WO2019062342A1 - 后台应用清理方法、装置、存储介质及电子设备 - Google Patents

后台应用清理方法、装置、存储介质及电子设备 Download PDF

Info

Publication number
WO2019062342A1
WO2019062342A1 PCT/CN2018/099364 CN2018099364W WO2019062342A1 WO 2019062342 A1 WO2019062342 A1 WO 2019062342A1 CN 2018099364 W CN2018099364 W CN 2018099364W WO 2019062342 A1 WO2019062342 A1 WO 2019062342A1
Authority
WO
WIPO (PCT)
Prior art keywords
application
feature information
sample
training
cleanable
Prior art date
Application number
PCT/CN2018/099364
Other languages
English (en)
French (fr)
Other versions
WO2019062342A9 (zh
Inventor
曾元清
Original Assignee
Oppo广东移动通信有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Oppo广东移动通信有限公司 filed Critical Oppo广东移动通信有限公司
Priority to EP18862386.2A priority Critical patent/EP3702912A4/en
Publication of WO2019062342A1 publication Critical patent/WO2019062342A1/zh
Publication of WO2019062342A9 publication Critical patent/WO2019062342A9/zh
Priority to US16/819,777 priority patent/US11544633B2/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/445—Program loading or initiating
    • G06F9/44594—Unloading
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/485—Task life-cycle, e.g. stopping, restarting, resuming execution
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02—Addressing or allocation; Relocation
    • G06F12/0223—User address space allocation, e.g. contiguous or non contiguous base addressing
    • G06F12/023—Free address space management
    • G06F12/0253—Garbage collection, i.e. reclamation of unreferenced memory
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the present application relates to the field of communications technologies, and in particular, to a background application cleaning method, apparatus, storage medium, and electronic device.
  • the system of the electronic device supports multiple applications running at the same time, that is, one application runs in the foreground, and other applications can run in the background. If the application running in the background is not cleaned for a long time, the available memory of the electronic device becomes smaller, and the central processing unit (CPU) usage rate is too high, causing the electronic device to run slower, jamming, and power consumption. Too fast and other issues. Therefore, it is necessary to provide a method to solve the above problems.
  • CPU central processing unit
  • the embodiment of the present application provides a background application cleaning method, device, storage medium, and electronic device, which can improve the running fluency of the electronic device and reduce power consumption.
  • the background application cleaning method provided by the embodiment of the present application includes:
  • the generated plurality of decision trees are used to predict current feature information of the application, and output multiple prediction results, where the prediction results include cleanable and non-cleanable;
  • the background application cleaning device provided by the embodiment of the present application includes:
  • a collecting unit configured to collect multi-dimensional feature information of the application as a sample, and construct a sample set of the application
  • An extracting unit configured to extract feature information from the sample set according to a preset rule, and construct a plurality of training sets
  • a training unit configured to train each training set to generate a corresponding decision tree
  • a prediction unit configured to: when the application enters the background, use the generated multiple decision trees to predict current feature information of the application, and output multiple prediction results, where the prediction results include cleanable and non-cleanable;
  • the first determining unit is configured to determine, according to the multiple prediction results, whether the application can be cleaned up.
  • the storage medium provided by the embodiment of the present application has a computer program stored thereon, and when the computer program runs on the computer, the computer is executed to perform background application cleaning according to the first aspect of the embodiment of the present application. method.
  • an electronic device provided by an embodiment of the present application includes a processor and a memory, where the memory has a computer program, and the processor is configured to execute the first embodiment of the present application by calling the computer program.
  • FIG. 1 is a schematic diagram of an application scenario of a background application cleaning method according to an embodiment of the present disclosure.
  • FIG. 2 is a schematic flowchart of a background application cleaning method provided by an embodiment of the present application.
  • FIG. 3 is a schematic flowchart of a method for generating a decision tree according to an embodiment of the present application.
  • FIG. 4a is another schematic flowchart of a background application cleaning method provided by an embodiment of the present application.
  • FIG. 4b is another schematic flowchart of a background application cleaning method provided by an embodiment of the present application.
  • FIG. 4c is a schematic structural diagram of a decision tree generated by an embodiment of the present application.
  • FIG. 5 is a schematic structural diagram of a background application cleaning apparatus according to an embodiment of the present application.
  • FIG. 6 is another schematic structural diagram of a background application cleaning apparatus according to an embodiment of the present application.
  • FIG. 7 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
  • FIG. 8 is another schematic structural diagram of an electronic device according to an embodiment of the present application.
  • module as used herein may be taken to mean a software object that is executed on the computing system.
  • the different components, modules, engines, and services described herein can be considered as implementation objects on the computing system.
  • the apparatus and method described herein may be implemented in software, and may of course be implemented in hardware, all of which are within the scope of the present application.
  • references to "an embodiment” herein mean that a particular feature, structure, or characteristic described in connection with the embodiments can be included in at least one embodiment of the present application.
  • the appearances of the phrases in various places in the specification are not necessarily referring to the same embodiments, and are not exclusive or alternative embodiments that are mutually exclusive. Those skilled in the art will understand and implicitly understand that the embodiments described herein can be combined with other embodiments.
  • the embodiment of the present application provides a background application cleaning method, including the following steps:
  • the generated plurality of decision trees are used to predict current feature information of the application, and output multiple prediction results, where the prediction results include cleanable and non-cleanable;
  • the training is performed on each training set to generate a corresponding decision tree, including:
  • the feature information with the largest information gain is used as the feature information of the root node, and the remaining feature information is sequentially used as the feature information of the leaf node according to the information gain from the largest to the smallest, and a corresponding decision tree is generated.
  • the calculating the information gain of each feature information included in each training set includes:
  • Information gain, H(S) represents the entropy of the training set S, and H(S
  • p i represents the probability that the prediction result of the i-th category appears in the training set S
  • n represents the prediction result category
  • the extracting feature information from the sample set according to a preset rule, and constructing a plurality of training sets including:
  • a preset number of feature information is randomly extracted to form a corresponding sub-sample, and a plurality of sub-samples constitute a training set, and after multiple extractions, multiple trainings are constructed. set.
  • the multi-dimensional feature information of the application includes Q pieces of feature information, and the preset number is q, and the method further includes:
  • the method further includes:
  • sample labels including cleanable and non-cleanable
  • the determining, according to the multiple prediction results, whether the application can be cleaned includes:
  • the statistics can be cleaned up.
  • the proportion of this prediction result in all prediction results is recorded as the cleanable probability;
  • the proportion of the statistical result that cannot be cleaned up in all the prediction results is recorded as the maintenance probability;
  • the cleanable probability of the application is greater than the hold probability, it is determined that the application may be cleaned, and if the hold probability of the application is greater than the cleanable probability, it is determined to keep the state of the application running in the background unchanged.
  • the method when there are multiple applications in the background, the method further includes:
  • the multi-dimensional feature information of the application is collected as a sample, and the sample set of the application is constructed, including:
  • the multi-dimensional feature information of the application is collected once every preset time period, where the multi-dimensional feature information of the application includes running feature information of the application and/or state feature information of the electronic device;
  • a plurality of the samples are acquired within a preset historical time period, and the sample set is constructed.
  • the method further includes:
  • the background application cleaning method provided by the embodiment of the present application may be the background application cleaning device provided by the embodiment of the present application or the electronic device integrated with the background application cleaning device, wherein the background application cleaning device may adopt hardware or software.
  • the electronic device may be a device such as a smart phone, a tablet computer, a palmtop computer, a notebook computer, or a desktop computer.
  • FIG. 1 is a schematic diagram of an application scenario of a background application cleaning method according to an embodiment of the present disclosure.
  • the background device cleaning device is an electronic device, and the electronic device can collect multi-dimensional feature information of the application as a sample to construct the application. a sample set; extracting feature information from the sample set according to a preset rule, constructing a plurality of training sets; training each training set to generate a corresponding decision tree; and using the generated multiple trees when the application enters the background
  • the decision tree predicts current feature information of the application, and outputs a plurality of prediction results, the prediction result includes cleanable and non-cleanable; and determining whether the application can be cleaned according to the multiple prediction results.
  • the multi-dimensional feature information of the mailbox application may be collected according to a preset frequency in a historical time period (for example, the length of time that the mailbox enters the background every day, As a sample, the number of mailboxes entering the foreground is taken as a sample, the sample set of the mailbox application is constructed, and the feature information in the sample set is extracted to construct a plurality of training sets, and each training set is trained to generate a corresponding decision tree, for example, as shown in FIG.
  • the embodiment of the present application will provide a background application cleaning method according to an embodiment of the present application, and the background application cleaning device may be specifically integrated in an electronic device.
  • the background application cleaning method comprises: collecting multi-dimensional feature information of the application as a sample, constructing a sample set of the application; extracting feature information from the sample set according to a preset rule, constructing a plurality of training sets; and training each training set Generating a corresponding decision tree; when the application enters the background, predicting current feature information of the application by using the generated multiple decision trees, and outputting multiple prediction results, where the prediction results include cleanable and non-cleanable Determining whether the application can be cleaned based on the plurality of prediction results.
  • a background application cleanup method is provided. As shown in FIG. 2, the specific process of the background application cleanup method provided in this embodiment may be as follows:
  • Step S201 Collect multi-dimensional feature information of the application as a sample, and construct a sample set of the application.
  • the application mentioned in this embodiment may be any application installed on the electronic device, such as an office application, a communication application, a game application, a shopping application, and the like.
  • the multi-dimensional feature information of the application has a dimension of a certain length, and the parameters in each dimension correspond to a feature information of the application, that is, the multi-dimensional feature information is composed of a plurality of feature information.
  • the plurality of feature information may include feature information related to the application itself, that is, running feature information of the application, for example, the duration of the application cutting into the background; the duration of the electronic device being cut off during the background cutting; the number of times the application enters the foreground; The time in the foreground; the way the application enters the background, such as being switched by the home button (home button), being switched back by the return button, being switched in by other applications, etc.; the type of application, including level one (common application), level two ( Other applications) and so on.
  • the plurality of feature information may further include related feature information of the electronic device where the application is located, that is, state feature information of the electronic device, for example, an off time of the electronic device, a bright time, a current power, and a wireless network connection state of the electronic device. Whether the electronic device is in a charging state or the like.
  • the sample set of the application may include multiple samples collected at a preset frequency during the historical time period.
  • the historical time period may be, for example, the past 7 days or 10 days; the preset frequency may be, for example, collected every 10 minutes and collected every half hour. It can be understood that the multi-dimensional feature data of the application acquired at one time constitutes one sample, and a plurality of samples constitute the sample set.
  • each sample in the sample set can be marked to obtain a sample label for each sample. Since the implementation of the present invention is to predict whether the application can be cleaned, the labeled sample label includes "cleanable”. And "cannot be cleaned up”. Specifically, the user may mark the historical usage habits of the application. For example, after the application enters the background for 30 minutes, the user closes the application and is marked as “cleanable”; for example, after the application enters the background for 3 minutes, the user will The application is switched to the foreground and is marked as “not cleanable”. Specifically, the value “1" can be used to indicate “can be cleaned up”, the value "0" is used to mean “not cleanable", and vice versa.
  • Step S202 Extract feature information from the sample set according to a preset rule, and construct a plurality of training sets.
  • a preset number of feature information may be randomly extracted from the multi-dimensional feature information of each sample to form a corresponding sub-sample, and multiple sub-samples constitute a training set, and multiple extractions are performed to construct multiple The training set, the preset number can be customized according to actual needs.
  • the feature information in different training sets can be repeated, and the feature information in the same training set can be repeated, which can effectively prevent the training result from being over-fitting.
  • the training set is only composed of extracting part of the feature information from each sample of the sample set, the number of subsamples in the training set is the same as the number of samples in the sample set.
  • the sample set includes 100 samples, each sample includes 15 feature information, and 5 feature information is randomly extracted from each sample of the sample set to form corresponding sub-samples, which can constitute 100 sub-samples, and each sub-sample includes Randomly extracted 5 feature information, 100 subsamples constitute a training set.
  • the sample label of each sub-sample may be marked as the sample label of the corresponding sample.
  • the sample label of sample 1 is "cleanable”
  • the sample label of the subsample composed of sample 1 is also marked as "cleanable”.
  • the multi-dimensional feature information of the application may include Q feature information, and the preset number may be q, and may be according to a formula. Determine the number of training sets built, and M denotes the number of training sets.
