for subsequent model training and inference processes. Through this sequence, the electronic device can retain the core temporal dynamics and dual-screen switching features of the user in the folded-screen dual-screen scenario.

[0111] In this way, the electronic device can significantly reduce the subsequent model operation overhead through the output vector sequence with the characteristics of lightweight and strong scenario relevance, and can accurately retain the temporal structure of attention switching in the complex environment of multi-application interface display.

[0112] Step 202b: The electronic device processes the sequence of scenario-based lightweight coding vectors to obtain a prediction vector.

[0113] In the embodiments of the present application, the above prediction vector is the result of processing the sequence of scenario-based lightweight coding vectors, and contains the prediction information on the attention degree of each application interface in the folded screen.

[0114] Optionally, in the embodiments of the present application, based on the sequence of scenario-based lightweight coding vectors, the electronic device can optimize the network structure for the folded-screen dual-screen separation mode, and construct a miniature vertical large model with both high efficiency and scenario adaptability. This model can realize the fine characterization and unified mapping of the user usage rules in the dual-screen parallel application environment through the dual-domain morphology aggregation strategy and the lightweight multi-task coupling module, and finally output the prediction vector of the user's attention state and possible switching timing as the input for the subsequent information entropy-driven diffusion model.

[0115] Optionally, in the embodiments of the present application, in combination with Figure 2 , as Figure 4 shown, the above step 202b can be specifically implemented through the following steps 202b1 to 202b5.

[0116] Step 202b1: The electronic device extracts features from the sequence of scenario-based lightweight coding vectors through a first morphology aggregation function to obtain a first feature vector.

[0117] In the embodiments of the present application, the above first feature vector characterizes the usage of the main screen of the folding screen, and the above second feature vector characterizes the usage of the secondary screen of the folding screen.

[0118] In the embodiments of the present application, the above first form aggregation function is a function used to extract features related to the usage of the main screen from the sequence of scene-based lightweight encoding vectors, and can aggregate and extract feature information from the operation events and application information of the main screen.

[0119] Step 202b2: The electronic device extracts features from the sequence of scene-based lightweight encoding vectors through the second form aggregation function to obtain a second feature vector.

[0120] In the embodiments of the present application, the above second form aggregation function is a function used to extract features related to the usage of the secondary screen from the sequence of scene-based lightweight encoding vectors, and can aggregate and extract feature information from the operation events and application information of the secondary screen.

[0121] Optionally, in the embodiments of the present application, the electronic device may perform data dimension adaptation on the sequence of scene-based lightweight encoding vectors to make it adapt to the input requirements of the subsequent dual-domain form aggregation layer and the micro backbone network. If the subsequent model components require higher-dimensional inputs, the dimension of the vector can be increased by adding zero values, average values, or other filling methods. If the subsequent model components require lower-dimensional inputs, dimensionality reduction techniques such as principal component analysis (PCA), linear discriminant analysis (LDA), etc. can be used to reduce the dimension of the vector.

[0122] Optionally, in the embodiments of the present application, since the dual-screen usage scenario of the folding screen often involves the interaction behavior between the main screen and the secondary screen at the same time, the electronic device may construct a dual-domain form aggregation layer to perform form transformation on the structural features of c j from the main screen domain and the secondary screen domain respectively. The dual-domain form aggregation layer is specifically two form aggregation functions, namely the above first form aggregation function and the second form aggregation function. The form of the first form aggregation function is shown in the following formula (4):

[0123]

[0124] The form of the second form aggregation function is shown in the following formula (5):

[0125]

[0126] where, Φ s (·) and Φ t (·) respectively represent the form aggregation functions of the main screen domain and the secondary screen domain, and Θ s , Θ t are the corresponding sets of trainable parameters. is the first eigenvector, and is the second eigenvector. The electronic device can use a morphological aggregation function to distinguish and extract the usage characteristics of the main screen and the secondary screen from both the statistical distribution and the behavior pattern, avoiding the decline in representation accuracy caused by the mixing of different screen characteristics.

[0127] Step 202b3: The electronic device fuses the first eigenvector and the second eigenvector through a micro-network function to obtain a characterization vector.

[0128] In the embodiment of the present application, the above micro-network function is a lightweight neural network function used to fuse the first eigenvector and the second eigenvector, which can perform non-linear transformation and combination on the two eigenvectors to generate a comprehensive characterization vector.

[0129] In the embodiment of the present application, the above characterization vector is the result obtained by the micro-network function fusing the first eigenvector and the second eigenvector, and is characterized by integrating the usage characteristics of the main screen and the secondary screen.

[0130] Optionally, in the embodiment of the present application, after obtaining the first eigenvector and the second eigenvector, the electronic device can input the first eigenvector and the second eigenvector into the micro-network function to complete the basic characterization mapping. The form of the micro-network function is shown in the following formula (6):

[0131]

[0132] where Θ Γ is the backbone network parameter, and a set of characterization vectors h j that takes into account the differences between the dual-screen domains is obtained.

[0133] Step 202b4: The electronic device fuses the characterization vector and the application type embedding vector through a cross-domain fusion mapping function to obtain a third eigenvector.

[0134] In the embodiment of the present application, the above cross-domain fusion mapping function is a function used to fuse the characterization vector and the application type embedding vector, and is used to synthesize the usage characteristics of the main screen and the secondary screen and the relevant information of the application type.

[0135] In the embodiment of the present application, the above application type embedding vector is a vector that encodes the application type and is used to represent the characteristics of different application types. For example, game applications, social applications, document editing applications, etc., and each application has its corresponding embedding vector.

[0136] In the embodiments of the present application, the above-mentioned third feature vector is the result obtained by fusing the characterization vector and the application type embedding vector through a cross-domain fusion mapping function. This third feature vector synthesizes the usage characteristics of the main screen and the secondary screen as well as the relevant information of the application type.

[0137] Optionally, in the embodiments of the present application, the electronic device can perform cross-domain fusion mapping on the characterization vector h j to capture the migration law during the use of the main and secondary screens. This mapping can, in view of the potential switching and parallel operation characteristics of the dual screens when using a folding screen, dynamically combine the dual-domain data formed by the main screen and the secondary screen to achieve a unified characterization of screen migration. The form of the cross-domain fusion mapping function is shown in the following formula (7):

[0138] z j =Φ cross ([h j ,u j ,Θ cross ) = W c ·[h j ,u j T +b c Formula (7)

[0139] Among them, the above-mentioned W c ,b c are trainable parameters, u j is the embedding vector related to the application type, and Θ cross is the overall parameter set related to cross-domain fusion mapping. The electronic device can, by adding u j and the h j obtained by aggregating the dual-domain forms into the mapping input, perform fusion modeling on the temporal distribution of the dual-screen domain, the application type characteristics, and the screen migration behavior, and obtain the third feature vector z j that synthesizes the usage characteristics of the main screen and the secondary screen as well as the relevant information of the application type.

