CN107678845A - Application program management-control method, device, storage medium and electronic equipment - Google Patents
Application program management-control method, device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN107678845A CN107678845A CN201710940355.2A CN201710940355A CN107678845A CN 107678845 A CN107678845 A CN 107678845A CN 201710940355 A CN201710940355 A CN 201710940355A CN 107678845 A CN107678845 A CN 107678845A
- Authority
- CN
- China
- Prior art keywords
- application program
- sample
- loss function
- prediction result
- probability
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Classifications
-
- 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
-
- 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
- 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
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- 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
- 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
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5011—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resources being hardware resources other than CPUs, Servers and Terminals
- G06F9/5022—Mechanisms to release resources
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
This application discloses a kind of application program management-control method, device, storage medium and electronic equipment, methods described includes:The multidimensional characteristic information of acquisition applications program is as sample, build the sample set of application program, characteristic information is extracted from sample set according to preset rules, build multiple training sets, Logic Regression Models are trained according to multiple training sets, with the forecast model after being trained, the current multidimensional characteristic information of application program is obtained and is simultaneously used as forecast sample, prediction result is generated according to the forecast model after forecast sample and training, and management and control is carried out to application program according to prediction result.The application can improve the accuracy being predicted to application program, so as to lift intellectuality and accuracy that management and control is carried out to the application program for entering backstage.
Description
Technical field
The application belongs to communication technical field, more particularly to a kind of application program management-control method, device, storage medium and electricity
Sub- equipment.
Background technology
With the development of electronic technology, people generally install many application programs on an electronic device.When user is in electronics
When multiple application programs are opened in equipment, if user retracts the desktop of electronic equipment or rests on the application of a certain application program
Interface or management and control electronic equipment screen, then multiple application programs that user opens still can be in the running background of electronic equipment.
But many application users in backstage are interior for a period of time to use, but the application program of these running backgrounds
The internal memory of electronic equipment can seriously be taken so that central processing unit (central processing unit, CPU) occupancy
It is too high, cause electronic equipment the speed of service occur slack-off, interim card, the problems such as power consumption is too fast, and cause the power consumption of electronic equipment
Speed is accelerated.
The content of the invention
The application provides a kind of application program management-control method, device, storage medium and electronic equipment, can be lifted to application
Program carries out intellectuality and the accuracy of management and control.
In a first aspect, the embodiment of the present application provides a kind of application program management-control method, including:
The multidimensional characteristic information of acquisition applications program builds the sample set of the application program as sample;
Characteristic information is extracted from the sample set according to preset rules, builds multiple training sets;
Logic Regression Models are trained according to the multiple training set, with the forecast model after being trained;
Obtain the current multidimensional characteristic information of the application program and be used as forecast sample, according to the forecast sample and instruction
Forecast model generation prediction result after white silk, and management and control is carried out to the application program according to the prediction result.
Second aspect, the embodiment of the present application provide a kind of application program control device, including:
Acquisition module, the multidimensional characteristic information for acquisition applications program build the sample of the application program as sample
This collection;
Module is built, for extracting characteristic information from the sample set according to preset rules, builds multiple training sets;
Training module, for being trained according to the multiple training set to Logic Regression Models, after being trained
Forecast model;
Management and control module, for obtaining the current multidimensional characteristic information of the application program and being used as forecast sample, according to institute
The forecast model generation prediction result after forecast sample and training is stated, and the application program is carried out according to the prediction result
Management and control.
The third aspect, the embodiment of the present application provide a kind of storage medium, computer program are stored thereon with, when the calculating
When machine program is run on computers so that the computer performs above-mentioned application program management-control method.
Fourth aspect, the embodiment of the present application provide a kind of electronic equipment, including processor and memory, the memory have
Computer program, the processor is by calling the computer program, for performing above-mentioned application program management-control method.
Application program management-control method, device, storage medium and the electronic equipment that the embodiment of the present application provides, should by collection
By the use of the multidimensional characteristic information of program as sample, the sample set of application program is built, is extracted according to preset rules from sample set
Characteristic information, multiple training sets are built, Logic Regression Models are trained according to multiple training sets, with pre- after being trained
Model is surveyed, the current multidimensional characteristic information of application program is obtained and is used as forecast sample, according to pre- after forecast sample and training
Model generation prediction result is surveyed, and management and control is carried out to application program according to prediction result.The application can be improved to application program
The accuracy being predicted, so as to lift intellectuality and accuracy that management and control is carried out to the application program for entering backstage.
Brief description of the drawings
In order to illustrate more clearly of the technical scheme in the embodiment of the present application, make required in being described below to embodiment
Accompanying drawing is briefly described.It should be evident that drawings in the following description are only some embodiments of the present application, for
For those skilled in the art, on the premise of not paying creative work, it can also be obtained according to these accompanying drawings other attached
Figure.
Fig. 1 is the system schematic for the application program control device that the embodiment of the present application provides.
Fig. 2 is the application scenarios schematic diagram for the application program control device that the embodiment of the present application provides.
Fig. 3 is the schematic flow sheet for the application program management-control method that the embodiment of the present application provides.
Fig. 4 is another schematic flow sheet for the application program management-control method that the embodiment of the present application provides.
Fig. 5 is the another application schematic diagram of a scenario for the application program control device that the embodiment of the present application provides.
Fig. 6 is the structural representation for the application program control device that the embodiment of the present application provides.
Fig. 7 is another structural representation for the application program control device that the embodiment of the present application provides.
Fig. 8 is the another structural representation for the application program control device that the embodiment of the present application provides.
Fig. 9 is the structural representation for the electronic equipment that the embodiment of the present application provides.
Figure 10 is another structural representation for the electronic equipment that the embodiment of the present application provides.
Embodiment
Schema is refer to, wherein identical element numbers represent identical component, and the principle of the application is to implement one
Illustrated in appropriate computing environment.The following description is based on illustrated the application specific embodiment, and it should not be by
It is considered as limitation the application other specific embodiments not detailed herein.
In the following description, the specific embodiment of the application is by with reference to as the step performed by one or multi-section computer
And symbol illustrates, unless otherwise stating clearly.Therefore, these steps and operation will have to mention for several times is performed by computer, this paper institutes
The computer of finger, which performs, to be included by representing with the computer processing unit of the electronic signal of the data in a structuring pattern
Operation.The data or the opening position being maintained in the memory system of the computer are changed in this operation, and its is reconfigurable
Or change the running of the computer in a manner of known to the tester of this area in addition.The data structure that the data are maintained
For the provider location of the internal memory, it has the particular characteristics as defined in the data format.But the application principle is with above-mentioned text
Word illustrates that it is not represented as a kind of limitation, this area tester will appreciate that plurality of step as described below and behaviour
Also may be implemented among hardware.
Term as used herein " module " can see the software object performed in the arithmetic system as.It is as described herein
Different components, module, engine and service can see the objective for implementation in the arithmetic system as.And device as described herein and side
Method can be implemented in a manner of software, can also be implemented certainly on hardware, within the application protection domain.
Term " first ", " second " and " the 3rd " in the application etc. is to be used to distinguish different objects, rather than for retouching
State particular order.In addition, term " comprising " and " having " and their any deformations, it is intended that cover non-exclusive include.
Such as contain the step of process, method, system, product or the equipment of series of steps or module is not limited to list or
Module, but some embodiments also include the step of not listing or module, or some embodiments also include for these processes,
Method, product or equipment intrinsic other steps or module.
Referenced herein " embodiment " is it is meant that the special characteristic, structure or the characteristic that describe can wrap in conjunction with the embodiments
It is contained at least one embodiment of the application.Each position in the description occur the phrase might not each mean it is identical
Embodiment, nor the independent or alternative embodiment with other embodiments mutual exclusion.Those skilled in the art explicitly and
Implicitly understand, embodiment described herein can be combined with other embodiments.