  • q feature information is randomly extracted from the Q feature information of each sample, including Combination.
  • the number of training sets ie, the number of random extractions
  • the number of training sets is determined as It can reduce the amount of calculation, ensure the number of decision trees, and improve the accuracy of prediction.
  • Step S203 Training each training set to generate a corresponding decision tree.
  • Step S2031 calculating an information gain of each feature information included in each training set.
  • p i represents the probability that the prediction result of the i-th category appears in the training set S
  • n represents the prediction result category.
  • the prediction result includes two types, which can be cleaned and cannot be cleaned, so n takes a value of 2.
  • the training set S there are 10 samples (that is, multi-dimensional feature data of 10 applications), wherein the sample number of the sample label is "cleanable", and the number of samples with the sample label of "non-cleanable" is 3 , the entropy of the training set S
  • the training set S including 10 samples, when the training set S is classified by the feature of “type of application”, when the application type is one level, the sample labels of four samples are “not cleanable”, and The sample labels of the three samples are “cleanable”; when the application type is two, the sample labels of the remaining three samples are “cleanable”, and the entropy after the training set S is divided by the feature “type of application”
  • step S2032 the feature information with the largest information gain is used as the feature information of the root node, and the remaining feature information is sequentially used as the feature information of the leaf node in descending order of the information gain to generate a corresponding decision tree.
  • each training set is trained to obtain a number of decision trees equal to the number of training sets, for example, when the number of training sets is When you get Decision tree.
  • the generated multiple decision trees constitute a random forest.
  • the constructed random forest is used to predict the application.
  • the above steps S201-S203 may be repeated to construct a corresponding random forest for different applications.
  • the process of constructing the random forest may be completed in the server in advance.
  • the electronic device may send sample data of each application to the server, and the server processes the sample data of each application to generate corresponding applications.
  • the electronic device can obtain a corresponding random forest from the server, and use the obtained random forest to predict the corresponding application.
  • Step S204 When the application enters the background, predicting current feature information of the application by using the generated multiple decision trees, and outputting multiple prediction results, where the prediction results include cleanable and non-cleanable.
  • the current feature information of the application has the same dimension as the multi-dimensional feature information of the application collected when the sample is composed, and the corresponding parameter values of the two may be the same or different in each dimension.
  • the process of predicting current feature information of the application by using any one decision tree includes:
  • the condition for stopping the traversal for example, whether the application can be cleaned up explicitly
  • Each decision tree outputs a prediction result for the application.
  • Decision tree can output Forecast results.
  • Step S205 Determine, according to the multiple prediction results, whether the application can be cleaned up.
  • the proportion of the “cleanable” prediction result in all prediction results can be counted as the cleanable probability; the statistical “non-cleanable” prediction result is included in all prediction results.
  • the proportion of the record is recorded as the probability of maintaining;
  • the cleanable probability of the application is greater than the hold probability, it is determined that the application may be cleaned, and if the hold probability of the application is greater than the cleanable probability, it is determined to keep the state of the application running in the background unchanged.
  • the forest forest corresponding to each application may be used to predict the corresponding application, and each application may be cleaned according to the prediction result; and the cleanable application may be cleaned.
  • Probability according to the order in which the cleanable probability can be cleaned up, select a preset number of applications for cleaning, or select an application that can clear the probability greater than the preset probability for cleaning.
  • the application by collecting the multi-dimensional feature information of the application as a sample, constructing a sample set of the application, extracting feature information from the sample set according to a preset rule, constructing a plurality of training sets, and training each training set, Generating a corresponding decision tree, when the application enters the background, using the generated multiple decision trees to predict the current feature information of the application, and outputting multiple prediction results, and determining whether the cleaning can be cleaned according to the multiple prediction results.
  • the application cleans up the application that can be cleaned, thereby realizing the automatic cleaning of the background application, improving the running fluency of the electronic device and reducing the power consumption.
  • each of the samples of the sample set includes a plurality of feature information that reflects the behavior habits of the user using the application
  • the embodiment of the present application may make the cleaning of the corresponding application more personalized.
  • the sample is constructed according to the multi-dimensional feature information of each application, and the decision tree is constructed to form a random forest.
  • the current feature information of each application and the exclusive random forest prediction application can be cleaned up, which can improve the accuracy of the cleaning.
  • another background application cleaning method is provided. As shown in FIG. 4a, the method in this embodiment includes:
  • Step S401 Collect multi-dimensional feature information of the application as a sample, and construct a sample set of the application.
  • the application mentioned in this embodiment may be any application installed on the electronic device, such as an office application, a communication application, a game application, a shopping application, and the like.
  • the multi-dimensional feature information of the application has a dimension of a certain length, and the parameters in each dimension correspond to a feature information of the application, that is, the multi-dimensional feature information is composed of a plurality of feature information.
  • the plurality of feature information may include feature information related to the application itself, for example, the duration of the application cutting into the background; the duration of the electronic device when the application is cut into the background; the number of times the application enters the foreground; the time when the application is in the foreground; the application enters
  • the background mode is, for example, switched by the home button (home button), switched back by the return button, switched by other applications, etc.; the type of application includes level one (common application), level two (other applications), and the like.
  • the plurality of feature information may further include related feature information of the electronic device where the application is located, for example, the time when the electronic device is off, the time of the bright screen, the current power, the wireless network connection status of the electronic device, whether the electronic device is in the charging state, or the like.
  • the sample set of the application may include multiple samples collected at a preset frequency during the historical time period.
  • the historical time period may be, for example, the past 7 days or 10 days; the preset frequency may be, for example, collected every 10 minutes and collected every half hour. It can be understood that the multi-dimensional feature data of the application acquired at one time constitutes one sample, and a plurality of samples constitute the sample set.
  • a specific sample may be as shown in Table 1 below, and includes feature information of multiple dimensions. It should be noted that the feature information shown in Table 1 is only an example. In practice, the number of feature information included in one sample may be increased. The number of information shown in Table 1 may be less than the number of information shown in Table 1. The specific feature information may be different from that shown in Table 1, and is not specifically limited herein.
  • Step S402 Mark the samples in the sample set to obtain a sample label of each sample.
  • the labeled sample tags include "cleanable” and "non-cleanable”.
  • the user may mark the historical usage habits of the application. For example, after the application enters the background for 30 minutes, the user closes the application and is marked as “cleanable”; for example, after the application enters the background for 3 minutes, the user will The application is switched to the foreground and is marked as “not cleanable”. Specifically, the value “1” can be used to indicate “can be cleaned up”, the value "0" is used to mean “not cleanable”, and vice versa.
  • Step S403 Each time from the multi-dimensional feature information of each sample, a preset number of feature information is randomly extracted to form a corresponding sub-sample, and the plurality of sub-samples constitute a training set, which is extracted multiple times to form multiple trainings. set.
  • the multi-dimensional feature information of each sample includes 15 pieces of feature information, and then five pieces of feature information can be randomly extracted from each sample to form a corresponding sub-sample.
  • a subsample constructed according to the sample shown in Table 1 can be as shown in Table 2:
  • Dimension Characteristic information 1 Background running time 2 Current battery power 3 Number of times you enter the front desk every day 4 Type of application, including level 1 (common application), level 2 (other applications) 5 Current wireless network status
  • Multiple sub-samples constitute a training set. After multiple extractions, multiple training sets are constructed, and the number of training sets constructed is constructed.
  • the training set is only composed of extracting part of the feature information from each sample of the sample set, the number of subsamples in the training set is the same as the number of samples in the sample set.
  • Step S404 marking the sample label of each subsample as a sample label of the corresponding sample.
  • sample label of the sample shown in Table 1 is "cleanable”
  • sample label of the subsample shown in Table 2 is also marked as “cleanable”.
  • Step S405 Train each training set to generate a corresponding decision tree.
  • the information gain of each feature information included, g(S, K) represents the information gain of the feature information K in the training set S
  • H(S) represents the entropy of the training set S
  • K) represents the use feature information K.
  • the entropy of the training set S after the training set S is divided.
  • p i represents the probability that the prediction result of the i-th category appears in the training set S
  • n represents the prediction result category.
  • the prediction result includes two types, which can be cleaned and cannot be cleaned, so n takes a value of 2.
  • the feature information with the largest information gain can be used as the feature information of the root node, and the remaining feature information is sequentially used as the feature information of the leaf node according to the information gain from the largest to the smallest, and the corresponding information is generated. Decision tree.
  • each subsample in a training set is as shown in Table 2, including “background running time”, “current power of electronic device”, “number of times to enter the foreground every day”, “type of application”, “current wireless network status”. "These five feature information are trained to generate a decision tree as shown in Figure 4c.
  • each training set is trained to obtain a number of decision trees equal to the number of training sets, see FIG. 4b, for example, when the number M of training sets is When you get the quantity M Decision tree, Decision trees can form random forests.
  • Step S406 When the application enters the background, predicting current feature information of the application by using the generated multiple decision trees, and outputting multiple prediction results.
  • the current feature information of the application has the same dimension as the multi-dimensional feature information of the application collected when the sample is composed, and the corresponding parameter values of the two may be the same or different in each dimension.
  • the multi-dimensional feature information of an application includes the current power: 30%, the background running time: 10 minutes, the number of backgrounds entering the background every day: 15 times, the application type: one level, the wireless network status: abnormal, the decision shown in Figure 4c.
  • the tree's prediction of the application will be "cleanable.”
  • the decision tree predicts an application and will output For each forecast, each forecast may be “cleanable” or “not cleanable”.
  • step S407 the statistics can be cleaned up, and the proportion of the prediction result in all the prediction results is recorded as the cleanable probability; the proportion of the statistically uncleanable prediction result in all the prediction results is recorded as the retention probability.
  • step S408 the relationship between the cleanable probability and the hold probability is determined. If the cleanable probability is greater than the hold probability, step S409 is performed. Otherwise, if the hold probability is greater than the cleanable probability, step S410 is performed.
  • Step S409 determining that the application can be cleaned up.
  • Step S410 determining to keep the state of the application unchanged.
  • each application is predicted by using a random forest built for each application.
  • the prediction result is shown in Table 3 below, and it is determined that the application A1 and the application A3 running in the background can be cleaned, and the application A2 is maintained. The state of running in the background does not change.
  • the application by collecting the multi-dimensional feature information of the application as a sample, constructing a sample set of the application, extracting feature information from the sample set according to a preset rule, constructing a plurality of training sets, and training each training set, Generating a corresponding decision tree, when the application enters the background, using the generated multiple decision trees to predict the current feature information of the application, and outputting multiple prediction results, and determining whether the cleaning can be cleaned according to the multiple prediction results.
  • the application cleans up the application that can be cleaned, thereby realizing the automatic cleaning of the background application, improving the running fluency of the electronic device and reducing the power consumption.
  • the embodiment of the present application further provides a background application cleaning device, including an acquisition unit, an extraction unit, a training unit, a prediction unit, and a first determining unit, as follows:
  • a collecting unit configured to collect multi-dimensional feature information of the application as a sample, and construct a sample set of the application
  • An extracting unit configured to extract feature information from the sample set according to a preset rule, and construct a plurality of training sets
  • a training unit configured to train each training set to generate a corresponding decision tree
  • a prediction unit configured to: when the application enters the background, use the generated multiple decision trees to predict current feature information of the application, and output multiple prediction results, where the prediction results include cleanable and non-cleanable;
  • the first determining unit is configured to determine, according to the multiple prediction results, whether the application can be cleaned up.
  • the training unit comprises:
  • a calculation subunit for calculating an information gain of each feature information included in each training set
  • a sub-unit is generated, and the feature information for maximizing the information gain is used as the feature information of the root node, and the remaining feature information is sequentially used as the feature information of the leaf node according to the information gain from the largest to the smallest, and a corresponding decision tree is generated.
  • the calculating subunit is specifically configured to:
  • Information gain, H(S) represents the entropy of the training set S, and H(S
  • p i represents the probability that the prediction result of the i-th category appears in the training set S
  • n represents the prediction result category
  • the extracting unit is configured to: randomly extract a preset number of feature information from the multi-dimensional feature information of each sample to form a corresponding sub-sample, and the plurality of sub-samples form a a training set, after multiple extractions, constructing multiple training sets;