[0140] Optionally, in the embodiments of the present application, the electronic device can, through a lightweight multi-task coupling module, realize the coupling processing of the third feature vector z j and the semantic distribution of multiple application scenarios, so as to be compatible with the different running forms of various types of applications in the dual-screen separation mode of the folding screen. Among them, the function of the multi-task coupling module, that is, the form of the multi-task coupling function, is shown in the following formula (8):

[0141] y j =Λ(z j ,Θ Λ ) Formula (8)

[0142] Among them, Θ Λis a parameter set for the multi-task coupling module. In the embodiments of the present application, in order to highlight the dual-screen separation scenario, Λ(·) processes the semantic features corresponding to different application scenarios through different sub-channels internally, and retains several key components most relevant to the screen domain difference through a screening function at the output end, so as to map to a more representative dimension in the dual-screen separation interaction process of the folding screen, so as to obtain a third feature vector y that synthesizes the application name on the basis of the third feature vector z j j .

[0143] Step 202b5: The electronic device transforms the third feature vector through an output layer mapping function to obtain a prediction vector.

[0144] In the embodiments of the present application, the above output layer mapping function is a function for converting the third feature vector into a prediction vector, which is a linear transformation or a non-linear transformation function for generating a final prediction result.

[0145] In the embodiments of the present application, the above prediction vector is the result obtained by the output layer mapping function transforming the third feature vector, and it contains prediction information on the attention degree of each application interface in the folding screen.

[0146] Optionally, in the embodiments of the present application, the electronic device may perform preliminary training on the third feature vector y j to obtain the prediction result of the user's attention state and potential switching timing in the dual-screen separation mode. The form of the output layer mapping function for preliminary training is shown in formula (9) below:

[0147] p j =Π(y j , Θ Π ) Formula (9)

[0148] where Π(·) is the output layer mapping function, Θ Π is the output layer parameter, and p j is the prediction vector.

[0149] Optionally, in the embodiments of the present application, the electronic device may compare p j with the true attention distribution and switching moment label in the folding screen usage scenario to construct a loss function and use a gradient-based optimization method to perform overall training on these parameters {Θ Γ , Θ s , Θ t , Θ cross , Θ Λ , Θ Π}. After the training is completed, the model can be given a scenario-based lightweight feature vector c j and application scenario information u​j When it is, an output prediction vector is generated, which includes estimated values for the current dual-screen attention state and the next switching timing.

[0150] In this way, the electronic device can significantly improve the accuracy of predicting user behavior through technologies such as adaptive information entropy evaluation, dynamic diffusion scheduling, and multi-layer variable diffusion channels.

[0151] Step 202c: The electronic device performs multi-layer variable diffusion on the prediction vector to obtain an attention switching timing distribution.

[0152] In the embodiments of the present application, the above multi-layer variable diffusion is a technology based on a diffusion model, which is used to perform multi-layer diffusion processing on the prediction vector and can capture the switching rules of the user's attention between different time points and different application interfaces.

[0153] In the embodiments of the present application, the above attention switching timing distribution is the attention allocation and switching situation of the user to different application interfaces at different time points.

[0154] Optionally, in the embodiments of the present application, the electronic device can further combine the multi-task complexity of the dual-screen separation scenario based on the prediction vector, and dynamically schedule the diffusion process based on information entropy to obtain a more refined attention switching timing distribution.

[0155] Optionally, in the embodiments of the present application, in combination with Figure 2 , as Figure 5 shown, the above step 202c can be specifically implemented through the following steps 202c1 to 202c3.

[0156] Step 202c1: The electronic device calculates the information entropy of the prediction vector through an information entropy evaluation function to obtain the information entropy.

[0157] Optionally, in the embodiments of the present application, the electronic device can construct an adaptive information entropy evaluation unit for the prediction vector p j to reflect the task mixing degree and the uncertainty of the user's attention distribution in the dual-screen parallel application environment in real time.

[0158] Optionally, in the embodiments of the present application, the electronic device can calculate the information entropy of the prediction vector through an information entropy evaluation function, and the specific form of the information entropy evaluation function is as shown in the following formula (10):

[0159]

[0160] Among them, represents the proportional value of the i-th prediction component in the vector p j , K is the dimension of the prediction vector, and H(p j) is the information entropy. Through this information entropy function, the certainty of the current prediction for attention allocation and switching nodes can be quantified. When the multi-task scenario is relatively complex or the dual-screen attention is in a highly uncertain distribution, H(p j ) will increase significantly, indicating that a more refined diffusion simulation process is required to capture user behavior.

[0161] Step 202c2: The electronic device iterates the information entropy through a diffusion iteration function to obtain a diffusion state vector.

[0162] In the embodiment of the present application, the number of iterations and the diffusion step size of the above iteration function are determined according to the magnitude of the information entropy.

[0163] Optionally, in the embodiment of the present application, the electronic device can dynamically schedule the diffusion model based on H(p j ) so as to allocate more diffusion steps when the user's attention is extremely uncertain or multi-tasks are highly parallel. Conversely, the iteration overhead can be reduced in simple scenarios, thereby improving the efficiency and accuracy of the model in the dual-screen scenario.

[0164] Optionally, in the embodiment of the present application, the electronic device can iterate the information entropy through a diffusion iteration function to finally obtain a diffusion state vector The specific form of this diffusion iteration function is shown in Formula (11) below:

[0165] X t+1 =X t +δ(H(p j ))·g(X t ,ε t ) Formula (11)

[0166] Among them, t represents the diffusion iteration step index, and X t represents the diffusion state obtained in the t-th step of iteration. g(·) is a diffusion increment function, and ε t is a noise vector, and δ(·) is a step size scheduling function. The specific form of this step size scheduling function is shown in Formula (12) below:

[0167]

[0168] Among them, α is the basic step size factor, is a constant coefficient used to improve the diffusion resolution in the case of high entropy. The electronic device can adjust the number of iterations and the diffusion step size through this adaptive step size mechanism. If the uncertainty degree of p j is relatively high, a larger number of iterations and a finer-grained diffusion step size can be allocated, so as to more accurately capture the dynamic process of the user's large-scale attention switching between the dual screens.

[0169] Optionally, in the embodiments of the present application, the electronic device may perform diffusion iteration through a multi-layer variable diffusion channel, thereby further enhancing the adaptability of the diffusion model to scenarios with drastic temporal changes and high multi-task parallelism. This multi-layer variable channel is also called a diffusion channel set, which includes several interconnected diffusion layers. Each layer has different noise sampling intensities and state update rules, and uses the information entropy value as a gating signal for dynamic path selection. The form of the multi-layer variable channel is {C1, C2, …, C L}, where C l represents the l-th layer channel, which has an independent noise sampling distribution and a diffusion function

[0170] Optionally, in the embodiments of the present application, at the t-th iteration, the electronic device may perform path gating according to the relevant dimension values in the information entropy evaluation result and the prediction vector p j . The specific formula is shown in the following formula (13):

[0171]

[0172] where the Gate(·) function is used to select the most suitable diffusion layer l j by comparing the applicable intervals of each channel with the feature distribution of p * to perform iterative update.

[0173] In this way, the electronic device can select a channel with weaker noise and finer diffusion steps when the attention jumps significantly or the competition degree of multiple applications is relatively high, so as to accurately capture the temporal discreteness; while in a simple scenario or with a lower uncertainty, it selects a channel with stronger noise and a larger step size to accelerate diffusion, thus taking into account both accuracy and efficiency.

[0174] Step 202c3: Analyze the diffusion state vector to obtain the attention switching time sequence distribution.

[0175] Optionally, in the embodiments of the present application, the electronic device may parse the diffusion state vector into an attention switching trajectory evolving with time where s t represents the attention allocation and potential switching position of the user between the main and secondary screens at the t-th time period.