In the prior art, when carrying out management and control to the application program on backstage, typically directly accounted for according to the internal memory of electronic equipment
With situation and the priority of each application program, the certain applications program to backstage is cleared up, with releasing memory.But some
Application program is critically important to user or user needs to reuse some application programs in a short time, if to rear progress
These application programs are cleaned out during cleaning, then need electronic equipment to reload this when user reuses these application programs
The processes of a little application programs is, it is necessary to take considerable time and memory source.Wherein, the electronic equipment can be smart mobile phone, put down
The equipment such as plate computer, desktop computer, notebook computer or palm PC.
Referring to Fig. 1, Fig. 1 is the system schematic for the application program control device that the embodiment of the present application provides.The application
Program control device is mainly used in:The multidimensional characteristic information of acquisition applications program builds the sample set of application program as sample,
Characteristic information is extracted from sample set according to preset rules, builds multiple training sets, is returned according to the multiple training Set Pair Logic
Return model to be trained, with the forecast model after being trained, obtain the current multidimensional characteristic information of application program simultaneously as pre-
Test sample sheet, prediction result is generated according to the forecast model after forecast sample and training, and application program entered according to prediction result
Row management and control, such as clear up or freeze.
Specifically, referring to Fig. 2, the application scenarios that Fig. 2 is the application program control device that the embodiment of the present application provides show
It is intended to.For example application program control device detects the application in the running background of electronic equipment when receiving management and control request
Program includes application program a, application program b and application program c.Then multidimensional characteristic corresponding to application program a is obtained respectively
Multidimensional characteristic information corresponding to multidimensional characteristic information corresponding to information, application program b and application program c, passes through logistic regression
Whether model needs probability to be used to be predicted application program a, obtains probability a ', by Logic Regression Models to application
Whether program b needs probability to be used to be predicted, and obtains probability b ', by Logic Regression Models to application program c whether
Need probability to be used to be predicted, obtain probability c ';Running background is answered according to probability a ', probability b ' and probability c '
Management and control is carried out with program a, application program b and application program c, such as the minimum application program b of probability is closed.
The embodiment of the present application provides a kind of application program management-control method, and the executive agent of the application program management-control method can be with
It is the application program control device that the embodiment of the present application provides, or is integrated with the electronic equipment of the application program control device,
Wherein the application program control device can be realized by the way of hardware or software.
The embodiment of the present application will be described from the angle of application program control device, and the application program control device is specific
It can integrate in the electronic device.The application program management-control method includes:The multidimensional characteristic information of acquisition applications program is as sample
This, builds the sample set of application program, characteristic information is extracted from sample set according to preset rules, build multiple training sets, root
Logic Regression Models are trained according to the multiple training set, with the forecast model after being trained, application program is obtained and works as
Preceding multidimensional characteristic information is simultaneously used as forecast sample, and prediction result is generated according to the forecast model after forecast sample and training, and
Management and control is carried out to application program according to prediction result.
Referring to Fig. 3, Fig. 3 is the schematic flow sheet for the application program management-control method that the embodiment of the present application provides.The application
The application program management-control method that embodiment provides is applied to electronic equipment, and idiographic flow can be as follows:
Step 101, the multidimensional characteristic information of acquisition applications program builds the sample set of application program as sample.
Wherein, default application program can be mounted in any application in electronic equipment, such as communication applications journey
Sequence, multimedia application, game application, information application program or shopping application program etc..
The multidimensional characteristic information of application has a dimension of certain length, and the parameter in each of which dimension is corresponding to characterize application
A kind of characteristic information, i.e. the multidimensional characteristic information is made up of multiple characteristic informations.The plurality of characteristic information can include application
Itself related characteristic information.Among an embodiment, 30 features of equipment can be collected, form 30 dimensional vectors, should
30 features are, for example,:
Using the duration of last incision backstage till now;
During cutting backstage till now using the last time, add up screen shut-in time length;
Using the last time duration is used on foreground;
Using the upper last time duration was used on foreground;
Using the upper last time duration was used on foreground;
Using the number for entering foreground in one day (by statistics daily);
Enter the number on foreground using (day off is separately counted by working day, day off) in one day;
Using in one day the time on foreground is in (by statistics daily);
The time on foreground is in using (day off is separately counted by working day, day off) in one day;
Intended application is daily 8:00-12:The time span that 00 this period was used;
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 0-5 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 5-10 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 10-15 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 15-20 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 15-20 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 25-30 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to after 30 minutes);
Intended application one-level type;
Intended application two-level type;
The mode that intended application is switched, it is divided into and switches by the switching of home keys, by the switching of recent keys, by other application;
Screen amount is gone out the time;
Current screen light on and off state;
Currently whether have and charging;
Current electricity;
Current wifi states;
The period index on the current time residing same day;
The background application counts gained immediately following the number that is opened after current foreground application regardless of day off on working day;
The background application divides day off on working day to count immediately following the number that is opened after current foreground application;
Current foreground, which is applied, enters backstage to intended application into foreground by the Mean Time Between Replacement counted daily;
Current foreground, which is applied, enters backstage to intended application into when extinguishing during foreground by the average screen counted daily
Between.
In the sample set of application, it can be included in historical time section, the multiple samples gathered according to predeterminated frequency.History
Period, such as can be 7 days, 10 days in the past;Predeterminated frequency, such as can be that collection in every 10 minutes once, per half an hour is adopted
Collection is once.It is understood that the multi-dimensional feature data of the application once gathered forms a sample, multiple samples, institute is formed
State sample set.
After sample set is formed, each sample in sample set can be marked, obtain the sample of each sample
Label, because this implementation will be accomplished that whether prediction application can clear up, therefore, the sample label marked includes clearing up
With can not clear up.The history use habit of application can be specifically marked according to user, such as:When application enters 30 points of backstage
Zhong Hou, user close the application, then are labeled as " can clear up ";For another example after using entering 3 minutes from the background, user will
Using front stage operation has been switched to, then " can not clear up " is labeled as.Specifically, numerical value " 1 " expression " can clear up " can be used, uses number
Value " 0 " expression " can not clear up ", vice versa.
, can be by characteristic information that in the multidimensional characteristic information of application, unused numerical value directly represents for ease of classifying, training
Come out with specific numerical quantization, such as this characteristic information of the wireless network connection status of electronic equipment, numerical value 1 can be used
Represent normal state, abnormal state is represented with numerical value 0 (vice versa);For another example whether charged for electronic equipment
This characteristic information of state, can represent charged state with numerical value 1, and uncharged state is represented with numerical value 0 (vice versa).
Step 102, characteristic information is extracted from sample set according to preset rules, builds multiple training sets.
In one embodiment, it is default that the random extraction in ground can be put back to every time from the multidimensional characteristic information of each sample
The characteristic information of number, subsample corresponding to composition, multiple subsamples form a training set, and repeatedly after extraction, structure is multiple
Training set, preset number can be according to being actually needed self-defined value.
In one embodiment, training set can be divided into two parts, and a part is monomer sample x, and now target should for mark
With next whether using, it is that, if being otherwise labeled as 0, form can be (x if can then marki, yi), wherein yi∈ 0,
1}.Another part is triple, i.e., by sampling two sample (xi, xj), if two sample labels are consistent, 1 is designated as, label is not
Unanimously, -1 is designated as, form is (xi, xj, γ), wherein γ ∈ { 1, -1 }.
Therefore, it is above-mentioned to extract characteristic information from the sample set according to preset rules, the step of building multiple training sets
It can include:
Sample in sample set is marked, obtains the first label of each sample;
Single sample is extracted from sample set, the first training set is formed according to sample and corresponding first label, repeatedly
Extract to obtain multiple first training sets;
Two samples are extracted from sample set, according to two samples respectively corresponding to the first label generate the of two samples
Two labels, the second training set is formed according to two samples and corresponding second label, repeatedly extraction is instructed to obtain multiple second
Practice collection;
Multiple training sets are built according to multiple first training sets and the second training set.