  • the multi-dimensional feature information of the application includes Q pieces of feature information, the preset number is q, and the device further includes:
  • the apparatus further includes:
  • a marking unit for marking the samples in the sample set, obtaining a sample label of each sample, the sample label including cleanable and non-cleanable; and marking the sample label of each subsample as a sample label of the corresponding sample .
  • the first determining unit comprises:
  • Determining a subunit configured to determine that the application may be cleaned when the cleanable probability of the application is greater than a maintenance probability, and determine to keep the application running in the background when the retention probability of the application is greater than the cleanable probability change.
  • the background of the electronic device runs in multiple applications, and the device further includes:
  • the cleaning unit is configured to select a preset number of applications according to the order of the cleanable probability, or select an application that can clear the probability greater than the preset probability, and clean the selected application.
  • the collecting unit is specifically configured to:
  • the multi-dimensional feature information of the application is collected once every preset time period, where the multi-dimensional feature information of the application includes running feature information of the application and/or state feature information of the electronic device;
  • a plurality of the samples are acquired within a preset historical time period, and the sample set is constructed.
  • the apparatus further includes:
  • a sending unit configured to send the sample set of the application to a server
  • a receiving unit configured to receive the multiple decision trees from the server.
  • a background application cleaning device is further provided.
  • the background application cleaning device is applied to an electronic device.
  • the background application cleaning device includes an acquisition unit 501, an extraction unit 502, a training unit 503, and Prediction unit 504 and determination unit 505 are as follows:
  • the collecting unit 501 is configured to collect multi-dimensional feature information of the application as a sample, and construct a sample set of the application;
  • the extracting unit 502 is configured to extract feature information from the sample set according to a preset rule, and construct a plurality of training sets;
  • a training unit 503 configured to train each training set to generate a corresponding decision tree
  • the prediction unit 504 is configured to: when the application enters the background, use the generated multiple decision trees to predict current feature information of the application, and output multiple prediction results, where the prediction results include cleanable and non-cleanable;
  • the first determining unit 505 is configured to determine, according to the multiple prediction results, whether the application can be cleaned up.
  • the training unit 503 includes a computing subunit 5031 and a generating subunit 5032, as follows:
  • a calculation subunit 5031 configured to calculate an information gain of each feature information included in each training set
  • the generating sub-unit 5032 is configured to use the feature information for maximizing the information gain as the feature information of the root node, and use the feature information as the feature information of the leaf node in descending order of the information gain to generate a corresponding decision tree.
  • the computing subunit 5031 is specifically configured to:
  • Information gain, H(S) represents the entropy of the training set S, and H(S
  • p i represents the probability that the prediction result of the i-th category appears in the training set S
  • n represents the prediction result category
  • the extracting unit 502 is specifically configured to randomly extract a preset number of feature information from the multi-dimensional feature information of each sample to form a corresponding sub-sample, and the plurality of sub-samples constitute one a training set, after multiple extractions, constructing multiple training sets;
  • the multi-dimensional feature information of the application includes Q pieces of feature information, and the preset number is q.
  • the device further includes:
  • a second determining unit 507 configured to follow a formula Determine the number of training sets built, and M denotes the number of training sets.
  • the apparatus further includes:
  • a marking unit 506 configured to mark the samples in the sample set, obtain a sample label of each sample, the sample label includes a cleanable and non-cleanable; and mark the sample label of each subsample as a sample of the corresponding sample label.
  • the first determining unit 505 includes a statistical subunit 5051 and a determining subunit 5052, as follows:
  • the statistical sub-element 5051 is used for statistically cleaning up the proportion of the predicted result in all the predicted results, and is recorded as the cleanable probability; the proportion of the statistically uncleanable prediction result in all the predicted results is recorded as keeping Probability
  • the determining subunit 5052 is configured to determine that the application can be cleaned when the cleanable probability of the application is greater than the maintaining probability, and determine to keep the application running in the background when the maintaining probability of the application is greater than the cleanable probability constant.
  • the device further includes:
  • the cleaning unit 508 is configured to select a preset number of applications according to the order of the cleanable probability, or select an application that can clear the probability greater than the preset probability, and clean the selected application.
  • the collecting unit 501 is specifically configured to:
  • the multi-dimensional feature information of the application is collected once every preset time period, where the multi-dimensional feature information of the application includes running feature information of the application and/or state feature information of the electronic device;
  • a plurality of the samples are acquired within a preset historical time period, and the sample set is constructed.
  • the apparatus further includes:
  • a sending unit 509 configured to send the sample set of the application to a server
  • the receiving unit 510 is configured to receive the multiple decision trees from the server.
  • the multi-dimensional feature information of the application is collected by the collecting unit 501 as a sample, and the sample set of the application is constructed, and the extracting unit 502 extracts the feature information from the sample set according to a preset rule.
  • a plurality of training sets are constructed, and each training set is trained by the training unit 503 to generate a corresponding decision tree.
  • the prediction unit 504 uses the generated multiple decision trees to compare the current application.
  • the feature information is predicted, and the plurality of prediction results are output.
  • the determining unit 505 determines whether the application can be cleaned according to the plurality of prediction results, and the application that can be cleaned is cleaned, thereby realizing automatic cleaning of the background application, and improving The smooth running of electronic equipment reduces power consumption.
  • the foregoing modules may be implemented as a separate entity, or may be implemented in any combination, and may be implemented as the same or a plurality of entities.
  • the foregoing modules refer to the foregoing method embodiments, and details are not described herein again.
  • the electronic device 600 includes a processor 601 and a memory 602.
  • the processor 601 is electrically connected to the memory 602.
  • the processor 600 is a control center of the electronic device 600 that connects various portions of the entire electronic device using various interfaces and lines, by running or loading a computer program stored in the memory 602, and recalling data stored in the memory 602, The various functions of the electronic device 600 are performed and data is processed to thereby perform overall monitoring of the electronic device 600.
  • the memory 602 can be used to store software programs and modules, and the processor 601 executes various functional applications and data processing by running computer programs and modules stored in the memory 602.
  • the memory 602 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, a computer program required for at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may be stored according to Data created by the use of electronic devices, etc.
  • memory 602 can include high speed random access memory, and can also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, memory 602 can also include a memory controller to provide processor 601 access to memory 602.
  • the processor 601 in the electronic device 600 loads the instructions corresponding to the process of one or more computer programs into the memory 602 according to the following steps, and is stored in the memory 602 by the processor 601.
  • the computer program in which to implement various functions, as follows:
  • the generated plurality of decision trees are used to predict current feature information of the application, and output multiple prediction results, where the prediction results include cleanable and non-cleanable;
  • the processor 601 when training each training set to generate a corresponding decision tree, the processor 601 specifically performs the following steps:
  • the feature information with the largest information gain is used as the feature information of the root node, and the remaining feature information is sequentially used as the feature information of the leaf node according to the information gain from the largest to the smallest, and a corresponding decision tree is generated.
  • the processor 601 when calculating the information gain of each feature information included in each training set, the processor 601 specifically performs the following steps:
  • Information gain, H(S) represents the entropy of the training set S, and H(S
  • p i represents the probability that the prediction result of the i-th category appears in the training set S
  • n represents the prediction result category
  • the processor 601 when extracting feature information from the sample set according to a preset rule to construct a plurality of training sets, the processor 601 specifically performs the following steps:
  • a preset number of feature information is randomly extracted to form a corresponding sub-sample, and a plurality of sub-samples constitute a training set, and after multiple extractions, multiple trainings are constructed. set;
  • the multi-dimensional feature information of the application includes Q pieces of feature information, and the preset number is q.
  • the processor 601 is further configured to perform the following steps:
  • the processor 601 is further configured to perform the following steps:
  • sample labels including cleanable and non-cleanable
  • the processor 601 is further configured to perform the following steps:
  • the processor 601 when determining whether the application can be cleaned according to the multiple prediction results, is specifically configured to perform the following steps:
  • the statistics can be cleaned up.
  • the proportion of this prediction result in all prediction results is recorded as the cleanable probability;
  • the proportion of the statistical result that cannot be cleaned up in all the prediction results is recorded as the maintenance probability;
  • the cleanable probability of the application is greater than the hold probability, it is determined that the application may be cleaned, and if the hold probability of the application is greater than the cleanable probability, it is determined to keep the state of the application running in the background unchanged.
  • the processor 601 when there are multiple applications in the background, the processor 601 is further configured to perform the following steps:
  • the processor 601 when collecting the multi-dimensional feature information of the application as a sample and constructing the sample set of the application, the processor 601 is specifically configured to perform the following steps:
  • the multi-dimensional feature information of the application is collected once every preset time period, where the multi-dimensional feature information of the application includes running feature information of the application and/or state feature information of the electronic device;
  • a plurality of the samples are acquired within a preset historical time period, and the sample set is constructed.
  • the processor 601 is further configured to perform the following steps:
  • the electronic device in the embodiment of the present application constructs a sample set of the application by collecting multi-dimensional feature information of the application as a sample, extracts feature information from the sample set according to a preset rule, and constructs multiple training sets.
  • Each training set is trained to generate a corresponding decision tree.
  • the generated plurality of decision trees are used to predict the current feature information of the application, and output multiple prediction results, according to the multiple The prediction result determines whether the application can be cleaned up, and the application that can be cleaned is cleaned, thereby realizing the automatic cleaning of the background application, improving the running fluency of the electronic device and reducing the power consumption.
  • the electronic device 600 may further include: a display 603, a radio frequency circuit 604, an audio circuit 605, and a power source 606.
  • the display 603, the radio frequency circuit 604, the audio circuit 605, and the power source 606 are electrically connected to the processor 601, respectively.
  • the display 603 can be used to display information entered by a user or information provided to a user, as well as various graphical user interfaces, which can be composed of graphics, text, icons, video, and any combination thereof.
  • the display 603 can include a display panel.
  • the display panel can be configured in the form of a liquid crystal display (LCD) or an organic light-emitting diode (OLED).
  • LCD liquid crystal display
  • OLED organic light-emitting diode
  • the radio frequency circuit 604 can be used for transceiving radio frequency signals to establish wireless communication with network devices or other electronic devices through wireless communication, and to transmit and receive signals with network devices or other electronic devices.
  • the audio circuit 605 can be used to provide an audio interface between a user and an electronic device through a speaker or a microphone.
  • the power source 606 can be used to power various components of the electronic device 600.
  • the power source 606 can be logically coupled to the processor 601 through a power management system to enable functions such as managing charging, discharging, and power management through the power management system.
  • the electronic device 600 may further include a camera, a Bluetooth module, and the like, and details are not described herein again.
  • the embodiment of the present application further provides a storage medium, where the storage medium stores a computer program, and when the computer program runs on a computer, causes the computer to execute the background application cleaning method in any of the above embodiments, such as: Collecting multi-dimensional feature information of the application as a sample, constructing a sample set of the application; extracting feature information from the sample set according to a preset rule, constructing a plurality of training sets; training each training set to generate a corresponding decision tree; When the application enters the background, predicting current feature information of the application by using the generated multiple decision trees, and outputting multiple prediction results, where the prediction results include cleanable and non-cleanable; according to the multiple predictions The result determines if the application can be cleaned up.
  • the storage medium may be a magnetic disk, an optical disk, a read only memory (ROM), or a random access memory (RAM).
  • ROM read only memory
  • RAM random access memory
  • the computer program may be stored in a computer readable storage medium, such as in a memory of the electronic device, and executed by at least one processor in the electronic device, and may include, for example, a background application during execution.
  • the storage medium may be a magnetic disk, an optical disk, a read only memory, a random access memory, or the like.
  • each functional module may be integrated into one processing chip, or each module may exist physically separately, or two or more modules may be integrated into one module.
  • the above integrated modules can be implemented in the form of hardware or in the form of software functional modules.
  • the integrated module if implemented in the form of a software functional module and sold or used as a standalone product, may also be stored in a computer readable storage medium, such as a read only memory, a magnetic disk or an optical disk, etc. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Medical Informatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请实施例公开了一种后台应用清理方法、装置、存储介质及电子设备，其中，后台应用清理方法包括：采集应用的多维特征信息作为样本，构建所述应用的样本集；按照预设规则从所述样本集中提取特征信息，构建多个训练集；对每个训练集进行训练，生成对应的决策树；当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；根据所述多个预测结果确定是否可以清理所述应用。