[0176] In this way, the electronic device can adapt to different user behavior patterns and usage scenarios by quantifying the uncertainty of the prediction, dynamically simulating the attention change, and accurately predicting the attention switching, thereby improving the accuracy of predicting user behavior.

[0177] Step 202d: The electronic device couples the prediction vector and the attention switching time sequence distribution to obtain a fused prediction vector.

[0178] It should be noted that the above coupling refers to combining two or more data sources to generate more comprehensive and accurate results. In the embodiments of the present application, the electronic device can combine the prediction vector and the attention switching time series distribution to obtain a fused prediction vector.

[0179] In the embodiments of the present application, the above fused prediction vector is the result after coupling, which synthesizes the information of the prediction vector and the attention switching time series distribution, and can provide a more accurate user attention prediction result.

[0180] Optionally, in the embodiments of the present application, the combination Figure 2 , such as Figure 6 shown, the above step 202d can be specifically implemented through the following steps 202d1 to 202d3.

[0181] Step 202d1: The electronic device inputs the prediction vector and the attention switching time series distribution into the coupling mapping function. Through the coupling mapping function, the prediction vector and the attention switching time series distribution are structurally aligned to obtain a fused vector.

[0182] In the embodiments of the present application, the above coupling mapping function is a function for combining two or more data sources.

[0183] In the embodiments of the present application, the above structural alignment is to align two data sources under the same structure or framework, so that the electronic device can perform effective combination and analysis.

[0184] In the embodiments of the present application, the above fused vector is a vector that synthesizes the information of the prediction vector and the attention switching time series distribution.

[0185] Optionally, in the embodiments of the present application, the electronic device can establish a coupling mechanism for the miniature vertical large model prediction vector p j obtained in S3 and the attention switching time series distribution obtained by the diffusion model based on information entropy in S4.

[0186] Optionally, in the embodiments of the present application, the miniature vertical large model can compactly and efficiently predict the user's attention allocation and switching trend at the current moment based on the structure of folding screen dual-domain morphology aggregation, cross-domain fusion, and multi-task coupling, while the diffusion model performs a more in-depth iterative simulation in the time dimension through multi-layer variable channels and adaptive information entropy scheduling. Therefore, the electronic device can take into account the advantages of the miniature vertical large model and the diffusion model through the coupling mapping function. The specific form of this coupling mapping function is shown in the following formula (14):

[0187]

[0188] Among them, F is the fusion vector, and Υ(·) is the fusion operator, which is used to synchronously analyze p j The representational ability of attention and switching mechanism in the short-term range, and the diffusion result of S in the time series are used to achieve the structural alignment of the two.

[0189] Step 202d2: The electronic device inputs the information entropy into the weight adjustment function and calculates the fusion weight.

[0190] In the embodiment of the present application, the weight adjustment function is a function used to calculate weights according to the input information entropy.

[0191] In the embodiment of the present application, the above-mentioned fusion weight is the output of the weight adjustment function, which is used to represent the degree of emphasis on different data sources during the fusion process. When the information entropy is relatively high, the fusion weight tends to trust the prediction result of the diffusion model; conversely, it tends to trust the prediction result of the micro vertical large model.

[0192] Optionally, in the embodiment of the present application, the electronic device can achieve a higher-confidence fusion prediction by mutually correcting the possible local errors of the micro vertical large model and the diffusion model respectively during the fusion process.

[0193] Optionally, in the embodiment of the present application, the electronic device can calculate the dynamic weight based on the weight adjustment function, and the specific form of the weight adjustment function is shown in the following formula (15):

[0194]

[0195] Among them, η is the modulation coefficient, which adapts to the double-screen uncertainty through information entropy, and ω p represents the weight of the prediction result of the diffusion model, reflecting the degree of emphasis on the result of the diffusion model during the fusion prediction. ω s represents the weight of the prediction result of the micro vertical large model, reflecting the degree of emphasis on the result of the micro vertical large model during the fusion prediction. When the uncertainty covered by p j is relatively high, the electronic device can increase its dependence on the output S of the diffusion model; conversely, the electronic device trusts the short-term and efficient prediction of the micro vertical large model more.

[0196] Step 202d3: The electronic device weights the fusion vector according to the fusion weight to obtain the fusion prediction vector.

[0197] It should be noted that the above weighting is to adjust the data according to the weights to reflect the importance of different data sources. In the embodiment of the present application, the electronic device can adjust each element of the fusion vector according to the fusion weight.

[0198] Optionally, in the embodiment of the present application, the electronic device can use ω pand ω s Weight each component in the fusion vector F to obtain a fusion prediction vector M fuse .

[0199] Step 202e: The electronic device analyzes the fusion prediction vector to obtain a prediction result

[0200] In the embodiments of the present application, the above prediction result is a prediction of the user's attention to each application interface in the folding screen and a set of possible switching nodes

[0201] Optionally, in the embodiments of the present application, the electronic device may, based on the fusion prediction vector M fuse , comprehensively infer the attention distribution and possible switching nodes of the user evolving over time in the dual-screen scenario, and map M fuse to two dimensions of the main screen attention curve and the secondary screen attention curve, and combine the estimated switching probability or switching time label to obtain a prediction of the attention allocation for the complete time series, that is, the prediction result

[0202] In this way, the electronic device can obtain an accurate prediction result through efficient data processing, enhanced model prediction, and dynamic attention switching prediction, and thus can use the prediction result to accurately allocate resources and improve the performance of the electronic device when displaying multiple application interfaces

[0203] Optionally, in the embodiments of the present application, the above prediction result includes a main screen attention sequence, a secondary screen attention sequence, and a set of switching nodes. Combining Figure 2 , as Figure 7 shown, the above step 202e can be specifically implemented through the following steps 202e1 to 202e3

[0204] Step 202e1: The electronic device extracts the main screen attention at each time point in the fusion prediction vector to obtain a main screen attention sequence

[0205] In the embodiments of the present application, the above main screen attention is the degree of attention of the user to the main screen application interface of the folding screen at a certain time point

[0206] In the embodiments of the present application, the above main screen attention sequence is a set of main screen attention values arranged in chronological order, reflecting the change in the user's attention to the main screen application interface at different time points

[0207] Optionally, in the embodiments of the present application, the electronic device may extract the main screen attention value from the vector at each time point of the fusion prediction vector and arrange the extracted main screen attention values in chronological order to form a main screen attention sequence, in the specific form of

[0208] Step 202e2: The electronic device extracts the attention degree of the secondary screen at each time point in the fusion prediction vector to obtain a secondary screen attention degree sequence.

[0209] In the embodiment of the present application, the above-mentioned attention degree of the secondary screen is the degree of attention of the user to the application interface of the secondary screen of the folding screen at a certain time point.

[0210] In the embodiment of the present application, the above-mentioned secondary screen attention degree sequence is a set of secondary screen attention degree values arranged in chronological order, reflecting the change of the user's attention degree to the secondary screen application interface at different time points.

[0211] Optionally, in the embodiment of the present application, the electronic device may extract the secondary screen attention degree value from the vector at each time point of the fusion prediction vector, and arrange the extracted secondary screen attention degree values in chronological order to form a secondary screen attention degree sequence, and the specific form is

[0212] Step 202e3: The electronic device extracts the switching probability at each time point in the fusion prediction vector, and based on the switching probability, determines at least one switching time node from the time points in the fusion prediction vector to obtain a switching time node set.