Step 103, Logic Regression Models are trained according to multiple training sets, with the forecast model after being trained.
Logistic regression (Logistic Regression, LR) model is a kind of disaggregated model in machine learning, due to calculating
Method it is simple and efficient, in practice using very extensive.Logistic regression is mainly by constructing an important index:Generation ratio
To judge the classification of dependent variable.It introduces the concept of probability, is Y=1 event (as application can clear up) genetic definition, event
(as application can not clear up), non-genetic definition is Y=0, then the probability that event occurs is p, and the nonevent probability of event is 1-p,
P is regarded as x linear function.
In actual applications, the form of expression of Logic Regression Models has a variety of, such as, in the form of grader, according to classification
The classification capacity of device, grader can be divided into:Weak Classifier and strong classifier.So the timing that grader refers generally to is patrolled
Collect regression model.
The embodiment of the present application, corresponding Logic Regression Models can be trained using training set, be instructed accordingly
Forecast model after white silk.Neutral net in the present invention is shallow-layer neutral net, and network structure is only two layers, i.e. embeding layer and complete
Articulamentum, layer parameter is embedded in train to obtain by single sample, triple simultaneously, embeding layer is divided after full articulamentum
Class, substantially increase accuracy rate.
In one embodiment, the step of Logic Regression Models being trained according to the multiple training set, including:
The first-loss function of the Logic Regression Models is obtained according to multiple first training sets;
The second loss function of the Logic Regression Models is obtained according to multiple second training sets;
According to first-loss function and the second loss function generation target loss function, and according to target loss Function Estimation
Model parameter in the Logic Regression Models.
Wherein, above-mentioned the step of target loss function is generated according to first-loss function and the second loss function, including:
The weighted value of first-loss function and the second loss function is obtained respectively;
The weighted sum of first-loss function and the second loss function is calculated, to obtain target loss function.
After target loss function is obtained, gradient descent method can be based on target loss function is calculated, to obtain logic
Model parameter in regression model.
For example, after multiple training sets are built, for each data x in training setiCalculate a value embedded, the process
Neutral net concealed nodes are formed by one by 8 neurons to realize.
For single sample, embeding layer is done by logistic regression and classified, using classification cross entropy as loss function:
Wherein:
I, k is positive integer,It is distributed for prediction probability, NsFor training
The batch size of classification, C are classification number, yiTo characterize the one-hot encoding of sample class, W is the weight of full articulamentum;By most
The smallization loss function, training obtain embeding layer.
For triple sample (xi, xj, γ), wherein γ is the label of sampling, if being unanimously 1, inconsistent is -1, is led to
Cross COS distance:
Similarity of two nodes on embeding layer is calculated, by minimizing logistic regression loss function:
Wherein, NgTo train the batch size of triple, further training learns obtained embeding layer.
The target loss function of final optimization pass is above-mentioned two weighted sums, i.e. L=Ls+λLu, λ is weight, to adjust list
The relative scale of individual sample and triple loss function;By the gradient descent method of autoadapted learning rate, obtain final embedding
Enter layer.
Wherein, loss function (loss function) is for estimating differing for the predicted value f (x) of model and actual value Y
Cause degree, it is a non-negative real-valued function, is represented usually using L (Y, f (x)), or L (w), loss function is smaller, mould
The robustness of type is better.Loss function is the core of empirical risk function, and structure risk function important composition portion
Point.
Step 104, obtain the current multidimensional characteristic information of application program and be used as forecast sample, according to forecast sample and instruction
Forecast model generation prediction result after white silk, and management and control is carried out to application program according to prediction result.
Such as can be according to the multidimensional characteristic of predicted time acquisition applications as forecast sample.Wherein, predicted time can be with
Set according to demand, such as can be current time.For example, can be used as in the multidimensional characteristic of predicted time point acquisition applications pre-
Test sample sheet.
Above-mentioned prediction result can include cleaning or not clear up, and if desired judge whether current background application can clear up, and obtain
Take the current multidimensional degree characteristic information of application program, such as the multiple features letter of application program use information and electronic equipment currently
Breath etc., to be input to forecast model, forecast model can obtain prediction result according to model parameter calculation, so as to judge to apply journey
Whether sequence, which needs, is cleared up.
It should be noted that the training process of forecast model can also can be completed in server end at electronic equipment end.
Training process, actual prediction process when forecast model are all when server end is completed, it is necessary to use the forecast model after training
When, the characteristic information of current multiple dimensions of application program can be input to server, will after the completion of server actual prediction
Prediction result is sent to electronic equipment end, and electronic equipment is further according to the prediction result management and control application program.
Training process, actual prediction process when forecast model are all when electronic equipment end is completed, it is necessary to after using training
Forecast model when, the current multidimensional characteristic information of application program can be input to electronic equipment, electronic equipment actual prediction
After the completion of, electronic equipment is according to the prediction result management and control application program.
From the foregoing, it will be observed that the multidimensional characteristic information for the application program management-control method acquisition applications program that the embodiment of the present application provides
As sample, the sample set of application program is built, characteristic information is extracted from sample set according to preset rules, builds multiple training
Logic Regression Models are trained by collection according to the multiple training set, with the forecast model after train, are obtained and are applied journey
The current multidimensional characteristic information of sequence is simultaneously used as forecast sample, according to the forecast model generation prediction knot after forecast sample and training
Fruit, and management and control is carried out to application program according to prediction result.The application can improve the accuracy being predicted to application program,
So as to lift intellectuality and accuracy that management and control is carried out to the application program for entering backstage.
On the basis of the method that will be described below in above-described embodiment, the method for cleaning of the application is described further.Ginseng
Fig. 4 is examined, the application program management-control method includes:
201, the multidimensional characteristic information of acquisition applications program builds the sample set of application program as sample.
The multidimensional characteristic information of application has a dimension of certain length, and the parameter in each of which dimension is corresponding to characterize application
A kind of characteristic information, i.e. the multidimensional characteristic information is made up of multiple characteristic informations.The plurality of characteristic information can include application
Itself related characteristic information.Among an embodiment, 30 features of equipment can be collected, form 30 dimensional vectors, should
30 features are, for example,:
Using the duration of last incision backstage till now;
During cutting backstage till now using the last time, add up screen shut-in time length;
Using the last time duration is used on foreground;
Using the upper last time duration was used on foreground;
Using the upper last time duration was used on foreground;
Using the number for entering foreground in one day (by statistics daily);
Enter the number on foreground using (day off is separately counted by working day, day off) in one day;
Using in one day the time on foreground is in (by statistics daily);
The time on foreground is in using (day off is separately counted by working day, day off) in one day;
Intended application is daily 8:00-12:The time span that 00 this period was used;
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 0-5 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 5-10 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 10-15 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 15-20 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 15-20 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 25-30 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to after 30 minutes);
Intended application one-level type;
Intended application two-level type;
The mode that intended application is switched, it is divided into and switches by the switching of home keys, by the switching of recent keys, by other application;
Screen amount is gone out the time;
Current screen light on and off state;
Currently whether have and charging;
Current electricity;
Current wifi states;
The period index on the current time residing same day;
The background application counts gained immediately following the number that is opened after current foreground application regardless of day off on working day;
The background application divides day off on working day to count immediately following the number that is opened after current foreground application;
Current foreground, which is applied, enters backstage to intended application into foreground by the Mean Time Between Replacement counted daily;
Current foreground, which is applied, enters backstage to intended application into when extinguishing during foreground by the average screen counted daily
Between.
202, characteristic information is extracted from sample set according to preset rules, builds multiple training sets.