Description

后台应用清理方法、装置、存储介质及电子设备
本申请要求于2017年9月30日提交中国专利局、申请号为201710922744.2、发明名称为“后台应用清理方法、装置、存储介质及电子设备”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信技术领域，具体涉及一种后台应用清理方法、装置、存储介质及电子设备。
背景技术
目前，智能手机等电子设备上，通常会安装多个不同功能的应用，以解决用户的不同需求。目前电子设备的系统支持多个应用同时运行，即一个应用在前台运行，其他应用可以在后台运行。如果长时间不清理后台运行的应用，则会导致电子设备的可用内存变小、中央处理器(central processing unit，CPU)占用率过高，导致电子设备出现运行速度变慢，卡顿，耗电过快等问题。因此，有必要提供一种方法解决上述问题。
技术解决方案
本申请实施例提供了一种后台应用清理方法、装置、存储介质及电子设备，能够提高电子设备的运行流畅度，降低功耗。
第一方面，本申请实施例提供的后台应用清理方法，包括：
采集应用的多维特征信息作为样本，构建所述应用的样本集；
按照预设规则从所述样本集中提取特征信息，构建多个训练集；
对每个训练集进行训练，生成对应的决策树；
当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
根据所述多个预测结果确定是否可以清理所述应用。
第二方面，本申请实施例提供的后台应用清理装置，包括：
采集单元，用于采集应用的多维特征信息作为样本，构建所述应用的样本集；
提取单元，用于按照预设规则从所述样本集中提取特征信息，构建多个训练集；
训练单元，用于对每个训练集进行训练，生成对应的决策树；
预测单元，用于当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
第一确定单元，用于根据所述多个预测结果确定是否可以清理所述应用。
第三方面，本申请实施例提供的存储介质，其上存储有计算机程序，当所述计算机程序在计算机上运行时，使得所述计算机执行如本申请实施例第一方面所述的后台应用清理方法。
第四方面，本申请实施例提供的电子设备，包括处理器和存储器，所述存储器有计算机程序，其特征在于，所述处理器通过调用所述计算机程序，用于执行如本申请实施例第一方面所述的后台应用清理方法。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的后台应用清理方法的应用场景示意图。
图2是本申请实施例提供的后台应用清理方法的一个流程示意图。
图3是本申请实施例提供的决策树的生成方法的一个流程示意图。
图4a是本申请实施例提供的后台应用清理方法的另一流程示意图。
图4b是本申请实施例提供的后台应用清理方法的另一流程示意图。
图4c是本申请实施例生成的决策树的一个结构示意图。
图5是本申请实施例提供的后台应用清理装置的一个结构示意图。
图6是本申请实施例提供的后台应用清理装置的另一结构示意图。
图7是本申请实施例提供的电子设备的一个结构示意图。
图8是本申请实施例提供的电子设备的另一结构示意图。
本发明的实施方式
请参照图式，其中相同的组件符号代表相同的组件，本申请的原理是以实施在一适当的运算环境中来举例说明。以下的说明是基于所例示的本申请具体实施例，其不应被视为限制本申请未在此详述的其它具体实施例。
在以下的说明中，本申请的具体实施例将参考由一部或多部计算机所执行的步骤及符号来说明，除非另有述明。因此，这些步骤及操作将有数次提到由计算机执行，本文所指的计算机执行包括了由代表了以一结构化型式中的数据的电子信号的计算机处理单元的操作。此操作转换该数据或将其维持在该计算机的内存系统中的位置处，其可重新配置或另外以本领域测试人员所熟知的方式来改变该计算机的运作。该数据所维持的数据结构为该内存的实体位置，其具有由该数据格式所定义的特定特性。但是，本申请原理以上述文字来说明，其并不代表为一种限制，本领域测试人员将可了解到以下所述的多种步骤及操作亦可实施在硬件当中。
本文所使用的术语“模块”可看做为在该运算系统上执行的软件对象。本文所述的不同组件、模块、引擎及服务可看做为在该运算系统上的实施对象。而本文所述的装置及方法可以以软件的方式进行实施，当然也可在硬件上进行实施，均在本申请保护范围之内。
本申请中的术语“第一”、“第二”和“第三”等是用于区别不同对象，而不是用于描述特定顺序。此外，术语“包括”和“具有”以及它们任何变形，意图在于覆盖不排他的包含。例如包含了一系列步骤或模块的过程、方法、系统、产品或设备没有限定于已列出的步骤或模块，而是某些实施例还包括没有列出的步骤或模块，或某些实施例还包括对于这些过程、方法、产品或设备固有的其它步骤或模块。
在本文中提及“实施例”意味着，结合实施例描述的特定特征、结构或特性可以包含在本申请的至少一个实施例中。在说明书中的各个位置出现该短语并不一定均是指相同的实施例，也不是与其它实施例互斥的独立的或备选的实施例。本领域技术人员显式地和隐式地理解的是，本文所描述的实施例可以与其它实施例相结合。
由于如果长时间不清理电子设备后台运行的应用，会导致电子设备出现运行速度变慢，卡顿，耗电过快等一系列问题。因而，本申请实施例提供了一种后台应用清理方法，包括以下步骤：
采集应用的多维特征信息作为样本，构建所述应用的样本集；
按照预设规则从所述样本集中提取特征信息，构建多个训练集；
对每个训练集进行训练，生成对应的决策树；
当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
根据所述多个预测结果确定是否可以清理所述应用。
一实施例中，所述对每个训练集进行训练，生成对应的决策树，包括：
计算每个训练集中包含的每个特征信息的信息增益；
将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
一实施例中，所述计算每个训练集中包含的每个特征信息的信息增益，包括：
按照公式g(S,K)＝H(S)-H(S|K)计算每个训练集中包含的每个特征信息的信息增 益，g(S,K)表示训练集S中特征信息K的信息增益，H(S)表示训练集S的熵，H(S|K)表示使用特征信息K划分训练集S后训练集S的熵；
其中，
Figure PCTCN2018099364-appb-000001
p i表示第i个类别的预测结果在训练集S中出现的概率，n表示预测结果类别；
其中，
Figure PCTCN2018099364-appb-000002
H(Y|K＝k i)表示特征信息K被固定为值k i时的条件熵。
一实施例中，所述按照预设规则从所述样本集中提取特征信息，构建多个训练集，包括：
每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，以构成对应的子样本，多个子样本构成一个训练集，经多次提取后，构建多个训练集。
一实施例中，所述应用的多维特征信息包括Q个特征信息，所述预设数目为q，所述方法还包括：
按照公式
Figure PCTCN2018099364-appb-000003
确定构建的训练集的数量，M表示训练集的数量。
一实施例中，在采集应用的多维特征信息作为样本之后，还包括：
对所述样本集中的样本进行标记，得到每个样本的样本标签，所述样本标签包括可清理和不可清理；
在构成对应的子样本之后，还包括：
将每个子样本的样本标签标记为对应的样本的样本标签。
一实施例中，所述根据所述多个预测结果确定是否可以清理所述应用，包括：
统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率；
若所述应用的可清理概率大于保持概率，则确定可以清理所述应用，若所述应用的保持概率大于可清理概率，则确定保持所述应用在后台运行的状态不变。
一实施例中，当后台有多个应用时，所述方法还包括：
按照可清理概率从大到小的顺序，选取预设数量的应用；或者选取可清理概率大于预设概率的应用；
清理所选取的应用。
一实施例中，所述采集应用的多维特征信息作为样本，构建所述应用的样本集，包括：
每隔预设时长采集一次所述应用的多维特征信息，其中，所述应用的多维特征信息包括所述应用的运行特征信息和/或所述电子设备的状态特征信息；
将每次采集的所述应用的多维特征信息确定为样本；
在预设历史时间段内获取多个所述样本，构建所述样本集。
一实施例中，在采集应用的多维特征信息作为样本，构建所述应用的样本集之后，还包括：
将所述应用的样本集发送给服务器；
从所述服务器接收所述多棵决策树。
本申请实施例提供的后台应用清理方法，其执行主体可以是本申请实施例提供的后台应用清理装置，或者集成了该后台应用清理装置的电子设备，其中该后台应用清理装置可以采用硬件或者软件的方式实现。其中，电子设备可以是智能手机、平板电脑、掌上电脑、笔记本电脑、或者台式电脑等设备。
请参阅图1，图1为本申请实施例提供的后台应用清理方法的应用场景示意图，以后台应用清理装置为电子设备为例，电子设备可以采集应用的多维特征信息作为样本，构建所述应用的样本集；按照预设规则从所述样本集中提取特征信息，构建多个训练集；对每个训练集进行训练，生成对应的决策树；当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；根据所述多个预测结果确定是否可以清理所述应用。
具体地，例如图1所示，以判断后台运行的邮箱应用是否可以清理为例，可以在历史时间段内，按照预设频率采集邮箱应用的多维特征信息(例如邮箱每天进入后台运行的时长、邮箱每天进入前台的次数等)作为样本，构建邮箱应用的样本集，提取样本集中的特征信息构建多个训练集，对每个训练集进行训练，生成对应的决策树，例如图1所示，生成了7棵决策树，利用每棵决策树对邮箱应用的当前特征进行预测，输出7个预测结果，用“1”表示预测结果“可清理”，用“0”表示预测结果“不可清理”，则图1所示的例子中，“可清理”的预测结果占多数，则确定可以清理后台运行的邮箱应用。
本申请实施例将从后台应用清理装置的角度，描述本申请实施例提供后台应用清理方法，该后台应用清理装置具体可以集成在电子设备中。该后台应用清理方法包括：采集应用的多维特征信息作为样本，构建所述应用的样本集；按照预设规则从所述样本集中提取特征信息，构建多个训练集；对每个训练集进行训练，生成对应的决策树；当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；根据所述多个预测结果确定是否可以清理所述应用。
在一优选实施例中，提供了一种后台应用清理方法，如图2所示，本实施例提供的后台应用清理方法的具体流程可以如下：
步骤S201、采集应用的多维特征信息作为样本，构建所述应用的样本集。
本实施例所提及的应用，可以是电子设备上安装的任何一个应用，例如办公应用、通信应用、游戏应用、购物应用等。
应用的多维特征信息具有一定长度的维度，其每个维度上的参数均对应表征应用的一种特征信息，即该多维特征信息由多个特征信息构成。该多个特征信息可以包括应用自身相关的特征信息，即应用的运行特征信息，例如：应用切入到后台的时长；应用切入到后台期间，电子设备的灭屏时长；应用进入前台的次数；应用处于前台的时间；应用进入后台的方式，例如被主页键(home键)切换进入、被返回键切换进入，被其他应用切换进入等；应用的类型，包括一级(常用应用)、二级(其他应用)等。该多个特征信息还可以包括应用所在的电子设备的相关特征信息，即电子设备的状态特征信息，例如：电子设备的灭屏时间、亮屏时间、当前电量，电子设备的无线网络连接状态，电子设备是否在充电状态等。
应用的样本集中，可以包括在历史时间段内，按照预设频率采集的多个样本。历史时间段，例如可以是过去7天、10天；预设频率，例如可以是每10分钟采集一次、每半小时采集一次。可以理解的是，一次采集的应用的多维特征数据构成一个样本，多个样本，构成所述样本集。
在构成样本集之后，可以对样本集中的每个样本进行标记，得到每个样本的样本标签，由于本实施要实现的是预测应用是否可以清理，因此，所标记的样本标签包括“可清理”和“不可清理”。具体可根据用户对应用的历史使用习惯进行标记，例如：当应用进入后 台30分钟后，用户关闭了该应用，则标记为“可清理”；再例如，当应用进入后台3分钟之后，用户将应用切换到了前台运行，则标记为“不可清理”。具体地，可以用数值“1”表示“可清理”，用数值“0”表示“不可清理”，反之亦可。
步骤S202、按照预设规则从所述样本集中提取特征信息，构建多个训练集。
例如，可以每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，构成对应的子样本，多个子样本构成一个训练集，多次提取后，构建多个训练集，预设数目可根据实际需要自定义取值。
由于是有放回地随机提取特征信息，因此，不同训练集中的特征信息可以重复，同一个训练集集中的特征信息也可以重复，这样可以有效地防止训练结果陷入过拟合。
由于训练集仅是从样本集的每个样本中提取部分特征信息构成的，因此训练集中的子样本数量与样本集中的样本数量是相同的。
例如，样本集中包括100个样本，每个样本包括15个特征信息，从样本集的每个样本中随机提取5个特征信息，构成对应的子样本，共可以构成100个子样本，每个子样本包括随机抽取的5个特征信息，100个子样本构成一个训练集。
在构成每个样本对应的子样本之后，可以将每个子样本的样本标签标记为对应的样本的样本标签。例如样本1的样本标签为“可清理”，则由样本1构成的子样本的样本标签也标记为“可清理”。
具体实现中，应用的多维特征信息可以包括Q个特征信息，上述预设数目可以为q，则可以根据公式
Figure PCTCN2018099364-appb-000004
确定构建的训练集的数量，M表示训练集的数量。
可以理解的是，根据随机组合原理，从每个样本的Q个特征信息中随机提取q个特征信息，包括
Figure PCTCN2018099364-appb-000005
种组合。在本申请实施例中，将训练集的数量(即随机抽取的次数)确定为
Figure PCTCN2018099364-appb-000006
既可以减少计算量，又可以保证决策树的数量，提高预测的准确性。
步骤S203、对每个训练集进行训练，生成对应的决策树。
具体的训练方法可参阅图3所示，包括以下步骤：
步骤S2031、计算每个训练集中包含的每个特征信息的信息增益。
具体地，可以按照公式g(S,K)＝H(S)-H(S|K)计算每个训练集中包含的每个特征信息的信息增益，g(S,K)表示训练集S中特征信息K的信息增益，H(S)表示训练集S的熵，H(S|K)表示使用特征信息K划分训练集S后训练集S的熵。