[0213] In the embodiment of the present application, each of the above-mentioned switching time nodes is a time node for switching the attention degree between the main screen and the secondary screen.

[0214] In the embodiment of the present application, the above-mentioned switching probability refers to the possibility that the user switches from one application interface to another application interface at a certain time point, and can be represented by a value between 0 and 1.

[0215] In the embodiment of the present application, the above-mentioned switching time node set is a series of time points determined according to the switching probability, and these time points represent the moments when the user may perform application interface switching. The specific form of the switching node set is

[0216] Optionally, in the embodiment of the present application, the electronic device may extract the switching probability from the vector at each time point of the fusion prediction vector, and then may compare it with each switching probability according to a preset threshold, for example, 0.5, and determine the time point with the switching probability greater than or equal to the threshold as the switching time node, and collect the determined switching time nodes into a set, which is the switching time node set.

[0217] Optionally, in the embodiment of the present application, the electronic device may index the time series dimension information in M fuse and combine the dual-screen domain features to obtain the main and secondary screen attention degree curves changing with time and the set of potential attention switching moments.

[0218] In this way, the electronic device can absorb the quick insights of the micro vertical large model into user operations during fusion, and also combine the iterative simulation ability of the information entropy-driven diffusion process for long-duration and high-parallel scenarios, thereby improving the accuracy of predicting critical switching nodes.

[0219] Optionally, in the embodiments of the present application, in combination with Figure 1 , such as Figure 8 shown, the above step 203 can be specifically implemented by the following steps 203a to 203c.

[0220] Step 203a: The electronic device determines the low-power priority and switching probability corresponding to each application interface based on the prediction result.

[0221] In the embodiments of the present application, the above low-power priority is the priority of the application interface marked as a low-power type interface, and the above switching probability is the probability of switching to the application interface.

[0222] Optionally, in the embodiments of the present application, the electronic device can analyze the current main and secondary screen attention distributions and the next switching time and probability according to the prediction result to obtain the low-power priority and switching probability, which are used as the basis for subsequent low-power scheduling. Among them:

[0223] (1) The low-power priority indicates that the application interface can be preferentially optimized for power consumption. For example, if a certain application interface detects low activity during the current period, it is given a label of "power saving first" or "low power first", restricting its operations such as UI updates, background animations, and component loading, and reducing the occupancy of the Central Processing Unit (CPU) and Graphics Processing Unit (GPU).

[0224] (2) The switching probability refers to the possibility that the user switches from the current application interface to another application interface at a certain time point. When the electronic device predicts that the user may perform a screen switch in a short time, the switching probability can be calculated. If the switching probability is high, the application interface in the target screen can be pre-loaded on the premise of keeping the power consumption controllable, otherwise no additional rendering synthesis will be performed, thus avoiding waste of electric energy.

[0225] Exemplarily, a mobile phone is used as an example of the electronic device for illustration. As Figure 9As shown in (a) of [description], assume that the mobile phone detects that the user switches to the application interface of social application A every time the game is loaded to 10% progress in the past 5 times. Then, when the mobile phone displays the game interface 10 and loads the game to 10% progress, the mobile phone predicts that the probability of the user switching to the application interface of social application A is extremely high. Thus, it can pre-render the chat list framework of social application A in advance and reserve 1 large core for social application A, thereby reducing the switching latency and reducing the utilization rate fluctuation of the mobile phone's GPU. According to the prediction result, pre-load the resources of social application A in the background, so that as Figure 9 As shown in (b) of [description], when the user switches to the application interface 11 of social application A, the electronic device can be opened in an extremely short time and receive messages before opening the application, so that the received messages can be immediately displayed while opening the application interface.

[0226] Optionally, in the embodiments of the present application, the electronic device can compare the attention degree in the attention degree sequence with a preset attention degree threshold. In the case where the attention degree is greater than or equal to the attention degree threshold, it is determined that the application interface is not a low-power type application interface. In the case where the attention degree is less than the attention degree threshold, it is determined that the application interface is a low-power type application interface, and according to the switching node set and the attention degree sequence, calculate the switching probability of each application interface.

[0227] Exemplarily, assume that the predicted result is the main screen attention degree sequence: [0.8, 0.6, 0.4], the secondary screen attention degree sequence: [0.2, 0.4, 0.6], the switching node set: [0, 2], and assume that the attention degree threshold is 0.5. Then the electronic device can determine:

[0228] The attention degree of the main screen at time point 0 is 0.8, which is greater than 0.5, and is not marked as a low-power type.

[0229] The attention degree of the main screen at time point 1 is 0.6, which is greater than 0.5, and is not marked as a low-power type.

[0230] The attention degree of the main screen at time point 2 is 0.4, which is less than 0.5, and is marked as a low-power type.

[0231] The attention degree of the secondary screen at time point 0 is 0.2, which is less than 0.5, and is marked as a low-power type.

[0232] The attention degree of the secondary screen at time point 1 is 0.4, which is less than 0.5, and is marked as a low-power type.

[0233] The attention degree of the secondary screen at time point 2 is 0.6, which is greater than 0.5, and is not marked as a low-power type.

[0234] Time point 0 is a switching node, and the switching probability is relatively high, for example, 0.7.

[0235] Time point 1 is not a switching node, and the switching probability is relatively low, such as 0.3.

[0236] Time point 2 is a switching node, and the switching probability is relatively high, such as 0.8.

[0237] Step 203b: The electronic device generates a low-power scheduling policy corresponding to each application interface based on the low-power priority and switching probability corresponding to each application interface.

[0238] In the embodiments of the present application, the above low-power scheduling policy is a resource scheduling and management plan formulated for each application interface according to the low-power priority and switching probability, and is used to minimize power consumption as much as possible on the premise of ensuring the user experience.

[0239] Optionally, in the embodiments of the present application, the electronic device integrates a set of low-power interaction scheduling policies, and continuously adaptively updates according to user operations and model predictions during system operation. The electronic device can set a "low-power scheduling management unit" at the operating system or application management service level through the policy decision and feedback module, periodically schedule the results of the fusion prediction, and evaluate the current main and secondary screen usage data and background task conditions.

[0240] Optionally, in the embodiments of the present application, if a form change or a user behavior indicating a tendency to turn off the screen is detected, such as folding to unfolding, or switching from the secondary screen to the main screen, the electronic device can lock the refresh rate, pause irrelevant animations, postpone some heavy computations, etc. As shown in (a) of Figure 10 For example, a certain folding screen mobile phone will turn off the screen when the included angle between the main screen and the secondary screen is less than 10°. After the mobile phone predicts the user's behavior, the mobile phone can predict that the user is about to close the mobile phone 1 second after the user closes all applications on the main screen and the secondary screen. As shown in (b) of Figure 10 The mobile phone can turn off the screen when the included angle between the main screen and the secondary screen is less than 150°, thereby saving the power consumption of the mobile phone.

[0241] Optionally, in the embodiments of the present application, the electronic device can maintain a basic match between the user's attention and the interface switching requirements through the "low-power interaction scheduling policy" in the folding screen dual-screen usage scenario, and finally significantly reduce the ineffective energy consumption of the background or weakly active screen, and improve the battery life of the electronic device.

[0242] Step 203c: The electronic device performs resource scheduling for each application interface based on the low-power scheduling policy corresponding to each application interface.