In one embodiment, training set can be divided into two parts, and a part is monomer sample x, and now target should for mark
With next whether using, it is that, if being otherwise labeled as 0, form can be (x if can then marki, yi), wherein yi∈ 0,
1}.Another part is triple, i.e., by sampling two sample (xi, xj), if two sample labels are consistent, 1 is designated as, label is not
Unanimously, -1 is designated as, form is (xi, xj, γ), wherein γ ∈ { 1, -1 }.
203, Logic Regression Models are trained according to multiple training sets, with the forecast model after being trained.
The embodiment of the present application, corresponding Logic Regression Models can be trained using training set, be instructed accordingly
Forecast model after white silk.Neutral net in the present invention is shallow-layer neutral net, and network structure is only two layers, i.e. embeding layer and complete
Articulamentum, layer parameter is embedded in train to obtain by single sample, triple simultaneously, embeding layer is divided after full articulamentum
Class, substantially increase accuracy rate.
204, obtain the current multidimensional characteristic information of application program and be used as forecast sample, after forecast sample and training
Forecast model generation application program the first probability that can be cleared up and the second probability that can not be cleared up.
According to Logic Regression Models after forecast set sample and its corresponding training, corresponding prediction probability is exported, is obtained more
Individual prediction probability.What one Logic Regression Models output, one the first probability that can be cleared up comprising application and application can not clear up
The prediction probability of second probability.
205, the first probability that can be cleared up application program obtains comparing knot compared with the second probability that can not be cleared up
Fruit.
206, second that exports application program the first prediction result that can be cleared up according to comparative result or can not clear up is pre-
Survey result.
Specifically, when the first probability is more than the second probability, export the first prediction result that can be cleared up, when the first probability not
During more than the second probability, the second prediction result that can not be cleared up is exported.
Such as some prediction probability P, it is false if Y=1 represents that application can clear up, Y=0 represents that application can not clear up
If P (Y=1 | x) is more than P (Y=0 | x), now, the first prediction result that output prediction application can clear up;Assuming that P (Y=1 | x)
No more than P (Y=0 | x), now, the second prediction result that output prediction application can not clear up.
207, according to the quantity of the first prediction result and the quantity of the second prediction result, determine whether application program can be clear
Reason.
When the quantity of the first prediction result is more than the quantity of the second prediction result, it is determined that application can clear up;
When the quantity of the first prediction result is not more than the quantity of the second prediction result, it is determined that using can not clear up.
In a specific example, it can utilize the multiple of Logic Regression Models prediction running background of training in advance should
With that whether can clear up, as shown in table 1, it is determined that the application A1 of running background can be cleared up and using A3, and keep existing using A2
The state of running background is constant.
Using | Prediction result |
Using A1 | It can clear up |
Using A2 | It can not clear up |
Using A3 | It can clear up |
Table 1
From the foregoing, it will be observed that the multidimensional characteristic information for the application program management-control method acquisition applications program that the embodiment of the present application provides
As sample, the sample set of application program is built, characteristic information is extracted from sample set according to preset rules, builds multiple training
Logic Regression Models are trained by collection according to multiple training sets, and with the forecast model after train, obtaining application program ought
Preceding multidimensional characteristic information is simultaneously used as forecast sample, and prediction result is generated according to the forecast model after forecast sample and training, and
Management and control is carried out to application program according to prediction result.The application can improve the accuracy being predicted to application program, so as to
Lift intellectuality and the accuracy that management and control is carried out to the application program for entering backstage.
Referring to Fig. 5, Fig. 5 is the another application scene signal for the application program control device that the embodiment of the present application provides
Figure.When the training process of forecast model is completed in server end, the actual prediction process of forecast model is completed at electronic equipment end
When, it is necessary to using optimization after forecast model when, the current multidimensional characteristic information of application program can be input to electronic equipment,
After the completion of electronic equipment actual prediction, electronic equipment is according to the prediction result management and control application program.Optionally, will can train
Forecast model file (model files) be transplanted on smart machine, if desired judge current background application whether can clear up, more
New current sample set, is input to the forecast model file (model files) trained, calculates and can obtain predicted value.
In certain embodiments, before the step of acquisition application program current multidimensional characteristic information, can also include:
Detect whether application program enters backstage, if entering backstage, obtain the current multidimensional characteristic information of application program.
Then input prediction model generation prediction result, and management and control is carried out to application program according to prediction result.
In certain embodiments, before the step of acquisition application program current multidimensional characteristic information, can also include:
Preset time is obtained, if present system time reaches preset time, it is special to obtain the current multidimensional of application program
Reference ceases.Wherein preset time can be a time point in one day, such as at 9 points in the morning, or during several in one day
Between point, such as at 9 points in the morning, 6 pm.It can also be one or several time points in more days.Then input prediction model is given birth to
Management and control is carried out to application program into prediction result, and according to prediction result.
Above-mentioned all technical schemes, any combination can be used to form the alternative embodiment of the application, it is not another herein
One repeats.
Referring to Fig. 6, Fig. 6 is the structural representation for the application program control device that the embodiment of the present application provides.Wherein should
Application program control device 300 is applied to electronic equipment, and the application program control device 300 includes acquisition module 301, structure mould
Block 302, training module 303 and management and control module 304.
Wherein, acquisition module 301, the multidimensional characteristic information for acquisition applications program build application program as sample
Sample set.
Wherein, default application program can be mounted in any application in electronic equipment, such as communication applications journey
Sequence, multimedia application, game application, information application program or shopping application program etc..
The multidimensional characteristic information of application has a dimension of certain length, and the parameter in each of which dimension is corresponding to characterize application
A kind of characteristic information, i.e. the multidimensional characteristic information is made up of multiple characteristic informations.The plurality of characteristic information can include application
Itself related characteristic information.Among an embodiment, 30 features of equipment can be collected, form 30 dimensional vectors, should
30 features are, for example,:
Using the duration of last incision backstage till now;
During cutting backstage till now using the last time, add up screen shut-in time length;
Using the last time duration is used on foreground;
Using the upper last time duration was used on foreground;
Using the upper last time duration was used on foreground;
Using the number for entering foreground in one day (by statistics daily);
Enter the number on foreground using (day off is separately counted by working day, day off) in one day;
Using in one day the time on foreground is in (by statistics daily);
The time on foreground is in using (day off is separately counted by working day, day off) in one day;
Intended application is daily 8:00-12:The time span that 00 this period was used;
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 0-5 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 5-10 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 10-15 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 15-20 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 15-20 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to 25-30 minutes);
Intended application is in first bin of backstage dwell histogram (number accounting corresponding to after 30 minutes);
Intended application one-level type;
Intended application two-level type;
The mode that intended application is switched, it is divided into and switches by the switching of home keys, by the switching of recent keys, by other application;
Screen amount is gone out the time;
Current screen light on and off state;
Currently whether have and charging;
Current electricity;
Current wifi states;
The period index on the current time residing same day;
The background application counts gained immediately following the number that is opened after current foreground application regardless of day off on working day;
The background application divides day off on working day to count immediately following the number that is opened after current foreground application;
Current foreground, which is applied, enters backstage to intended application into foreground by the Mean Time Between Replacement counted daily;
Current foreground, which is applied, enters backstage to intended application into when extinguishing during foreground by the average screen counted daily
Between.
Module 302 is built, for extracting characteristic information from sample set according to preset rules, builds multiple training sets.
In one embodiment, it is default that the random extraction in ground can be put back to every time from the multidimensional characteristic information of each sample
The characteristic information of number, subsample corresponding to composition, multiple subsamples form a training set, and repeatedly after extraction, structure is multiple
Training set, preset number can be according to being actually needed self-defined value.