其中，
Figure PCTCN2018099364-appb-000007
p i表示第i个类别的预测结果在训练集S中出现的概率，n表示预测结果类别。具体在本实施例中，预测结果包括两类，可清理和不可清理，所以n取值为2。例如，训练集S中，包括10个样本(即10个应用的多维特征数据)，其中样本标签为“可清理”的样本数量有7个，样本标签为“不可清理”的样本数量为3个，则 训练集S的熵
Figure PCTCN2018099364-appb-000008
其中，
Figure PCTCN2018099364-appb-000009
H(Y|K＝k i)表示特征信息K被固定为值k i时的条件熵。例如，训练集S中，包括10个样本，采用“应用的类型”这个特征对训练集S进行分类时，当应用类型为一级时，有4个样本的样本标签为“不可清理”，有3个样本的样本标签为“可清理”；当应用类型为二级时，剩余3个样本的样本标签均为“可清理”，则使用特征“应用的类型”划分训练集S后的熵
Figure PCTCN2018099364-appb-000010
步骤S2032、将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
按照上述方法，对每个训练集进行训练，可以得到与训练集数量相等的数量的决策树，例如当训练集的数量为
Figure PCTCN2018099364-appb-000011
时，即可得到
Figure PCTCN2018099364-appb-000012
棵决策树。
生成的多棵决策树即构成随机森林，本实施例即采用构成的随机森林对应用进行预测。
在某些实施例中，可以重复上述步骤S201～S203，以为不同的应用构建对应的随机森林。在某些实施例中，随机森林的构建过程可以预先在服务器中完成，例如，电子设备可以将各个应用的样本数据发送给服务器，由服务器对各个应用的样本数据进行处理，生成各个应用对应的随机森林，当需要预测某个应用是否可以清理时，电子设备可以从服务器获取对应的随机森林，利用获取的随机森林对对应的应用进行预测。
步骤S204、当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理。
需要说明的是，所述应用的当前特征信息与构成样本时采集的所述应用的多维特征信息，具有相同的维度，二者在每个维度上对应的参数值可能相同，也可能不同。
采用任意一棵决策树对所述应用的当前特征信息进行预测的过程包括：
首先，从所述应用的多维特征信息中，提取与所述决策树的根节点的分裂特征对应的特征信息，根据该决策树的根节点的分裂条件对提取的特征信息进行判断，以得到根节点决策结果；如果决策结果满足停止遍历的条件(例如已明确应用是否可以清理)，则输出应用的预测结果；否则，则根据根节点决策结果确定待遍历的叶子节点。
接下来，从所述应用的多维特征信息中，提取与所确定的叶子节点的分裂特征对应的特征信息，根据所述叶子节点的分裂条件，对提取的特征信息进行判断，以得到叶子节点决策结果；如果决策结果满足停止遍历的条件(例如已明确应用是否可以清理)，则输出应用的预测结果；否则，则根据该叶子节点决策结果确定待遍历的下一叶子节点。
重复上述步骤，直至得到所述应用的预测结果。
每个决策树针对所述应用，都会输出一个预测结果，
Figure PCTCN2018099364-appb-000013
棵决策树可以输出
Figure PCTCN2018099364-appb-000014
个预测结果。
步骤S205、根据所述多个预测结果确定是否可以清理所述应用。
在得到所有决策树的预测结果之后，可以统计“可清理”这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计“不可清理”这一预测结果在所有预测结果中所占的比例，记为保持概率；
若所述应用的可清理概率大于保持概率，则确定可以清理所述应用，若所述应用的保持概率大于可清理概率，则确定保持所述应用在后台运行的状态不变。
在某些实施方式中，当后台存在多个应用时，可以利用每个应用对应的森林森林对对应的应用进行预测，根据预测结果确定每个应用是否可以清理；获取可以清理的应用的可清理概率，按照可清理概率从大到小的顺序，选取预设数量的应用进行清理，或者选取可清理概率大于预设概率的应用进行清理。
本实施例中，通过采集应用的多维特征信息作为样本，构建所述应用的样本集，按照预设规则从所述样本集中提取特征信息，构建多个训练集，对每个训练集进行训练，生成对应的决策树，当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，根据所述多个预测结果确定是否可以清理所述应用，清理可以清理的应用，以此实现了后台应用的自动清理，提高了电子设备的运行流畅度，降低了功耗。
进一步地，由于样本集的每个样本中，包括了反映用户使用应用的行为习惯的多个特征信息，因此本申请实施例可以使得对对应应用的清理更加个性化。
进一步地，根据每个应用程序的多维特征信息构建样本，生成决策树构成随机森林，采用每个应用程序的当前特征信息及专属的随机森林预测应用是否可清理，可以提高清理的准确度。
在一优选实施例中，提供了另一种后台应用清理方法，如图4a所示，本实施例的方法包括：
步骤S401、采集应用的多维特征信息作为样本，构建所述应用的样本集。
本实施例所提及的应用，可以是电子设备上安装的任何一个应用，例如办公应用、通信应用、游戏应用、购物应用等。
应用的多维特征信息具有一定长度的维度，其每个维度上的参数均对应表征应用的一种特征信息，即该多维特征信息由多个特征信息构成。该多个特征信息可以包括应用自身相关的特征信息，例如：应用切入到后台的时长；应用切入到后台期间，电子设备的灭屏时长；应用进入前台的次数；应用处于前台的时间；应用进入后台的方式，例如被主页键(home键)切换进入、被返回键切换进入，被其他应用切换进入等；应用的类型，包括一级(常用应用)、二级(其他应用)等。该多个特征信息还可以包括应用所在的电子设备的相关特征信息，例如：电子设备的灭屏时间、亮屏时间、当前电量，电子设备的无线网络连接状态，电子设备是否在充电状态等。
应用的样本集中，可以包括在历史时间段内，按照预设频率采集的多个样本。历史时间段，例如可以是过去7天、10天；预设频率，例如可以是每10分钟采集一次、每半小时采集一次。可以理解的是，一次采集的应用的多维特征数据构成一个样本，多个样本，构成所述样本集。
一个具体的样本可如下表1所示，包括多个维度的特征信息，需要说明的是，表1所示的特征信息仅为举例，实际中，一个样本所包含的特征信息的数量，可以多于表1所示信息的数量，也可以少于表1所示信息的数量，所取的具体特征信息也可以与表1所示不同，此处不作具体限定。
Figure PCTCN2018099364-appb-000015
Figure PCTCN2018099364-appb-000016
表1
步骤S402、对所述样本集中的样本进行标记，得到每个样本的样本标签。
由于本实施要实现的是预测应用是否可以清理，因此，所标记的样本标签包括“可清理”和“不可清理”。具体可根据用户对应用的历史使用习惯进行标记，例如：当应用进入后台30分钟后，用户关闭了该应用，则标记为“可清理”；再例如，当应用进入后台3分钟之后，用户将应用切换到了前台运行，则标记为“不可清理”。具体地，可以用数值“1”表示“可清理”，用数值“0”表示“不可清理”，反之亦可。
步骤S403、每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，构成对应的子样本，多个子样本构成一个训练集，多次提取，构成多个训练集。
具体地，例如表1所示，每个样本的多维特征信息包括15个特征信息，则可以每次从每个样本中有放回地随机提取5个特征信息，构成对应的子样本。根据表1所示的样本构建的一个子样本，可如表2所示：
维度 特征信息
1 后台运行时长
2 电子设备的当前电量
3 每天进入前台的次数
4 应用的类型，包括一级(常用应用)，二级(其他应用)
5 当前无线网状态
表2
多个子样本构成一个训练集，多次提取后，构建多个训练集，构建的训练集的数量
Figure PCTCN2018099364-appb-000017
由于训练集仅是从样本集的每个样本中提取部分特征信息构成的，因此训练集中的子样本数量与样本集中的样本数量是相同的。
步骤S404、将每个子样本的样本标签标记为对应的样本的样本标签。
例如表1所示的样本的样本标签为“可清理”，则将表2所示的子样本的样本标签也标记为“可清理”。
步骤S405、对每个训练集进行训练，生成对应的决策树。
具体地，针对任意一个训练集，可以先计算该训练集中包含的每个特征信息的信息增 益，可以按照公式g(S,K)＝H(S)-H(S|K)计算该训练集中包含的每个特征信息的信息增益，g(S,K)表示训练集S中特征信息K的信息增益，H(S)表示训练集S的熵，H(S|K)表示使用特征信息K划分训练集S后训练集S的熵。
其中，
Figure PCTCN2018099364-appb-000018
p i表示第i个类别的预测结果在训练集S中出现的概率，n表示预测结果类别。具体在本实施例中，预测结果包括两类，可清理和不可清理，所以n取值为2。
其中，
Figure PCTCN2018099364-appb-000019
H(Y|K＝k i)表示特征信息K被固定为值k i时的条件熵。
在得到每个特征信息的信息增益之后，可以将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
比如，一个训练集中的每个子样本均如表2所示，包含“后台运行时长”、“电子设备的当前电量”、“每天进入前台的次数”、“应用的类型”、“当前无线网状态”这5个特征信息，则经训练可以生成如图4c所示的决策树。
按照上述方法，对每个训练集进行训练，可以得到与训练集数量相等的数量的决策树，参阅图4b，例如当训练集的数量M为
Figure PCTCN2018099364-appb-000020
时，即可得到数量M为
Figure PCTCN2018099364-appb-000021
的决策树，
Figure PCTCN2018099364-appb-000022
棵决策树可以构成随机森林。
步骤S406、当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果。
需要说明的是，所述应用的当前特征信息与构成样本时采集的所述应用的多维特征信息，具有相同的维度，二者在每个维度上对应的参数值可能相同，也可能不同。
例如，某个应用的多维特征信息包括当前电量：30％，后台运行时长：10分钟，每天进入后台次数：15次，应用类型：一级，无线网状态：异常，则图4c所示的决策树对该应用的预测结果将是“可清理”。
在上面的例子中，
Figure PCTCN2018099364-appb-000023
棵决策树对一个应用进行预测，将输出
Figure PCTCN2018099364-appb-000024
个预测结果，每个预测结果可能是“可清理”，也可能是“不可清理”。
步骤S407、统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率。
步骤S408、判断可清理概率与保持概率的关系，若可清理概率大于保持概率，则执行步骤S409，反之，若保持概率大于可清理概率，则直至步骤S410。
步骤S409、确定可以清理所述应用。
步骤S410、确定保持所述应用的状态不变。
例如，后台运行了多个应用，利用针对每个应用构建的随机森林对每个应用进行预测，预测结果如下表3所示，则确定可以清理后台运行的应用A1和应用A3，而保持应用A2 在后台运行的状态不变。
Figure PCTCN2018099364-appb-000025
表3
本实施例中，通过采集应用的多维特征信息作为样本，构建所述应用的样本集，按照预设规则从所述样本集中提取特征信息，构建多个训练集，对每个训练集进行训练，生成对应的决策树，当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，根据所述多个预测结果确定是否可以清理所述应用，清理可以清理的应用，以此实现了后台应用的自动清理，提高了电子设备的运行流畅度，降低了功耗。
本申请实施例还提供了一种后台应用清理装置，包括采集单元、提取单元、训练单元、预测单元及第一确定单元，如下：
采集单元，用于采集应用的多维特征信息作为样本，构建所述应用的样本集；
提取单元，用于按照预设规则从所述样本集中提取特征信息，构建多个训练集；
训练单元，用于对每个训练集进行训练，生成对应的决策树；
预测单元，用于当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
第一确定单元，用于根据所述多个预测结果确定是否可以清理所述应用。
一实施例中，所述训练单元包括：
计算子单元，用于计算每个训练集中包含的每个特征信息的信息增益；
生成子单元，用于将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
一实施例中，所述计算子单元具体用于：
按照公式g(S,K)＝H(S)-H(S|K)计算每个训练集中包含的每个特征信息的信息增益，g(S,K)表示训练集S中特征信息K的信息增益，H(S)表示训练集S的熵，H(S|K)表示使用特征信息K划分训练集S后训练集S的熵；
其中，
Figure PCTCN2018099364-appb-000026
p i表示第i个类别的预测结果在训练集S中出现的概率，n表示预测结果类别；
其中，
Figure PCTCN2018099364-appb-000027
H(Y|K＝k i)表示特征信息K被固定为值k i时的条件熵。
一实施例中，所述提取单元具体用于，每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，以构成对应的子样本，多个子样本构成一个训练集，经多次提取后，构建多个训练集；
所述应用的多维特征信息包括Q个特征信息，所述预设数目为q，所述装置还包括：
第二确定单元，用于按照公式
Figure PCTCN2018099364-appb-000028
确定构建的训练集的数量，M表示训练集的数量。
一实施例中，所述装置还包括：
标记单元，用于对所述样本集中的样本进行标记，得到每个样本的样本标签，所述样本标签包括可清理和不可清理；以及将每个子样本的样本标签标记为对应的样本的样本标签。
一实施例中，所述第一确定单元包括：