[0243] Optionally, in the embodiments of the present application, the electronic device can combine the low-power priority and the switching probability to perform targeted layout and rendering allocation on the dual-screen interface, so as to achieve more efficient energy management. For the screen with "low power consumption priority", such as the secondary screen that is not currently being focused on, the electronic device can reduce the refresh frame rate and the level of animation special effects, and reduce the CPU and GPU occupancy.

[0244] Optionally, in the embodiments of the present application, the electronic device can implement the delayed loading and lazy loading strategies. For the function modules that are not currently in the usage focus or have a very low switching probability, the lazy loading is adopted, and the initialization is only performed when the user may click next time. If the electronic device determines according to the switching probability that the next screen is very likely to be activated, it can perform a small amount of necessary preloading for its key user interface (UI) components, and strictly control the background parallelism to avoid sacrificing too much power consumption.

[0245] Exemplarily, as Figure 11 shown, assume that the user frequently operates the game in the game screen 10 displayed on the main screen of the mobile phone. After the mobile phone performs local event collection and model training prediction, it can reduce the refresh rate of the application interface 11 of the social application A on the secondary screen, and pause receiving the background messages of the social application A, thereby reducing the power consumption.

[0246] Optionally, in the embodiments of the present application, the electronic device can further reduce the power consumption when multiple applications are running in parallel on the folding screen. For the upcoming and possible screen switching, perform "minimal necessary" pre-layout, and for the tasks that are frequently used, reserve the large core for the predicted target application, and set the rendering queue priority through the Android SurfaceFlinger service.

[0247] Optionally, in the embodiments of the present application, the electronic device can follow the "minimalist" pre-layout principle, that is, when the switching probability exceeds the probability threshold, such as 0.5, only pre-load the key UI elements necessary for the next screen, such as the main menu, play controls, etc., without performing full-scale rendering, so as to avoid high-overhead operations such as the GPU or input / output (IO).

[0248] Optionally, in the embodiments of the present application, the electronic device can set a preloading time window. If a change in the user operation is detected within this time window, such as changing from switching to no longer switching, the electronic device can quickly recycle the loaded but unused resources.

[0249] Optionally, in the embodiments of the present application, the electronic device can dynamically shut down the background parallel tasks, and for the applications and processes that are not currently in the usage focus and have a low switching probability, reduce their network request frequency, thread priority or automatically put them into sleep mode, reducing the resource occupancy of the CPU and memory.

[0250] In this way, the electronic device can apply the prediction results to the resource scheduling and interactive layout in the dual-screen separation mode of the folding screen, so as to dynamically adjust the resource scheduling according to the prediction results of user behavior and application usage. It ensures that when multiple application interfaces are running simultaneously, each application interface can obtain a resource share that matches its importance and usage frequency, avoiding ineffective calculations and renderings, and improving the resource utilization efficiency.

[0251] In the embodiments of this application, the electronic device can achieve millisecond-level prediction of the user's attention switching behavior through a micro vertical large model, information entropy-driven diffusion, and fusion prediction dynamic behavior modeling, and accordingly pre-allocate computing resources intelligently. This reduces the switching latency, improves the prediction accuracy, and at the same time controls the additional power consumption through a hierarchical preloading mechanism, significantly enhancing the user experience and the battery life of the electronic device in the dual-screen separation mode.

[0252] Figure 12 It is a schematic diagram of the execution process of the resource scheduling method provided by the embodiments of this application. As Figure 12 shown, the resource scheduling method provided by the embodiments of this application may include the following steps 10 to step 15.

[0253] Step 10: The electronic device constructs a multi-source data collection model for the dual-screen separation interaction scenario of the folding screen.

[0254] Optionally, in the embodiments of this application, the above step 10 may be implemented through the following steps 10a to 10d.

[0255] Step 10a: The electronic device collects dual-screen interaction events and constructs an initial event sequence.

[0256] Step 10b: The electronic device labels the foreground and background states and generates a switching label sequence.

[0257] Step 10c: The electronic device collects and records the interface visibility and operation interval duration and constructs a visibility time series model.

[0258] Step 10d: The electronic device outputs the fused multi-source information.

[0259] Step 11: The electronic device performs scene-based feature extraction and lightweight encoding.

[0260] Optionally, in the embodiments of this application, the above step 11 may be implemented through the following steps 11a to 11c.

[0261] Step 11a: The electronic device constructs a scene mapping function and generates a preprocessed feature matrix.

[0262] Step 11b: The electronic device compresses and retains key information in the scene feature matrix using a lightweight encoding operator.

[0263] Step 11c: The electronic device outputs the scene-based lightweight encoding vector and improves the docking with the subsequent model.

[0264] Step 12: The electronic device constructs and preliminarily trains a micro vertical large model optimized based on the scene.

[0265] Optionally, in the embodiments of the present application, the above Step 12 can be implemented through the following Steps 12a to 12d.

[0266] Step 12a: The electronic device constructs a dual-domain morphological aggregation layer based on the scene-based lightweight feature vector and initializes the micro backbone network.

[0267] Step 12b: The electronic device establishes a cross-domain fusion mapping and realizes the dynamic screen migration characterization.

[0268] Step 12c: The electronic device uses a lightweight multi-task coupling module to enhance the adaptation to multiple application scenarios.

[0269] Step 12d: The electronic device performs preliminary training and outputs the attention state and potential switching timing prediction vector.

[0270] Step 13: The electronic device constructs an information entropy-driven diffusion model.

[0271] Optionally, in the embodiments of the present application, the above Step 13 can be implemented through the following Steps 13a to 13d.

[0272] Step 13a: The electronic device introduces an adaptive information entropy evaluation unit and obtains the dual-screen task hybrid complexity.

[0273] Step 13b: The electronic device schedules and dynamically controls the step size based on the information entropy-driven diffusion process.

[0274] Step 13c: The electronic device designs a multi-layer variable diffusion channel and combines the prediction vector gating to select the path.

[0275] Step 13d: The electronic device outputs the improved attention switching timing distribution.

[0276] Step 14: The electronic device fuses prediction, dynamic attention switching, and scene prediction.

[0277] Optionally, in the embodiments of the present application, the above Step 14 can be implemented through the following Steps 14a to 14c.

[0278] Step 14a: The electronic device establishes a coupling mechanism between the output of the micro vertical large model and the diffusion timing distribution.

[0279] Step 14b: The electronic device performs dynamic error correction and generates a fusion prediction vector.

[0280] Step 14c: The electronic device obtains the comprehensive scenario prediction and outputs the main and secondary screen attention curves and switching nodes.

[0281] Step 15: The electronic device performs interactive resource scheduling and management based on the comprehensive scenario prediction.

[0282] Optionally, in the embodiments of the present application, the above Step 15 can be implemented through the following Steps 15a to 15d.

[0283] Step 15a: The electronic device extracts the screen low-power priority and switching probability.

[0284] Step 15b: The electronic device determines the interactive layout strategy based on the low-power priority and switching prediction.

[0285] Step 15c: The electronic device performs pre-layout for the next switching and background parallel task control.

[0286] Step 15d: The electronic device outputs the low-power scheduling strategy and applies it to the actual system.

[0287] Each of the above method embodiments, or various possible implementation manners in each method embodiment, can be executed alone, or any two or more of them can be combined with each other, which can be specifically determined according to actual usage requirements. The embodiments of the present application do not limit this.