In one embodiment, training set can be divided into two parts, and a part is monomer sample x, and now target should for mark
With next whether using, it is that, if being otherwise labeled as 0, form can be (x if can then marki, yi), wherein yi∈ 0,
1}.Another part is triple, i.e., by sampling two sample (xi, xj), if two sample labels are consistent, 1 is designated as, label is not
Unanimously, -1 is designated as, form is (xi, xj, γ), wherein γ ∈ { 1, -1 }.
Training module 303, for being trained according to multiple training sets to Logic Regression Models, with pre- after being trained
Survey model.
Logistic regression (Logistic Regression, LR) model is a kind of disaggregated model in machine learning, due to calculating
Method it is simple and efficient, in practice using very extensive.Logistic regression is mainly by constructing an important index:Generation ratio
To judge the classification of dependent variable.It introduces the concept of probability, is Y=1 event (as application can clear up) genetic definition, event
(as application can not clear up), non-genetic definition is Y=0, then the probability that event occurs is p, and the nonevent probability of event is 1-p,
P is regarded as x linear function.
In actual applications, the form of expression of Logic Regression Models has a variety of, such as, in the form of grader, according to classification
The classification capacity of device, grader can be divided into:Weak Classifier and strong classifier.So the timing that grader refers generally to is patrolled
Collect regression model.
The embodiment of the present application, corresponding Logic Regression Models can be trained using training set, be instructed accordingly
Forecast model after white silk.Neutral net in the present invention is shallow-layer neutral net, and network structure is only two layers, i.e. embeding layer and complete
Articulamentum, layer parameter is embedded in train to obtain by single sample, triple simultaneously, embeding layer is divided after full articulamentum
Class, substantially increase accuracy rate.
In one embodiment, after multiple training sets are built, for each data x in training setiCalculate an insertion
Value, the process form neutral net concealed nodes by 8 neurons by one and realized.
For single sample, embeding layer is done by logistic regression and classified, using classification cross entropy as loss function:
Wherein:
NsFor the batch size of training classification, C is classification number, yiFor
The one-hot encoding of sample class is characterized, W is the weight of full articulamentum;By minimizing the loss function, training obtains embeding layer.
For triple sample (xi, xj, γ), wherein γ is the label of sampling, if being unanimously 1, inconsistent is -1, is led to
Cross COS distance:
Similarity of two nodes on embeding layer is calculated, by minimizing logistic regression loss function:
Wherein, NgTo train the batch size of triple, further training learns obtained embeding layer.
The target loss function of final optimization pass is above-mentioned two weighted sums, i.e. L=Ls+λLu, λ is weight, to adjust list
The relative scale of individual sample and triple loss function;By the gradient descent method of autoadapted learning rate, obtain final embedding
Enter layer.
Management and control module 304, for obtaining the current multidimensional characteristic information of application program and being used as forecast sample, according to prediction
Forecast model generation prediction result after sample and training, and management and control is carried out to application program according to prediction result.
Such as can be according to the multidimensional characteristic of predicted time acquisition applications as forecast sample.Wherein, predicted time can be with
Set according to demand, such as can be current time.For example, can be used as in the multidimensional characteristic of predicted time point acquisition applications pre-
Test sample sheet.
Above-mentioned prediction result can include cleaning or not clear up, and if desired judge whether current background application can clear up, and obtain
Take the current multidimensional degree characteristic information of application program, such as the multiple features letter of application program use information and electronic equipment currently
Breath etc., to be input to forecast model, forecast model can obtain prediction result according to model parameter calculation, so as to judge to apply journey
Whether sequence, which needs, is cleared up.
It should be noted that the training process of forecast model can also can be completed in server end at electronic equipment end.
Training process, actual prediction process when forecast model are all when server end is completed, it is necessary to use the forecast model after training
When, the characteristic information of current multiple dimensions of application program can be input to server, will after the completion of server actual prediction
Prediction result is sent to electronic equipment end, and electronic equipment is further according to the prediction result management and control application program.
Training process, actual prediction process when forecast model are all when electronic equipment end is completed, it is necessary to after using training
Forecast model when, the current multidimensional characteristic information of application program can be input to electronic equipment, electronic equipment actual prediction
After the completion of, electronic equipment is according to the prediction result management and control application program.
Referring to Fig. 7, above-mentioned structure module 302 can specifically include:
Submodule 3021 is marked, for the sample in sample set to be marked, obtains the first label of each sample;
First extracting sub-module 3022, for extracting single sample from sample set, according to sample and corresponding first
Label forms the first training set, repeatedly extracts to obtain multiple first training sets;
Second extracting sub-module 3023, for extracting two samples from sample set, according to corresponding to two samples difference
First label generates the second label of two samples, and the second training set is formed according to two samples and corresponding second label,
Repeatedly extract to obtain multiple second training sets;
Submodule 3024 is built, for building multiple training sets according to multiple first training sets and the second training set.
With continued reference to Fig. 8, above-mentioned training module 303 specifically includes:
First function acquisition submodule 3031, for obtaining the first damage of Logic Regression Models according to multiple first training sets
Lose function;
Second function acquisition submodule 3032, for obtaining the second damage of Logic Regression Models according to multiple second training sets
Lose function;
Parameter estimation sub-module 3033, for generating target loss letter according to first-loss function and the second loss function
Number, and the model parameter in target loss Function Estimation Logic Regression Models.
In one embodiment, parameter estimation sub-module 3033 is specifically used for obtaining first-loss function and the second loss respectively
The weighted value of function, the weighted sum of first-loss function and the second loss function is calculated, to obtain target loss function.
Above-mentioned parameter estimates submodule 3033, and target loss function is calculated also particularly useful for based on gradient descent method, with
Obtain the model parameter in Logic Regression Models.
In one embodiment, prediction result includes:The first probability that application program can clear up and it can not clear up second general
Rate, management and control module 304, including:
Output sub-module, the first probability for that can clear up application program compare with the second probability that can not be cleared up
Compared with obtaining comparative result, application program the first prediction result that can be cleared up or can not clear up the exported according to comparative result
Two prediction results;
Determination sub-module, for the quantity according to the first prediction result and the quantity of the second prediction result, it is determined that described should
Whether can be cleared up with program.
Wherein, above-mentioned output sub-module, specifically for when the first probability is more than second probability, exporting what can be cleared up
First prediction result;
When the first probability is not more than second probability, the second prediction result that can not be cleared up is exported.
Above-mentioned all technical schemes, any combination can be used to form the alternative embodiment of the application, it is not another herein
One repeats.
From the foregoing, the application program control device of the embodiment of the present application, passes through the multidimensional characteristic of acquisition applications program
Information builds the sample set of application program, characteristic information is extracted from sample set according to preset rules, structure is multiple as sample
Training set, Logic Regression Models are trained according to multiple training sets, with the forecast model after being trained, journey is applied in acquisition
The current multidimensional characteristic information of sequence is simultaneously used as forecast sample, according to the forecast model generation prediction knot after forecast sample and training
Fruit, and management and control is carried out to application program according to prediction result.The application can improve the accuracy being predicted to application program,
So as to lift intellectuality and accuracy that management and control is carried out to the application program for entering backstage.
In the embodiment of the present application, application program control device belongs to same with the application program management-control method in foregoing embodiments
One design, can run the either method provided in application program management-control method embodiment on application program control device, its
Specific implementation process refers to the embodiment of application program management-control method, and here is omitted.
The embodiment of the present application also provides a kind of electronic equipment.Referring to Fig. 9, electronic equipment 400 include processor 401 and
Memory 402.Wherein, processor 401 is electrically connected with memory 402.
Processor 400 is the control centre of electronic equipment 400, utilizes various interfaces and the whole electronic equipment of connection
Various pieces, by the computer program of operation or load store in memory 402, and call and be stored in memory 402
Interior data, the various functions and processing data of electronic equipment 400 are performed, so as to carry out integral monitoring to electronic equipment 400.