统计子单元，用于统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率；
确定子单元，用于在所述应用的可清理概率大于保持概率时，确定可以清理所述应用，在所述应用的保持概率大于可清理概率时，确定保持所述应用在后台运行的状态不变。
一实施例中，所述电子设备的后台运行有多个应用，所述装置还包括：
清理单元，用于按照可清理概率从大到小的顺序，选取预设数量的应用；或者选取可清理概率大于预设概率的应用，清理所选取的应用。
一实施例中，所述采集单元具体用于，
每隔预设时长采集一次所述应用的多维特征信息，其中，所述应用的多维特征信息包括所述应用的运行特征信息和/或所述电子设备的状态特征信息；
将每次采集的所述应用的多维特征信息确定为样本；
在预设历史时间段内获取多个所述样本，构建所述样本集。
一实施例中，所述装置还包括：
发送单元，用于将所述应用的样本集发送给服务器；
接收单元，用于从所述服务器接收所述多棵决策树。
在一优选实施例中，还提供一种后台应用清理装置，该后台应用清理装置应用于电子设备，如图5所示，该后台应用清理装置包括采集单元501、提取单元502、训练单元503、预测单元504和确定单元505，如下：
采集单元501，用于采集应用的多维特征信息作为样本，构建所述应用的样本集；
提取单元502，用于按照预设规则从所述样本集中提取特征信息，构建多个训练集；
训练单元503，用于对每个训练集进行训练，生成对应的决策树；
预测单元504，用于当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
第一确定单元505，用于根据所述多个预测结果确定是否可以清理所述应用。
在一些实施例中，如图6所示，训练单元503包括计算子单元5031和生成子单元5032，如下：
计算子单元5031，用于计算每个训练集中包含的每个特征信息的信息增益；
生成子单元5032，用于将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
在一些实施例中，所述计算子单元5031具体用于：
按照公式g(S,K)＝H(S)-H(S|K)计算每个训练集中包含的每个特征信息的信息增益，g(S,K)表示训练集S中特征信息K的信息增益，H(S)表示训练集S的熵，H(S|K)表示使用特征信息K划分训练集S后训练集S的熵；
其中，
Figure PCTCN2018099364-appb-000029
p i表示第i个类别的预测结果在训练集S中出现的概率，n表示预测结果类别；
其中，
Figure PCTCN2018099364-appb-000030
H(Y|K＝k i)表示特征信息K被固定为值k i时的条件熵。
在一些实施例中，提取单元502具体用于，每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，以构成对应的子样本，多个子样本构成一个训练集，经多次提取后，构建多个训练集；
所述应用的多维特征信息包括Q个特征信息，所述预设数目为q，如图6所示，所述装置还包括：
第二确定单元507，用于按照公式
Figure PCTCN2018099364-appb-000031
确定构建的训练集的数量，M表示训练集的数量。
在一些实施例中，如图6所示，所述装置还包括：
标记单元506，用于对所述样本集中的样本进行标记，得到每个样本的样本标签，所述样本标签包括可清理和不可清理；以及将每个子样本的样本标签标记为对应的样本的样本标签。
在一些实施例中，如图6所示，第一确定单元505包括统计子单元5051和确定子单元5052，如下：
统计子单元5051，用于统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率；
确定子单元5052，用于在所述应用的可清理概率大于保持概率时，确定可以清理所述应用，在所述应用的保持概率大于可清理概率时，确定保持所述应用在后台运行的状态不变。
在一些实施例中，电子设备的后台有多个应用，如图6所示，所述装置还包括：
清理单元508，用于按照可清理概率从大到小的顺序，选取预设数量的应用；或者选取可清理概率大于预设概率的应用，清理所选取的应用。
在一些实施例中，采集单元501具体用于：
每隔预设时长采集一次所述应用的多维特征信息，其中，所述应用的多维特征信息包括所述应用的运行特征信息和/或所述电子设备的状态特征信息；
将每次采集的所述应用的多维特征信息确定为样本；
在预设历史时间段内获取多个所述样本，构建所述样本集。
在一些实施例中，如图6所示，所述装置还包括：
发送单元509，用于将所述应用的样本集发送给服务器；
接收单元510，用于从所述服务器接收所述多棵决策树。
由上可知，本实施例采用在电子设备中，由采集单元501采集应用的多维特征信息作为样本，构建所述应用的样本集，提取单元502按照预设规则从所述样本集中提取特征信息，构建多个训练集，再由训练单元503对每个训练集进行训练，生成对应的决策树，当所述应用进入后台时，由预测单元504利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，最后，确定单元505根据所述多个预测结果确定是否可 以清理所述应用，清理可以清理的应用，以此实现了后台应用的自动清理，提高了电子设备的运行流畅度，降低了功耗。
具体实施时，以上各个模块可以作为独立的实体来实现，也可以进行任意组合，作为同一或若干个实体来实现，以上各个模块的具体实施可参见前面的方法实施例，在此不再赘述。
本申请实施例还提供一种电子设备。请参阅图7，电子设备600包括处理器601以及存储器602。其中，处理器601与存储器602电性连接。
所述处理器600是电子设备600的控制中心，利用各种接口和线路连接整个电子设备的各个部分，通过运行或加载存储在存储器602内的计算机程序，以及调用存储在存储器602内的数据，执行电子设备600的各种功能并处理数据，从而对电子设备600进行整体监控。
所述存储器602可用于存储软件程序以及模块，处理器601通过运行存储在存储器602的计算机程序以及模块，从而执行各种功能应用以及数据处理。存储器602可主要包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需的计算机程序(比如声音播放功能、图像播放功能等)等；存储数据区可存储根据电子设备的使用所创建的数据等。此外，存储器602可以包括高速随机存取存储器，还可以包括非易失性存储器，例如至少一个磁盘存储器件、闪存器件、或其他易失性固态存储器件。相应地，存储器602还可以包括存储器控制器，以提供处理器601对存储器602的访问。
在本申请实施例中，电子设备600中的处理器601会按照如下的步骤，将一个或一个以上的计算机程序的进程对应的指令加载到存储器602中，并由处理器601运行存储在存储器602中的计算机程序，从而实现各种功能，如下：
采集应用的多维特征信息作为样本，构建所述应用的样本集；
按照预设规则从所述样本集中提取特征信息，构建多个训练集；
对每个训练集进行训练，生成对应的决策树；
当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
根据所述多个预测结果确定是否可以清理所述应用。
在某些实施方式中，在对每个训练集进行训练，生成对应的决策树时，处理器601具体执行以下步骤：
计算每个训练集中包含的每个特征信息的信息增益；
将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
在某些实施方式中，在计算每个训练集中包含的每个特征信息的信息增益时，处理器601具体执行以下步骤：
按照公式g(S,K)＝H(S)-H(S|K)计算每个训练集中包含的每个特征信息的信息增益，g(S,K)表示训练集S中特征信息K的信息增益，H(S)表示训练集S的熵，H(S|K)表示使用特征信息K划分训练集S后训练集S的熵；
其中，
Figure PCTCN2018099364-appb-000032
p i表示第i个类别的预测结果在训练集S中出现的概率，n表示预测结果类别；
其中，
Figure PCTCN2018099364-appb-000033
H(Y|K＝k i)表示特征信息K被固定为值k i时的条件熵。
在某些实施方式中，在按照预设规则从所述样本集中提取特征信息，构建多个训练集时，处理器601具体执行以下步骤：
每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，以构成对应的子样本，多个子样本构成一个训练集，经多次提取后，构建多个训练集；
所述应用的多维特征信息包括Q个特征信息，所述预设数目为q，处理器601还用于执行以下步骤：
按照公式
Figure PCTCN2018099364-appb-000034
确定构建的训练集的数量，M表示训练集的数量。
在某些实施方式中，在采集应用的多维特征信息作为样本之后，处理器601还用于执行以下步骤：
对所述样本集中的样本进行标记，得到每个样本的样本标签，所述样本标签包括可清理和不可清理；
在构成对应的子样本之后，处理器601还用于执行以下步骤：
将每个子样本的样本标签标记为对应的样本的样本标签。
在某些实施方式中，在根据所述多个预测结果确定是否可以清理所述应用时，处理器601具体用于执行以下步骤：
统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率；
若所述应用的可清理概率大于保持概率，则确定可以清理所述应用，若所述应用的保持概率大于可清理概率，则确定保持所述应用在后台运行的状态不变。
在某些实施方式中，当后台有多个应用时，处理器601还用于执行以下步骤：
按照可清理概率从大到小的顺序，选取预设数量的应用；或者选取可清理概率大于预设概率的应用；
清理所选取的应用。
在某些实施方式中，在采集应用的多维特征信息作为样本，构建所述应用的样本集时，处理器601具体用于执行以下步骤：
每隔预设时长采集一次所述应用的多维特征信息，其中，所述应用的多维特征信息包括所述应用的运行特征信息和/或所述电子设备的状态特征信息；
将每次采集的所述应用的多维特征信息确定为样本；
在预设历史时间段内获取多个所述样本，构建所述样本集。
在某些实施方式中，在采集应用的多维特征信息作为样本，构建所述应用的样本集之后，处理器601还用于执行以下步骤：
将所述应用的样本集发送给服务器；
从所述服务器接收所述多棵决策树。
由上述可知，本申请实施例的电子设备，通过采集应用的多维特征信息作为样本，构建所述应用的样本集，按照预设规则从所述样本集中提取特征信息，构建多个训练集，对每个训练集进行训练，生成对应的决策树，当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，根据所述多个预测结果确定是否可以清理所述应用，清理可以清理的应用，以此实现了后台应用的自动清理，提高 了电子设备的运行流畅度，降低了功耗。
请一并参阅图8，在某些实施方式中，电子设备600还可以包括：显示器603、射频电路604、音频电路605以及电源606。其中，其中，显示器603、射频电路604、音频电路605以及电源606分别与处理器601电性连接。
所述显示器603可以用于显示由用户输入的信息或提供给用户的信息以及各种图形用户接口，这些图形用户接口可以由图形、文本、图标、视频和其任意组合来构成。显示器603可以包括显示面板，在某些实施方式中，可以采用液晶显示器(Liquid Crystal Display，LCD)、或者有机发光二极管(Organic Light-Emitting Diode，OLED)等形式来配置显示面板。
所述射频电路604可以用于收发射频信号，以通过无线通信与网络设备或其他电子设备建立无线通讯，与网络设备或其他电子设备之间收发信号。
所述音频电路605可以用于通过扬声器、传声器提供用户与电子设备之间的音频接口。
所述电源606可以用于给电子设备600的各个部件供电。在一些实施例中，电源606可以通过电源管理系统与处理器601逻辑相连，从而通过电源管理系统实现管理充电、放电、以及功耗管理等功能。
尽管图8中未示出，电子设备600还可以包括摄像头、蓝牙模块等，在此不再赘述。
本申请实施例还提供一种存储介质，所述存储介质存储有计算机程序，当所述计算机程序在计算机上运行时，使得所述计算机执行上述任一实施例中的后台应用清理方法，比如：采集应用的多维特征信息作为样本，构建所述应用的样本集；按照预设规则从所述样本集中提取特征信息，构建多个训练集；对每个训练集进行训练，生成对应的决策树；当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；根据所述多个预测结果确定是否可以清理所述应用。
在本申请实施例中，存储介质可以是磁碟、光盘、只读存储器(Read Only Memory，ROM，)、或者随机存取记忆体(Random Access Memory，RAM)等。
在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述的部分，可以参见其他实施例的相关描述。
需要说明的是，对本申请实施例的后台应用清理方法而言，本领域普通决策人员可以理解实现本申请实施例的后台应用清理方法的全部或部分流程，是可以通过计算机程序来控制相关的硬件来完成，所述计算机程序可存储于一计算机可读取存储介质中，如存储在电子设备的存储器中，并被该电子设备内的至少一个处理器执行，在执行过程中可包括如后台应用清理方法的实施例的流程。其中，所述的存储介质可为磁碟、光盘、只读存储器、随机存取记忆体等。
对本申请实施例的后台应用清理装置而言，其各功能模块可以集成在一个处理芯片中，也可以是各个模块单独物理存在，也可以两个或两个以上模块集成在一个模块中。上述集成的模块既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。所述集成的模块如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中，所述存储介质譬如为只读存储器，磁盘或光盘等。
以上对本申请实施例所提供的一种后台应用清理方法、装置、存储介质及电子设备进行了详细介绍，本文中应用了具体个例对本申请的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本申请的方法及其核心思想；同时，对于本领域的技术人员，依据本申请的思想，在具体实施方式及应用范围上均会有改变之处，综上所述，本说明书内容不应理解为对本申请的限制。