[0288] In the resource scheduling method provided by the embodiments of the present application, the execution subject can be a resource scheduling device. In the embodiments of the present application, taking the resource scheduling device executing the resource scheduling method as an example, the resource scheduling device provided by the embodiments of the present application is described.

[0289] Figure 13 Shows a possible structural schematic diagram of a resource scheduling device involved in some embodiments of the present application. As Figure 13 shown, the resource scheduling device 70 may include: an acquisition module 71 and an execution module 72.

[0290] Among them, the above acquisition module 71 is used to obtain an interface data set if an operation event on the application interface is detected when at least two application interfaces are displayed on the folding screen. The interface data set includes the operation information of the operation event and the application information corresponding to the operation event.

[0291] The above acquisition module 71 is used to obtain an interface prediction result based on the interface data set. The interface prediction result represents the attention degree to the application interface in the folding screen.

[0292] The above execution module 72 is used to perform resource scheduling on at least two application interfaces based on the interface prediction result obtained by the acquisition module 71.

[0293] In a possible implementation, the operation information of the above operation event includes the operation type of the operation event and the operation time of the operation event. The application information corresponding to the above operation event includes the identifier of the application interface corresponding to the operation event, the running state of the application corresponding to the operation event, and the visibility of the application interface corresponding to the operation event.

[0294] In a possible implementation, the above acquisition module 71 is specifically configured to: extract scene features and perform lightweight encoding on the interface data set to obtain a sequence of scene lightweight encoded vectors; process the sequence of scene lightweight encoded vectors to obtain a prediction vector; perform multi-layer variable diffusion on the prediction vector to obtain an attention switching time series distribution; couple the prediction vector and the attention switching time series distribution to obtain a fused prediction vector; and analyze the fused prediction vector to obtain a prediction result.

[0295] In a possible implementation, the above acquisition module 71 is specifically configured to: extract scene features from the interface data set through a scene mapping function to obtain a scene feature matrix; compress the scene feature matrix through a lightweight encoding function to obtain an encoded mapping result; and integrate the encoded mapping result to obtain a sequence of scene lightweight encoded vectors.

[0296] In a possible implementation, the above acquisition module 71 is specifically configured to: extract features from the sequence of scene lightweight encoded vectors through a first form aggregation function to obtain a first feature vector, where the first feature vector represents the usage of the main screen of the folding screen; extract features from the sequence of scene lightweight encoded vectors through a second form aggregation function to obtain a second feature vector, where the second feature vector represents the usage of the secondary screen of the folding screen; fuse the first feature vector and the second feature vector through a micro network function to obtain a representation vector; fuse the representation vector and an application type embedding vector through a cross-domain fusion mapping function to obtain a third feature vector; and transform the third feature vector through an output layer mapping function to obtain a prediction vector.

[0297] In a possible implementation, the above acquisition module 71 is specifically configured to: calculate the information entropy of the prediction vector through an information entropy evaluation function to obtain the information entropy; perform iteration on the information entropy through a diffusion iteration function to obtain a diffusion state vector, where the number of iterations and the diffusion step of the iteration function are determined according to the magnitude of the information entropy; and analyze the diffusion state vector to obtain an attention switching time series distribution.

[0298] In a possible implementation manner, the above-mentioned obtaining module 71 is specifically configured to: input the prediction vector and the attention switching time sequence distribution into a coupling mapping function, and through the coupling mapping function, perform structured alignment on the prediction vector and the attention switching time sequence distribution to obtain a fusion vector; input the information entropy into a weight adjustment function, and calculate to obtain a fusion weight; weight the fusion vector according to the fusion weight to obtain a fusion prediction vector.

[0299] In a possible implementation manner, the above-mentioned prediction result includes a main screen attention degree sequence, a secondary screen attention degree sequence, and a switching node set. The obtaining module 71 is specifically configured to: extract the main screen attention degree at each time point in the fusion prediction vector to obtain the main screen attention degree sequence; extract the secondary screen attention degree at each time point in the fusion prediction vector to obtain the secondary screen attention degree sequence; extract the switching probability at each time point in the fusion prediction vector, and based on the switching probability, determine at least one switching time node from the time points in the fusion prediction vector to obtain a switching time node set, and each switching time node is a time node for switching the attention degrees of the main screen and the secondary screen.

[0300] In a possible implementation manner, the above-mentioned execution module 72 is specifically configured to: based on the prediction result obtained by the obtaining module 71, determine the low-power priority and the switching probability corresponding to each application interface, where the low-power priority is the priority of the application interface marked as a low-power type interface, and the switching probability is the probability of switching to the application interface; generate a low-power scheduling policy corresponding to each application interface based on the low-power priority and the switching probability corresponding to each application interface; and allocate resources for each application interface based on the low-power scheduling policy corresponding to each application interface.

[0301] In the embodiment of the present application, a resource scheduling device is provided. Since the resource scheduling device can detect the operation events of the user on the application interfaces in the folding screen to obtain the operation conditions of the application interfaces and the information of the applications corresponding to the operations, and then the resource scheduling device can analyze and process this information to obtain the attention degrees of the user on the application interfaces in the folding screen, and thus allocate resources for the multiple displayed application interfaces based on the interface prediction result, that is, this solution can dynamically adjust the resource scheduling according to the attention degrees of the user on the application interfaces. Therefore, sufficient resources can be allocated for the application interfaces concerned by the user, avoiding wasting resources by allocating too many resources for the application interfaces not concerned by the user, improving the flexibility of allocating resources for the application interfaces, and thus ensuring the smoothness of the application operation.

[0302] The resource scheduling device in the embodiments of the present application may be an electronic device or a component in an electronic device, such as an integrated circuit or a chip. The electronic device may be a terminal or other devices other than terminals. Exemplarily, the electronic device may be a mobile phone, a tablet computer, a laptop computer, a palmtop computer, a vehicle-mounted electronic device, a Mobile Internet Device (MID), an augmented reality (AR) / virtual reality (VR) device, a robot, a wearable device, an ultra-mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc. It may also be a server, a Network Attached Storage (NAS), a personal computer (PC), a television (TV), a teller machine, or a self-service machine, etc. The embodiments of the present application do not make specific limitations.

[0303] The resource scheduling device in the embodiments of the present application may be a device with an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems. The embodiments of the present application do not make specific limitations.

[0304] The resource scheduling device provided in the embodiments of the present application can implement each process implemented in the above method embodiments. To avoid repetition, it will not be elaborated here.

[0305] Optionally, as Figure 14 shown, the embodiments of the present application further provide an electronic device 1000, including a processor 1001 and a memory 1002. A program or instruction that can run on the processor 1001 is stored on the memory 1002. When the program or instruction is executed by the processor 1001, it implements each step of the above resource scheduling method embodiments and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0306] It should be noted that the electronic devices in the embodiments of the present application include the above-mentioned mobile electronic devices and non-mobile electronic devices.

[0307] Figure 15 It is a schematic diagram of the hardware structure of an electronic device for implementing the embodiments of the present application.

[0308] The electronic device 100 includes, but is not limited to, components such as a radio frequency unit 101, a network module 102, an audio output unit 103, an input unit 104, a sensor 105, a display unit 106, a user input unit 107, an interface unit 108, a memory 109, and a processor 110.