Memory 402 can be used for storage software program and module, and processor 401 is stored in memory 402 by operation
Computer program and module, so as to perform various function application and data processing.Memory 402 can mainly include storage
Program area and storage data field, wherein, storing program area can storage program area, the computer program needed at least one function
(such as sound-playing function, image player function etc.) etc.;Storage data field can store to be created according to using for electronic equipment
Data etc..In addition, memory 402 can include high-speed random access memory, nonvolatile memory, example can also be included
Such as at least one disk memory, flush memory device or other volatile solid-state parts.Correspondingly, memory 402 may be used also
With including Memory Controller, to provide access of the processor 401 to memory 402.
In the embodiment of the present application, the processor 401 in electronic equipment 400 can be according to the steps, by one or one
Instruction is loaded into memory 402 corresponding to the process of computer program more than individual, and is stored in by the operation of processor 401
Computer program in reservoir 402, it is as follows so as to realize various functions:
The multidimensional characteristic information of acquisition applications program builds the sample set of application program, according to preset rules as sample
Characteristic information is extracted from sample set, builds multiple training sets, Logic Regression Models are instructed according to the multiple training set
Practice, with the forecast model after train, the current multidimensional characteristic information of acquisition application program is simultaneously used as forecast sample, according to pre-
Forecast model generation prediction result after test sample sheet and training, and management and control is carried out to application program according to prediction result.The application
The accuracy being predicted to application program can be improved, so as to lift the intelligence that management and control is carried out to the application program for entering backstage
Change and accuracy.
Also referring to Figure 10, in some embodiments, electronic equipment 400 can also include:Display 403, radio frequency
Circuit 404, voicefrequency circuit 405 and power supply 406.Wherein, wherein, display 403, radio circuit 404, voicefrequency circuit 405 with
And power supply 406 is electrically connected with processor 401 respectively.
Display 403 is displayed for the information inputted by user or the information for being supplied to user and various figures are used
Family interface, these graphical user interface can be made up of figure, text, icon, video and its any combination.Display 403
Display panel can be included, in some embodiments, can use liquid crystal display (Liquid Crystal Display,
LCD) or the form such as Organic Light Emitting Diode (Organic Light-Emitting Diode, OLED) configures display surface
Plate.
Radio circuit 404 can be used for transceiving radio frequency signal, to be set by radio communication and the network equipment or other electronics
It is standby to establish wireless telecommunications, the receiving and transmitting signal between the network equipment or other electronic equipments.
Voicefrequency circuit 405 can be used for providing the COBBAIF between user and electronic equipment by loudspeaker, microphone.
Power supply 406 is used to all parts power supply of electronic equipment 400.In certain embodiments, power supply 406 can be with
It is logically contiguous by power-supply management system and processor 401, so as to by power-supply management system realize management charging, electric discharge, with
And the function such as power managed.
Although not shown in Figure 10, electronic equipment 400 can also include camera, bluetooth module etc., will not be repeated here.
The embodiment of the present application also provides a kind of storage medium, and storage medium is stored with computer program, works as computer program
When running on computers so that computer performs the application program management-control method in any of the above-described embodiment.
In the embodiment of the present application, storage medium can be magnetic disc, CD, read-only storage (Read Only Memory,
) or random access memory (Random Access Memory, RAM) etc. ROM.
In the above-described embodiments, the description to each embodiment all emphasizes particularly on different fields, and does not have the portion being described in detail in some embodiment
Point, it may refer to the associated description of other embodiment.
It should be noted that for the application program management-control method of the embodiment of the present application, this area common test personnel
It is that can pass through computer program it is appreciated that realizing all or part of flow of the embodiment of the present application application program management-control method
To control the hardware of correlation to complete, computer program can be stored in a computer read/write memory medium, be such as stored in electricity
In the memory of sub- equipment, and by least one computing device in the electronic equipment, it may include in the process of implementation as should
With the flow of the embodiment of program management-control method.Wherein, storage medium can be magnetic disc, CD, read-only storage, arbitrary access
Memory body etc..
For the application program control device of the embodiment of the present application, its each functional module can be integrated in a processing core
In piece or modules are individually physically present, can also two or more modules be integrated in a module.On
Stating integrated module can both be realized in the form of hardware, can also be realized in the form of software function module.Integrated
If module is realized in the form of software function module and as independent production marketing or in use, can also be stored in one
In computer read/write memory medium, storage medium is for example read-only storage, disk or CD etc..
A kind of application program management-control method, device, storage medium and the electronics provided above the embodiment of the present application is set
Standby to be described in detail, specific case used herein is set forth to the principle and embodiment of the application, the above
The explanation of embodiment is only intended to help and understands the present processes and its core concept;Meanwhile for those skilled in the art
Member, according to the thought of the application, there will be changes in specific embodiments and applications, in summary, this explanation
Book content should not be construed as the limitation to the application.
Claims (17)
1. a kind of application program management-control method, it is characterised in that the described method comprises the following steps:
The multidimensional characteristic information of acquisition applications program builds the sample set of the application program as sample;
Characteristic information is extracted from the sample set according to preset rules, builds multiple training sets;
Logic Regression Models are trained according to the multiple training set, with the forecast model after being trained;
Obtain the current multidimensional characteristic information of the application program and be used as forecast sample, after the forecast sample and training
Forecast model generation prediction result, and according to the prediction result to the application program carry out management and control.
2. application program management-control method according to claim 1, it is characterised in that it is described according to preset rules from the sample
This concentration extracts characteristic information, and the step of building multiple training sets includes:
Sample in the sample set is marked, obtains the first label of each sample;
Single sample is extracted from the sample set, the first training set is formed according to the sample and corresponding first label,
Repeatedly extract to obtain multiple first training sets;
Two samples are extracted from the sample set, it is described two according to the first label generation corresponding to described two samples difference
Second label of sample, the second training set is formed according to described two samples and corresponding second label, repeatedly extracted to obtain
To multiple second training sets;
Multiple training sets are built according to the multiple first training set and second training set.
3. application program management-control method according to claim 1, it is characterised in that described according to the multiple training set pair
The step of Logic Regression Models are trained, including:
The first-loss function of the Logic Regression Models is obtained according to the multiple first training set;
The second loss function of the Logic Regression Models is obtained according to the multiple second training set;
According to the first-loss function and the second loss function generation target loss function, and according to the target loss function
Estimate the model parameter in the Logic Regression Models.
4. application program management-control method according to claim 3, it is characterised in that according to preset formula and the multiple
First training set obtains the first-loss function of the Logic Regression Models.Wherein described preset formula is:
<mrow>
<msub>
<mi>L</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>N</mi>
<mi>s</mi>
</msub>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>N</mi>
<mi>s</mi>
</msub>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>C</mi>
</munderover>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mi>log</mi>
<mrow>
<mo>(</mo>
<msub>
<mover>
<mi>y</mi>
<mo>~</mo>
</mover>
<mrow>
<mi>i</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
</mrow>
Wherein:
I, k is positive integer,It is distributed for prediction probability, NsFor the batch size of training classification, C is classification number, yiTo characterize
The one-hot encoding of sample class.
5. application program management-control method according to claim 3, it is characterised in that according to the first-loss function and
Two loss functions generate the step of target loss function, including:
The weighted value of the first-loss function and the second loss function is obtained respectively;
The weighted sum of the first-loss function and the second loss function is calculated, to obtain the target loss function.
6. application program management-control method according to claim 3, it is characterised in that according to the target loss Function Estimation
The step of model parameter in the Logic Regression Models, including:
The target loss function is calculated based on gradient descent method, to obtain the model parameter in the Logic Regression Models.