Claims (20)

  1. 一种后台应用清理方法，其中，包括：
    采集应用的多维特征信息作为样本，构建所述应用的样本集；
    按照预设规则从所述样本集中提取特征信息，构建多个训练集；
    对每个训练集进行训练，生成对应的决策树；
    当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
    根据所述多个预测结果确定是否可以清理所述应用。
  2. 根据权利要求1所述的方法，其中，所述对每个训练集进行训练，生成对应的决策树，包括：
    计算每个训练集中包含的每个特征信息的信息增益；
    将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
  3. 根据权利要求2所述的方法，其中，所述计算每个训练集中包含的每个特征信息的信息增益，包括：
    按照公式g(S,K)＝H(S)-H(S|K)计算每个训练集中包含的每个特征信息的信息增益，g(S,K)表示训练集S中特征信息K的信息增益，H(S)表示训练集S的熵，H(S|K)表示使用特征信息K划分训练集S后训练集S的熵；
    其中，
    Figure PCTCN2018099364-appb-100001
    p i表示第i个类别的预测结果在训练集S中出现的概率，n表示预测结果类别；
    其中，
    Figure PCTCN2018099364-appb-100002
    H(Y|K＝k i)表示特征信息K被固定为值k i时的条件熵。
  4. 根据权利要求1至3任意一项所述的方法，其中，所述按照预设规则从所述样本集中提取特征信息，构建多个训练集，包括：
    每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，以构成对应的子样本，多个子样本构成一个训练集，经多次提取后，构建多个训练集。
  5. 根据权利要求4所述的方法，其中，所述应用的多维特征信息包括Q个特征信息，所述预设数目为q，所述方法还包括：
    按照公式
    Figure PCTCN2018099364-appb-100003
    确定构建的训练集的数量，M表示训练集的数量。
  6. 根据权利要求4所述的方法，其中，在采集应用的多维特征信息作为样本之后，还包括：
    对所述样本集中的样本进行标记，得到每个样本的样本标签，所述样本标签包括可清理和不可清理；
    在构成对应的子样本之后，还包括：
    将每个子样本的样本标签标记为对应的样本的样本标签。
  7. 根据权利要求1所述的方法，其中，所述根据所述多个预测结果确定是否可以清理所述应用，包括：
    统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率；
    若所述应用的可清理概率大于保持概率，则确定可以清理所述应用，若所述应用的保持概率大于可清理概率，则确定保持所述应用在后台运行的状态不变。
  8. 根据权利要求7所述的方法，其中，当后台有多个应用时，所述方法还包括：
    按照可清理概率从大到小的顺序，选取预设数量的应用；或者选取可清理概率大于预设概率的应用；
    清理所选取的应用。
  9. 根据权利要求1所述的方法，其中，所述采集应用的多维特征信息作为样本，构建所述应用的样本集，包括：
    每隔预设时长采集一次所述应用的多维特征信息，其中，所述应用的多维特征信息包括所述应用的运行特征信息和/或所述电子设备的状态特征信息；
    将每次采集的所述应用的多维特征信息确定为样本；
    在预设历史时间段内获取多个所述样本，构建所述样本集。
  10. 根据权利要求1所述的方法，其中，在采集应用的多维特征信息作为样本，构建所述应用的样本集之后，还包括：
    将所述应用的样本集发送给服务器；
    从所述服务器接收所述多棵决策树。
  11. 一种后台应用清理装置，其中，包括：
    采集单元，用于采集应用的多维特征信息作为样本，构建所述应用的样本集；
    提取单元，用于按照预设规则从所述样本集中提取特征信息，构建多个训练集；
    训练单元，用于对每个训练集进行训练，生成对应的决策树；
    预测单元，用于当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
    第一确定单元，用于根据所述多个预测结果确定是否可以清理所述应用。
  12. 根据权利要求11所述的装置，其中，所述训练单元包括：
    计算子单元，用于计算每个训练集中包含的每个特征信息的信息增益；
    生成子单元，用于将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
  13. 根据权利要求11或12所述的装置，其中，
    所述提取单元具体用于，每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，以构成对应的子样本，多个子样本构成一个训练集，经多次提取后，构建多个训练集；
    所述应用的多维特征信息包括Q个特征信息，所述预设数目为q，所述装置还包括：
    第二确定单元，用于按照公式
    Figure PCTCN2018099364-appb-100004
    确定构建的训练集的数量，M表示训练集的数量。
  14. 根据权利要求13所述的装置，其中，所述装置还包括：
    标记单元，用于对所述样本集中的样本进行标记，得到每个样本的样本标签，所述样本标签包括可清理和不可清理；以及将每个子样本的样本标签标记为对应的样本的样本标 签。
  15. 根据权利要求11所述的装置，其中，所述第一确定单元包括：
    统计子单元，用于统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率；
    确定子单元，用于在所述应用的可清理概率大于保持概率时，确定可以清理所述应用，在所述应用的保持概率大于可清理概率时，确定保持所述应用在后台运行的状态不变。
  16. 一种存储介质，其上存储有计算机程序，其中，当所述计算机程序在计算机上运行时，使得所述计算机执行如权利要求1至10任一项所述的后台应用清理方法。
  17. 一种电子设备，包括处理器和存储器，所述存储器存储有计算机程序，其中，所述处理器通过调用所述计算机程序，从而执行以下步骤：
    采集应用的多维特征信息作为样本，构建所述应用的样本集；
    按照预设规则从所述样本集中提取特征信息，构建多个训练集；
    对每个训练集进行训练，生成对应的决策树；
    当所述应用进入后台时，利用生成的多棵决策树对所述应用的当前特征信息进行预测，并输出多个预测结果，所述预测结果包括可清理和不可清理；
    根据所述多个预测结果确定是否可以清理所述应用。
  18. 根据权利要求17所述的电子设备，其中，在对每个训练集进行训练，生成对应的决策树时，所述处理器具体用于执行以下步骤：
    计算每个训练集中包含的每个特征信息的信息增益；
    将信息增益最大的特征信息作为根节点的特征信息，将其余特征信息按照信息增益从大到小的顺序依次作为叶子节点的特征信息，生成对应的决策树。
  19. 根据权利要求17或18所述的电子设备，其中，在按照预设规则从所述样本集中提取特征信息，构建多个训练集时，所述处理器具体用于执行以下步骤：
    每次从每个样本的多维特征信息中，有放回地随机提取预设数目的特征信息，以构成对应的子样本，多个子样本构成一个训练集，经多次提取后，构建多个训练集；
    所述应用的多维特征信息包括Q个特征信息，所述预设数目为q，所述处理器还用于执行以下步骤：
    按照公式
    Figure PCTCN2018099364-appb-100005
    确定构建的训练集的数量，M表示训练集的数量。
  20. 根据权利要求17所述的电子设备，其中，在根据所述多个预测结果确定是否可以清理所述应用时，所述处理器具体用于执行以下步骤：
    统计可清理这一预测结果在所有预测结果中所占的比例，记为可清理概率；统计不可清理这一预测结果在所有预测结果中所占的比例，记为保持概率；
    若所述应用的可清理概率大于保持概率，则确定可以清理所述应用，若所述应用的保持概率大于可清理概率，则确定保持所述应用在后台运行的状态不变。
PCT/CN2018/099364 2017-09-30 2018-08-08 后台应用清理方法、装置、存储介质及电子设备 WO2019062342A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP18862386.2A EP3702912A4 (en) 2017-09-30 2018-08-08 BACKGROUND APPLICATION CLEANING METHOD AND DEVICE AND STORAGE MEDIUM AND ELECTRONIC DEVICE
US16/819,777 US11544633B2 (en) 2017-09-30 2020-03-16 Method for cleaning up background application, storage medium, and electronic device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710922744.2A CN107678800B (zh) 2017-09-30 2017-09-30 后台应用清理方法、装置、存储介质及电子设备
CN201710922744.2 2017-09-30