[0309] Those skilled in the art can understand that the electronic device 100 may further include a power source (such as a battery) for supplying power to each component. The power source can be logically connected to the processor 110 through a power management system, so as to manage functions such as charging, discharging, and power consumption management through the power management system. Figure 15 The structure of the electronic device shown does not limit the electronic device. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.

[0310] Among them, the above-mentioned processor 110 is used to, when at least two application interfaces are displayed on the folding screen, if an operation event on the application interface is detected, obtain an interface data set, and the interface data set includes the operation information of the operation event and the application information corresponding to the operation event.

[0311] The above-mentioned processor 110 is used to obtain an interface prediction result based on the interface data set, and the interface prediction result characterizes the attention degree to the application interface in the folding screen.

[0312] The above-mentioned processor 110 is used to perform resource scheduling on at least two application interfaces based on the interface prediction result.

[0313] In a possible implementation manner, the operation information of the above-mentioned operation event includes the operation type of the operation event and the operation time of the operation event. The application information corresponding to the operation event includes the identifier of the application interface corresponding to the operation event, the running state of the application corresponding to the operation event, and the visibility of the application interface corresponding to the operation event.

[0314] In a possible implementation manner, the above-mentioned processor 110 is specifically used to: extract scene-based features and perform lightweight encoding on the interface data set to obtain a sequence of scene-based lightweight encoding vectors; process the sequence of scene-based lightweight encoding vectors to obtain a prediction vector; perform multi-layer variable diffusion on the prediction vector to obtain an attention switching time sequence distribution; couple the prediction vector and the attention switching time sequence distribution to obtain a fused prediction vector; and analyze the fused prediction vector to obtain a prediction result.

[0315] In a possible implementation, the above-mentioned processor 110 is specifically configured to: extract scene features from the interface data set through a scene mapping function to obtain a scene feature matrix; compress the scene feature matrix through a lightweight encoding function to obtain an encoded mapping result; and integrate the encoded mapping result to obtain a sequence of scene-based lightweight encoding vectors.

[0316] In a possible implementation, the above-mentioned processor 110 is specifically configured to: extract features from the sequence of scene-based lightweight encoding vectors through a first form aggregation function to obtain a first feature vector, which characterizes the usage of the main screen of the folding screen; extract features from the sequence of scene-based lightweight encoding vectors through a second form aggregation function to obtain a second feature vector, which characterizes the usage of the secondary screen of the folding screen; fuse the first feature vector and the second feature vector through a micro-network function to obtain a characterization vector; fuse the characterization vector and the application type embedding vector through a cross-domain fusion mapping function to obtain a third feature vector; and transform the third feature vector through an output layer mapping function to obtain a prediction vector.

[0317] In a possible implementation, the above-mentioned processor 110 is specifically configured to: calculate the information entropy of the prediction vector through an information entropy evaluation function to obtain the information entropy; iterate the information entropy through a diffusion iteration function to obtain a diffusion state vector, where the number of iterations and the diffusion step of the iteration function are determined according to the magnitude of the information entropy; and analyze the diffusion state vector to obtain the attention switching time sequence distribution.

[0318] In a possible implementation, the above-mentioned processor 110 is specifically configured to: input the prediction vector and the attention switching time sequence distribution into a coupling mapping function, and through the coupling mapping function, structurally align the prediction vector and the attention switching time sequence distribution to obtain a fusion vector; input the information entropy into a weight adjustment function to calculate a fusion weight; and weight the fusion vector according to the fusion weight to obtain a fusion prediction vector.

[0319] In a possible implementation, the above-mentioned prediction result includes a main screen attention sequence, a secondary screen attention sequence, and a switching node set. The above-mentioned processor 110 is specifically configured to: extract the main screen attention at each time point in the fusion prediction vector to obtain the main screen attention sequence; extract the secondary screen attention at each time point in the fusion prediction vector to obtain the secondary screen attention sequence; extract the switching probability at each time point in the fusion prediction vector, and based on the switching probability, determine at least one switching time node from the time points in the fusion prediction vector to obtain the switching time node set, where each switching time node is a time node for switching the attention to the main screen and the secondary screen.

[0320] In a possible implementation, the above-mentioned processor 110 is specifically configured to: based on the prediction result, determine the low-power priority and switching probability corresponding to each application interface, where the low-power priority is the priority of the application interface marked as a low-power type interface, and the switching probability is the probability of switching to the application interface; based on the low-power priority and switching probability corresponding to each application interface, generate a low-power scheduling policy corresponding to each application interface; and based on the low-power scheduling policy corresponding to each application interface, allocate resources to each application interface.

[0321] The embodiment of the present application provides an electronic device. Since the electronic device can detect the operation events of the user on the application interface in the folding screen to obtain the operation situation of the application interface and the information of the application corresponding to the operation, and then the electronic device can analyze and process this information to obtain the attention degree of the user to the application interface in the folding screen. Thus, based on the interface prediction result, resources are allocated to multiple displayed application interfaces. That is, this solution can dynamically adjust resource scheduling according to the attention degree of the user to the application interface. Therefore, sufficient resources can be allocated to the application interfaces that the user pays attention to, avoiding wasting resources by allocating too many resources to the application interfaces that the user does not pay attention to, improving the flexibility of resource allocation for application interfaces, and thus ensuring the smoothness of application operation.

[0322] The electronic device provided by the embodiment of the present application can implement each process implemented by the above method embodiment and achieve the same technical effect. To avoid repetition, it will not be elaborated here. The beneficial effects of various implementation manners in this embodiment can be specifically referred to the beneficial effects of the corresponding implementation manners in the above method embodiment. To avoid repetition, it will not be elaborated here.

[0323] It should be understood that in the embodiment of the present application, the input unit 104 may include a Graphics Processing Unit (GPU) 1041 and a microphone 1042. The graphics processor 1041 processes the image data of static pictures or videos obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The display unit 106 may include a display panel 1061, and the display panel 1061 may be configured in the form of a liquid crystal display, an organic light-emitting diode, etc. The user input unit 107 includes at least one of a touch panel 1071 and other input devices 1072. The touch panel 1071 is also called a touch screen. The touch panel 1071 may include two parts: a touch detection device and a touch controller. The other input devices 1072 may include, but are not limited to, a physical keyboard, function keys (such as volume control keys, power on / off keys, etc.), a trackball, a mouse, and a joystick, which will not be elaborated here.

[0324] The memory 109 can be used to store software programs and various data. The memory 109 may mainly include a first storage area for storing programs or instructions and a second storage area for storing data. Among them, the first storage area may store an operating system, application programs or instructions required for at least one function (such as a sound playback function, an image playback function, etc.). In addition, the memory 109 may include a volatile memory or a non-volatile memory, or the memory 109 may include both a volatile memory and a non-volatile memory. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous dynamic random access memory (SDRAM), a double data rate synchronous dynamic random access memory (DDR SDRAM), an enhanced synchronous dynamic random access memory (ESDRAM), a synch link dynamic random access memory (SLDRAM), and a direct rambus random access memory (DRRAM). The memory 109 in the embodiments of the present application includes but is not limited to these and any other suitable types of memories.

[0325] The processor 110 may include one or more processing units; optionally, the processor 110 integrates an application processor and a modem processor. Among them, the application processor mainly processes operations related to the operating system, user interface, and application programs, etc., and the modem processor mainly processes wireless communication signals, such as a baseband processor. It can be understood that the above modem processor may not be integrated into the processor 110 either.