7. application program management-control method according to claim 1, it is characterised in that the prediction result includes:It is described to answer
The first probability that can be cleared up with program and the second probability that can not be cleared up;
The step of management and control is carried out to the application program according to the prediction result, including:
The first probability that can be cleared up application program obtains comparative result compared with the second probability that can not be cleared up;
Application program the first prediction result that can be cleared up or the second prediction result that can not be cleared up are exported according to comparative result;
According to the quantity of the first prediction result and the quantity of the second prediction result, determine whether the application program can clear up.
8. application program management-control method according to claim 7, it is characterised in that application program is exported according to comparative result
The step of the first prediction result that can be cleared up or the second prediction result that can not be cleared up, including:
When first probability is more than second probability, the first prediction result that can be cleared up is exported;
When first probability is not more than second probability, the second prediction result that can not be cleared up is exported.
9. a kind of application program control device, it is characterised in that described device includes:
Acquisition module, the multidimensional characteristic information for acquisition applications program build the sample set of the application program as sample;
Module is built, for extracting characteristic information from the sample set according to preset rules, builds multiple training sets;
Training module, for being trained according to the multiple training set to Logic Regression Models, with the prediction after being trained
Model;
Management and control module, for obtaining the current multidimensional characteristic information of the application program and being used as forecast sample, according to described pre-
Forecast model generation prediction result after test sample sheet and training, and pipe is carried out to the application program according to the prediction result
Control.
10. application program control device according to claim 9, it is characterised in that the structure module specifically includes:
Submodule is marked, for the sample in the sample set to be marked, obtains the first label of each sample;
First extracting sub-module, for extracting single sample from the sample set, according to the sample and corresponding first
Label forms the first training set, repeatedly extracts to obtain multiple first training sets;
Second extracting sub-module, for extracting two samples from the sample set, according to corresponding to described two samples difference
First label generates the second label of described two samples, and second is formed according to described two samples and corresponding second label
Training set, repeatedly extract to obtain multiple second training sets;
Submodule is built, for building multiple training sets according to the multiple first training set and second training set.
11. application program control device according to claim 9, it is characterised in that the training module specifically includes:
First function acquisition submodule, for obtaining the first damage of the Logic Regression Models according to the multiple first training set
Lose function;
Second function acquisition submodule, for obtaining the second damage of the Logic Regression Models according to the multiple second training set
Lose function;
Parameter estimation sub-module, for generating target loss function according to the first-loss function and the second loss function, and
According to the model parameter in Logic Regression Models described in the target loss Function Estimation.
12. application program control device according to claim 9, it is characterised in that
The parameter estimation sub-module, the weight specifically for obtaining the first-loss function and the second loss function respectively
Value, the weighted sum of the first-loss function and the second loss function is calculated, to obtain the target loss function.
13. application program control device according to claim 11, it is characterised in that
The parameter estimation sub-module, the target loss function is calculated also particularly useful for based on gradient descent method, to obtain
Model parameter in the Logic Regression Models.
14. application program control device according to claim 9, it is characterised in that the prediction result includes:It is described to answer
The first probability that can be cleared up with program and the second probability that can not be cleared up, the management and control module, including:
Output sub-module, the first probability for that can clear up application program obtain compared with the second probability that can not be cleared up
To comparative result, second that exports application program the first prediction result that can be cleared up according to comparative result or can not clear up is pre-
Survey result;
Determination sub-module, for the quantity according to the first prediction result and the quantity of the second prediction result, determine described to apply journey
Whether sequence can clear up.
15. application program control device according to claim 14, it is characterised in that
The output sub-module, specifically for when first probability is more than second probability, exporting can clear up first
Prediction result;
When first probability is not more than second probability, the second prediction result that can not be cleared up is exported.
16. a kind of storage medium, is stored thereon with computer program, it is characterised in that when the computer program is in computer
During upper operation so that the computer performs the application program management-control method as described in any one of claim 1 to 7.
17. a kind of electronic equipment, including processor and memory, the memory have computer program, it is characterised in that described
Processor is by calling the computer program, for performing the application program management and control side as described in any one of claim 1 to 7
Method.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710940355.2A CN107678845B (en) | 2017-09-30 | 2017-09-30 | Application program control method and device, storage medium and electronic equipment |
PCT/CN2018/102254 WO2019062414A1 (en) | 2017-09-30 | 2018-08-24 | Method and apparatus for managing and controlling application program, storage medium, and electronic device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710940355.2A CN107678845B (en) | 2017-09-30 | 2017-09-30 | Application program control method and device, storage medium and electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107678845A true CN107678845A (en) | 2018-02-09 |
CN107678845B CN107678845B (en) | 2020-03-10 |
Family
ID=61140234
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710940355.2A Active CN107678845B (en) | 2017-09-30 | 2017-09-30 | Application program control method and device, storage medium and electronic equipment |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN107678845B (en) |
WO (1) | WO2019062414A1 (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019062417A1 (en) * | 2017-09-30 | 2019-04-04 | Oppo广东移动通信有限公司 | Application cleaning method and apparatus, storage medium and electronic device |
WO2019062414A1 (en) * | 2017-09-30 | 2019-04-04 | Oppo广东移动通信有限公司 | Method and apparatus for managing and controlling application program, storage medium, and electronic device |
CN110163460A (en) * | 2018-03-30 | 2019-08-23 | 腾讯科技(深圳)有限公司 | A kind of method and apparatus determined using score value |
CN110796513A (en) * | 2019-09-25 | 2020-02-14 | 北京三快在线科技有限公司 | Multitask learning method and device, electronic equipment and storage medium |
CN111797880A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | Data processing method, data processing device, storage medium and electronic equipment |
CN112130991A (en) * | 2020-08-28 | 2020-12-25 | 北京思特奇信息技术股份有限公司 | Application program control method and system based on machine learning |
CN112770002A (en) * | 2019-10-17 | 2021-05-07 | 荣耀终端有限公司 | Heartbeat control method and electronic equipment |
CN113325998A (en) * | 2020-02-29 | 2021-08-31 | 杭州海康存储科技有限公司 | Read-write speed control method and device |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111797861A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | Information processing method, information processing apparatus, storage medium, and electronic device |
CN111797423A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | Model training method, data authorization method, device, storage medium and equipment |
CN110232403B (en) * | 2019-05-15 | 2024-02-27 | 腾讯科技(深圳)有限公司 | Label prediction method and device, electronic equipment and medium |
CN110442516B (en) * | 2019-07-12 | 2024-02-09 | 未鲲(上海)科技服务有限公司 | Information processing method, apparatus, and computer-readable storage medium |