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/819,777 Continuation US11544633B2 (en) 2017-09-30 2020-03-16 Method for cleaning up background application, storage medium, and electronic device

Publications (2)

Publication Number Publication Date
WO2019062342A1 true WO2019062342A1 (zh) 2019-04-04
WO2019062342A9 WO2019062342A9 (zh) 2019-07-11

Family

ID=61139811

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/099364 WO2019062342A1 (zh) 2017-09-30 2018-08-08 后台应用清理方法、装置、存储介质及电子设备

Country Status (4)

Country Link
US (1) US11544633B2 (zh)
EP (1) EP3702912A4 (zh)
CN (1) CN107678800B (zh)
WO (1) WO2019062342A1 (zh)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107678800B (zh) * 2017-09-30 2020-02-14 Oppo广东移动通信有限公司 后台应用清理方法、装置、存储介质及电子设备
CN110086835B (zh) * 2018-01-24 2021-08-03 腾讯科技（深圳）有限公司 应用程序管控方法、终端、服务器及系统
CN108427582B (zh) * 2018-03-09 2022-02-08 北京小米移动软件有限公司 卡顿状态确定方法、装置及计算机可读存储介质
CN108427618B (zh) * 2018-03-09 2021-09-07 北京小米移动软件有限公司 卡顿状态确定方法、装置及计算机可读存储介质
CN108984369A (zh) * 2018-07-13 2018-12-11 厦门美图移动科技有限公司 卡顿预测方法、装置及移动终端
US11625640B2 (en) * 2018-10-05 2023-04-11 Cisco Technology, Inc. Distributed random forest training with a predictor trained to balance tasks
CN112101397B (zh) * 2019-06-18 2024-09-20 北京京东振世信息技术有限公司 一种预测图书重量区间的方法和装置
CN112130991B (zh) * 2020-08-28 2024-07-23 北京思特奇信息技术股份有限公司 一种基于机器学习的应用程序控制方法和系统
CN113609031B (zh) * 2021-07-08 2024-03-29 深圳市晨北科技有限公司 数据清理模型构建方法、数据清理方法及相关设备、介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120290530A1 (en) * 2011-05-13 2012-11-15 Piriform Commercial Ltd. Network registry and file cleaner
CN104866366A (zh) * 2015-06-15 2015-08-26 广东欧珀移动通信有限公司 应用程序清理方法及装置
CN105335099A (zh) * 2015-09-25 2016-02-17 深圳市金立通信设备有限公司 一种内存清理方法及终端
CN107678800A (zh) * 2017-09-30 2018-02-09 广东欧珀移动通信有限公司 后台应用清理方法、装置、存储介质及电子设备

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7174354B2 (en) * 2002-07-31 2007-02-06 Bea Systems, Inc. System and method for garbage collection in a computer system, which uses reinforcement learning to adjust the allocation of memory space, calculate a reward, and use the reward to determine further actions to be taken on the memory space
CN103970395A (zh) * 2013-01-30 2014-08-06 腾讯科技（深圳）有限公司 关闭后台程序的方法和装置
KR20160091786A (ko) * 2015-01-26 2016-08-03 삼성전자주식회사 사용자 관리 방법 및 사용자 관리 장치
CN105389193B (zh) * 2015-12-25 2019-04-26 北京奇虎科技有限公司 应用的加速处理方法、装置和系统、服务器
CN105608476B (zh) 2016-02-16 2019-03-15 北京小米移动软件有限公司 基于随机森林分类器的分类方法及装置
CN107291549B (zh) * 2016-03-31 2020-11-24 阿里巴巴集团控股有限公司 一种管理应用程序的方法及装置
CN106648023A (zh) * 2016-10-02 2017-05-10 上海青橙实业有限公司 移动终端及其基于神经网络的省电方法
WO2018152734A1 (zh) * 2017-02-23 2018-08-30 华为技术有限公司 一种清理终端的应用任务列表的方法、装置及终端
CN107133094B (zh) * 2017-06-05 2021-11-02 努比亚技术有限公司 应用管理方法、移动终端及计算机可读存储介质
CN107894827B (zh) * 2017-10-31 2020-07-07 Oppo广东移动通信有限公司 应用清理方法、装置、存储介质及电子设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120290530A1 (en) * 2011-05-13 2012-11-15 Piriform Commercial Ltd. Network registry and file cleaner
CN104866366A (zh) * 2015-06-15 2015-08-26 广东欧珀移动通信有限公司 应用程序清理方法及装置
CN105335099A (zh) * 2015-09-25 2016-02-17 深圳市金立通信设备有限公司 一种内存清理方法及终端
CN107678800A (zh) * 2017-09-30 2018-02-09 广东欧珀移动通信有限公司 后台应用清理方法、装置、存储介质及电子设备

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3702912A4 *

Also Published As

Publication number Publication date
US20200219016A1 (en) 2020-07-09
US11544633B2 (en) 2023-01-03
CN107678800A (zh) 2018-02-09
CN107678800B (zh) 2020-02-14
EP3702912A4 (en) 2021-03-03
EP3702912A1 (en) 2020-09-02
WO2019062342A9 (zh) 2019-07-11

Similar Documents

Publication Publication Date Title
WO2019062342A9 (zh) 后台应用清理方法、装置、存储介质及电子设备
US20230109816A1 (en) Notification display method and terminal
WO2019062414A1 (zh) 应用程序管控方法、装置、存储介质及电子设备
CN109299387B (zh) 一种基于智能推荐的消息推送方法、装置及终端设备
CN105427129B (zh) 一种信息的投放方法及系统
CN108337358B (zh) 应用清理方法、装置、存储介质及电子设备
JP6991163B2 (ja) 情報をプッシュする方法及びデバイス
WO2019062418A1 (zh) 应用清理方法、装置、存储介质及电子设备
CN107832132B (zh) 应用控制方法、装置、存储介质及电子设备
WO2019062413A1 (zh) 应用程序管控方法、装置、存储介质及电子设备
CN104537000B (zh) 一种用于推送信息的方法和装置
WO2019062405A1 (zh) 应用程序的处理方法、装置、存储介质及电子设备
US11422831B2 (en) Application cleaning method, storage medium and electronic device
CN107894827B (zh) 应用清理方法、装置、存储介质及电子设备
CN108011928A (zh) 一种信息推送方法、终端设备及计算机可读介质
CN107870810B (zh) 应用清理方法、装置、存储介质及电子设备
WO2019120007A1 (zh) 用户性别预测方法、装置及电子设备
CN107748697B (zh) 应用关闭方法、装置、存储介质及电子设备
WO2019062341A1 (zh) 后台应用清理方法、装置、存储介质及电子设备
WO2019062419A1 (zh) 应用清理方法、装置、存储介质及电子设备
WO2018153179A1 (zh) 消息推送方法、设备及可编程设备
WO2019062462A1 (zh) 应用控制方法、装置、存储介质以及电子设备
CN109587328B (zh) 消息管理方法和装置、存储介质及电子设备
CN103092348A (zh) 一种基于用户行为的移动终端广告播放方法
WO2020258773A1 (zh) 确定推送用户群的方法、装置、设备及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18862386

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 2018862386

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2018862386

Country of ref document: EP

Effective date: 20200430