[0326] The embodiments of the present application also provide a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, it implements each process of the above resource scheduling method embodiment and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0327] Among them, the processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media such as computer read-only memory ROM, random access memory RAM, magnetic disks, or optical discs.

[0328] Another embodiment of the present application provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement each process of the above embodiment of the resource scheduling method and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0329] It should be understood that the chip mentioned in the embodiments of the present application may also be referred to as a system-on-chip, system chip, chip system, or system-on-chip.

[0330] The embodiments of the present application provide a computer program product. The program product is stored in a storage medium and is executed by at least one processor to implement each process of the above embodiment of the resource scheduling method and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0331] It should be noted that in this article, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements not only includes those elements but also includes other elements not explicitly listed, or further includes elements inherent to such a process, method, article, or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article, or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the reverse order according to the functions involved. For example, the described methods may be performed in an order different from that described, and various steps may be added, omitted, or combined. Additionally, the features described with reference to certain examples may be combined in other examples.

[0332] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described example methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present application, in essence or the part that contributes to the prior art, can be embodied in the form of a computer software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present application.

[0333] The embodiments of the present application have been described above in conjunction with the accompanying drawings. However, the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims, and all of them fall within the protection scope of the present application.

Claims

1. A resource scheduling method, characterized in that: Applied to an electronic device including a folding screen, the method comprises: In the case where the folding screen displays at least two application interfaces, if an operation event on an application interface is detected, acquiring an interface data set, the interface data set including operation information of the operation event and application information corresponding to the operation event; Based on the interface data set, obtaining an interface prediction result, wherein the interface prediction result represents the degree of attention paid to the application interface in the folding screen; Based on the interface prediction result, resources are scheduled for the at least two application interfaces.

2. The method according to claim 1, characterized in that The operation information of the operation event includes the operation type of the operation event and the operation time of the operation event; The application information corresponding to the operation event includes an identifier of the application interface corresponding to the operation event, a running state of the application corresponding to the operation event, and visibility of the application interface corresponding to the operation event.

3. The method according to claim 1 or 2, characterized in that: The obtaining of the interface prediction result based on the interface data set includes: Performing scenario-based feature extraction and lightweight encoding on the interface data set to obtain a scenario-based lightweight encoding vector sequence; Processing the scene-based lightweight coding vector sequence to obtain a prediction vector; Performing multi-layer variable diffusion on the prediction vector to obtain the attention switching time series distribution; coupling the prediction vector and the attention switching timing distribution to obtain a fused prediction vector; The fused prediction vector is parsed to obtain the prediction result.

4. The method according to claim 3, characterized in that The performing scene-based feature extraction and lightweight encoding on the interface data set to obtain a scene-based lightweight encoding vector sequence includes: By using a scene mapping function, scene feature extraction is performed on the interface data set to obtain a scene feature matrix; The scene feature matrix is ​​compressed by a lightweight coding function to obtain a coding mapping result; The encoding mapping results are integrated to obtain the scenario-based lightweight encoding vector sequence.

5. The method according to claim 3, characterized in that: The step of processing the scene-based lightweight coding vector sequence to obtain a prediction vector includes: Performing feature extraction on the scene-based lightweight coding vector sequence through a first morphological aggregation function to obtain a first feature vector, where the first feature vector represents usage of the main screen of the folding screen; Performing feature extraction on the scene-based lightweight coding vector sequence through a second morphological aggregation function to obtain a second feature vector, where the second feature vector represents usage of the auxiliary screen of the folding screen; By using a micro-network function, the first feature vector and the second feature vector are fused to obtain a representation vector; The representation vector is fused with the application type embedding vector through a cross-domain fusion mapping function to obtain a third feature vector; The third eigenvector is transformed through an output layer mapping function to obtain the prediction vector.

6. The method according to claim 3, characterized in that: The step of performing multi-layer variable diffusion on the prediction vector to obtain the attention switching timing distribution includes: By using an information entropy evaluation function, information entropy is calculated for the prediction vector to obtain information entropy; The information entropy is iterated by a diffusion iteration function to obtain a diffusion state vector, wherein the number of iterations and the diffusion stride of the iterative function are determined according to the size of the information entropy; The diffusion state vector is analyzed to obtain the attention switching timing distribution.

7. The method according to claim 3, characterized in that The coupling of the prediction vector and the attention switching timing distribution to obtain a fused prediction vector includes: Inputting the prediction vector and the attention switching timing distribution into a coupling mapping function, and performing structured alignment on the prediction vector and the attention switching timing distribution through the coupling mapping function to obtain a fusion vector; Input the information entropy into a weight adjustment function to calculate a fusion weight; The fusion vector is weighted according to the fusion weight to obtain the fusion prediction vector.

8. The method according to claim 3, characterized in that The prediction results include a main screen attention sequence, an auxiliary screen attention sequence and a switch node set; The step of parsing the fused prediction vector to obtain the prediction result includes: Extracting the main screen attention degree at each time point in the fused prediction vector to obtain a main screen attention degree sequence; Extracting the auxiliary screen attention degree at each time point in the fused prediction vector to obtain an auxiliary screen attention degree sequence; The switching probability of each time point in the fused prediction vector is extracted, and based on the switching probability, at least one switching time node is determined from the time points in the fused prediction vector to obtain a set of switching time nodes, each switching time node being a time node for switching the attention to the main screen and the auxiliary screen.

9. The method according to any one of claims 1 to 8, characterized in that The allocating resources to the at least two application interfaces based on the interface prediction result includes: Based on the prediction result, determine the low power consumption priority and switching probability corresponding to each application interface, wherein the low power consumption priority is the priority of the application interface marked as a low power consumption type interface, and the switching probability is the probability of switching to the application interface; Based on the low power priority and switching probability corresponding to each application interface, a low power scheduling strategy corresponding to each application interface is generated; Based on the low power scheduling strategy corresponding to each application interface, resources are scheduled for each application interface.

10. A resource scheduling device, characterized in that: Applied to an electronic device including a folding screen, the device includes: an acquisition module and an execution module; The acquisition module is configured to acquire an interface data set when an operation event on an application interface is detected when the folding screen displays at least two application interfaces, wherein the interface data set includes operation information of the operation event and application information corresponding to the operation event; The acquisition module is used to acquire an interface prediction result based on the interface data set, wherein the interface prediction result represents the attention paid to the application interface in the folding screen; The execution module is used to schedule resources for the at least two application interfaces based on the interface prediction result obtained by the acquisition module.

11. The device according to claim 10, characterized in that The acquisition module is specifically used for: Performing scenario-based feature extraction and lightweight encoding on the interface data set to obtain a scenario-based lightweight encoding vector sequence; Processing the scene-based lightweight coding vector sequence to obtain a prediction vector; Performing multi-layer variable diffusion on the prediction vector to obtain the attention switching time series distribution; coupling the prediction vector and the attention switching timing distribution to obtain a fused prediction vector; The fused prediction vector is parsed to obtain the prediction result.

12. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores programs or instructions that can be run on the processor, and when the programs or instructions are executed by the processor, the steps of the resource scheduling method according to any one of claims 1 to 9 are implemented.

13. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by a processor, the steps of the resource scheduling method according to any one of claims 1 to 9 are implemented.

Images