CN112396445A (en) * | 2019-08-16 | 2021-02-23 | 京东数字科技控股有限公司 | Method and device for identifying user identity information |
CN112651534A (en) * | 2019-10-10 | 2021-04-13 | 顺丰科技有限公司 | Method, device and storage medium for predicting resource supply chain demand |
CN111124632B (en) * | 2019-12-06 | 2024-02-13 | 西安易朴通讯技术有限公司 | Optimization method and device of mobile terminal, terminal equipment and storage medium |
CN113034260A (en) * | 2019-12-09 | 2021-06-25 | 中国移动通信有限公司研究院 | Credit evaluation method, model construction method, display method and related equipment |
CN111144473B (en) * | 2019-12-23 | 2024-04-23 | 中国医学科学院肿瘤医院 | Training set construction method, training set construction device, electronic equipment and computer readable storage medium |
CN111460150B (en) * | 2020-03-27 | 2023-11-10 | 北京小米松果电子有限公司 | Classification model training method, classification method, device and storage medium |
CN111949530B (en) * | 2020-08-07 | 2024-02-20 | 北京灵汐科技有限公司 | Test result prediction method and device, computer equipment and storage medium |
CN112506556B (en) * | 2020-11-19 | 2023-08-25 | 杭州云深科技有限公司 | Application program classification method, device, computer equipment and storage medium |
CN113342335A (en) * | 2021-05-11 | 2021-09-03 | 北京大学 | Fast application page selection method, device, equipment and storage medium |
CN113239799A (en) * | 2021-05-12 | 2021-08-10 | 北京沃东天骏信息技术有限公司 | Training method, recognition method, device, electronic equipment and readable storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104102917A (en) * | 2014-07-03 | 2014-10-15 | 中国石油大学(北京) | Construction method of domain self-adaptive classifier, construction device for domain self-adaptive classifier, data classification method and data classification device |
CN106096538A (en) * | 2016-06-08 | 2016-11-09 | 中国科学院自动化研究所 | Face identification method based on sequencing neural network model and device |
CN106093612A (en) * | 2016-05-26 | 2016-11-09 | 国网江苏省电力公司电力科学研究院 | A kind of method for diagnosing fault of power transformer |
CN106993083A (en) * | 2017-02-21 | 2017-07-28 | 北京奇虎科技有限公司 | A kind of method and apparatus of recommendation intelligent terminal operation prompt information |
CN107133094A (en) * | 2017-06-05 | 2017-09-05 | 努比亚技术有限公司 | Application management method, mobile terminal and computer-readable recording medium |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104298549B (en) * | 2014-09-30 | 2018-03-30 | 北京金山安全软件有限公司 | Method and device for cleaning application programs in mobile terminal and mobile terminal |
CN104991803B (en) * | 2015-07-10 | 2018-04-06 | 上海斐讯数据通信技术有限公司 | To the managing and control system and method for the self-starting under given conditions of android application programs |
CN107133436A (en) * | 2016-02-26 | 2017-09-05 | 阿里巴巴集团控股有限公司 | A kind of multiple sample model training method and device |
CN107678799B (en) * | 2017-09-30 | 2019-10-25 | Oppo广东移动通信有限公司 | Application program management-control method, device, storage medium and electronic equipment |
CN107678845B (en) * | 2017-09-30 | 2020-03-10 | Oppo广东移动通信有限公司 | Application program control method and device, storage medium and electronic equipment |
CN107704364B (en) * | 2017-09-30 | 2019-08-06 | Oppo广东移动通信有限公司 | Background application management-control method, device, storage medium and electronic equipment |
CN107643948B (en) * | 2017-09-30 | 2020-06-02 | Oppo广东移动通信有限公司 | Application program control method, device, medium and electronic equipment |
-
2017
- 2017-09-30 CN CN201710940355.2A patent/CN107678845B/en active Active
-
2018
- 2018-08-24 WO PCT/CN2018/102254 patent/WO2019062414A1/en active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104102917A (en) * | 2014-07-03 | 2014-10-15 | 中国石油大学(北京) | Construction method of domain self-adaptive classifier, construction device for domain self-adaptive classifier, data classification method and data classification device |
CN106093612A (en) * | 2016-05-26 | 2016-11-09 | 国网江苏省电力公司电力科学研究院 | A kind of method for diagnosing fault of power transformer |
CN106096538A (en) * | 2016-06-08 | 2016-11-09 | 中国科学院自动化研究所 | Face identification method based on sequencing neural network model and device |
CN106993083A (en) * | 2017-02-21 | 2017-07-28 | 北京奇虎科技有限公司 | A kind of method and apparatus of recommendation intelligent terminal operation prompt information |
CN107133094A (en) * | 2017-06-05 | 2017-09-05 | 努比亚技术有限公司 | Application management method, mobile terminal and computer-readable recording medium |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019062417A1 (en) * | 2017-09-30 | 2019-04-04 | Oppo广东移动通信有限公司 | Application cleaning method and apparatus, storage medium and electronic device |
WO2019062414A1 (en) * | 2017-09-30 | 2019-04-04 | Oppo广东移动通信有限公司 | Method and apparatus for managing and controlling application program, storage medium, and electronic device |
CN110163460A (en) * | 2018-03-30 | 2019-08-23 | 腾讯科技(深圳)有限公司 | A kind of method and apparatus determined using score value |
CN110163460B (en) * | 2018-03-30 | 2023-09-19 | 腾讯科技(深圳)有限公司 | Method and equipment for determining application score |
CN111797880A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | Data processing method, data processing device, storage medium and electronic equipment |
CN110796513A (en) * | 2019-09-25 | 2020-02-14 | 北京三快在线科技有限公司 | Multitask learning method and device, electronic equipment and storage medium |
CN112770002A (en) * | 2019-10-17 | 2021-05-07 | 荣耀终端有限公司 | Heartbeat control method and electronic equipment |
CN112770002B (en) * | 2019-10-17 | 2022-04-19 | 荣耀终端有限公司 | Heartbeat control method and electronic equipment |
CN113325998A (en) * | 2020-02-29 | 2021-08-31 | 杭州海康存储科技有限公司 | Read-write speed control method and device |
CN112130991A (en) * | 2020-08-28 | 2020-12-25 | 北京思特奇信息技术股份有限公司 | Application program control method and system based on machine learning |
Also Published As
Publication number | Publication date |
---|---|
CN107678845B (en) | 2020-03-10 |
WO2019062414A1 (en) | 2019-04-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107678845A (en) | Application program management-control method, device, storage medium and electronic equipment | |
CN107678799B (en) | Application program management-control method, device, storage medium and electronic equipment | |
CN108337358A (en) | Using method for cleaning, device, storage medium and electronic equipment | |
CN110032632A (en) | Intelligent customer service answering method, device and storage medium based on text similarity | |
CN107678800A (en) | Background application method for cleaning, device, storage medium and electronic equipment | |
CN107077489A (en) | Automatic for multidimensional data is seen clearly | |
CN107894827A (en) | Using method for cleaning, device, storage medium and electronic equipment | |
CN108108455A (en) | Method for pushing, device, storage medium and the electronic equipment of destination | |
CN110019790A (en) | Text identification, text monitoring, data object identification, data processing method | |
CN110457481A (en) | A kind of method, apparatus, equipment and the storage medium of disaggregated model training | |
CN108228325A (en) | Application management method and device, electronic equipment, computer storage media | |
CN107870810B (en) | Application cleaning method and device, storage medium and electronic equipment | |
CN107704289A (en) | Using method for cleaning, device, storage medium and electronic equipment | |
CN107943534A (en) | Method for closing, device, storage medium and the electronic equipment of background application | |
CN107678531A (en) | Using method for cleaning, device, storage medium and electronic equipment | |
CN107807730B (en) | Using method for cleaning, device, storage medium and electronic equipment | |
CN110046981A (en) | A kind of credit estimation method, device and storage medium | |
CN107943582A (en) | Characteristic processing method, apparatus, storage medium and electronic equipment | |
CN108197225A (en) | Sorting technique, device, storage medium and the electronic equipment of image | |
CN110458296A (en) | The labeling method and device of object event, storage medium and electronic device | |
CN107885544A (en) | Application program management-control method, device, medium and electronic equipment | |
CN109934194A (en) | Picture classification method, edge device, system and storage medium | |
CN107643925A (en) | Background application method for cleaning, device, storage medium and electronic equipment | |
CN107748697A (en) | Using method for closing, device, storage medium and electronic equipment | |
CN109961075A (en) | User gender prediction method, apparatus, medium and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB02 | Change of applicant information |
Address after: 523860 No. 18, Wu Sha Beach Road, Changan Town, Dongguan, Guangdong Applicant after: OPPO Guangdong Mobile Communications Co., Ltd. Address before: 523860 No. 18, Wu Sha Beach Road, Changan Town, Dongguan, Guangdong Applicant before: Guangdong OPPO Mobile Communications Co., Ltd. |
|
CB02 | Change of applicant information | ||
GR01 | Patent grant | ||
GR01 | Patent grant |