CN109635923A - Method and apparatus for handling data - Google Patents
Method and apparatus for handling data Download PDFInfo
- Publication number
- CN109635923A CN109635923A CN201811384897.7A CN201811384897A CN109635923A CN 109635923 A CN109635923 A CN 109635923A CN 201811384897 A CN201811384897 A CN 201811384897A CN 109635923 A CN109635923 A CN 109635923A
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- sequence
- recognition
- data sequence
- 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.)
- Pending
Links
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
Abstract
The embodiment of the present application discloses the method and apparatus for handling data.One specific embodiment of this method includes: to obtain data sequence that is to be processed, including at least one default mark, the default missing values identified for indicating data sequence;For the data in data sequence, execute following processing step: in response to determining that the data are default mark, it determines preassigned, it is corresponding with the data, data in data sequence are as the corresponding target data of the data, according to training in advance, for handling included by the Recognition with Recurrent Neural Network of data sequence, the output result of sub-neural network for processing target data, assignment is carried out to the data, and determine that the data include as Recognition with Recurrent Neural Network by assigned value, for handling the input of the sub-neural network of the data, to obtain the output result of the sub-neural network for handling the data.The embodiment realizes Recognition with Recurrent Neural Network being effectively treated to the data sequence with missing values.
Description
Technical field
The invention relates to field of computer technology, and in particular to the method and apparatus for handling data.
Background technique
Recognition with Recurrent Neural Network is a kind of artificial neural network of node orientation connection cyclization.The processing list of Recognition with Recurrent Neural Network
The feedback link of existing inside has feedforward to connect again between member.Therefore, the internal state of Recognition with Recurrent Neural Network can show dynamic
Timing behavior.Recognition with Recurrent Neural Network can receive time series data as input, and analyze time series data.
Summary of the invention
The embodiment of the present application is proposed for handling data method and device.
In a first aspect, the embodiment of the present application provides a kind of method for handling data, this method comprises: obtaining wait locate
Reason, the default mark that includes at least one data sequence, wherein the default missing values identified for indicating data sequence;
For the data in data sequence, following processing step is executed: in response to determining that the data are default mark, determining and preassign
, data corresponding with the data, in data sequence as the corresponding target data of the data, according to training in advance, use
Included by the Recognition with Recurrent Neural Network of processing data sequence, the output of sub-neural network for processing target data as a result,
To the data carry out assignment, and determine the data include as Recognition with Recurrent Neural Network by assigned value, for handling the data
Sub-neural network input, to obtain the output result of the sub-neural network for handling the data.
In some embodiments, processing step further include: the data are not default mark in response to determination, determine the data
As the input of the sub-neural network included by Recognition with Recurrent Neural Network, for handling the data, to obtain for handling the number
According to sub-neural network output result.
In some embodiments, the method for handling data further include: determine that the data in data sequence are corresponding defeated
Result is as the son nerve included by Recognition with Recurrent Neural Network, for handling the corresponding output result of the data in data sequence out
The input of network, and will be used to handle the output result of the sub-neural network of the corresponding output result of the data in data sequence
It is determined as the corresponding processing result of data sequence.
In some embodiments, target data be according to Recognition with Recurrent Neural Network to the data in data sequence by arriving first after
Processing order, the processed data of preceding preset number of the data.
In some embodiments, training obtains Recognition with Recurrent Neural Network as follows: obtaining training sample set, wherein
Training sample includes the data sequence and the corresponding processing result of data sequence for including at least one default mark;Utilize machine
The method of study, the data sequence in training sample that training sample is concentrated, will as the input of initial cycle neural network
Desired output of the processing result corresponding with the data sequence of input as initial cycle neural network, training obtain circulation nerve
Network.
In some embodiments, the data in data sequence are used to indicate the attribute value of target object;And circulation nerve
The attribute value or Recognition with Recurrent Neural Network that network is used to predict target object are for determining whether target object belongs to pre-set categories.
Second aspect, the embodiment of the present application provide it is a kind of for handling the device of data, the device include: obtain it is single
Member is configured to obtain data sequence that is to be processed, including at least one default mark, wherein default mark is used for table
Show the missing values of data sequence;Processing unit is configured to execute following processing step for the data in data sequence: ring
It should determine that data preassigned, corresponding with the data, in data sequence are used as in determining that the data are default mark and be somebody's turn to do
The corresponding target data of data according to included by Recognition with Recurrent Neural Network training in advance, for handling data sequence, is used for
The output of the sub-neural network of processing target data is as a result, carry out assignment to the data, and determine that the data are made by assigned value
The input of sub-neural network that include for Recognition with Recurrent Neural Network, for handling the data, to obtain for handling the data
The output result of sub-neural network.
In some embodiments, processing unit is further configured in response to determining the data not be default mark, really
Fixed input of the data as the sub-neural network included by Recognition with Recurrent Neural Network, for handling the data, to be used for
Handle the output result of the sub-neural network of the data.
In some embodiments, for handling the device of data further include: determination unit is configured to determine data sequence
In the corresponding output result of data as included by Recognition with Recurrent Neural Network, it is corresponding for handling the data in data sequence
Export the input of the sub-neural network of result, and the son mind that will be used to handle the corresponding output result of the data in data sequence
Output result through network is determined as the corresponding processing result of data sequence.
In some embodiments, target data be according to Recognition with Recurrent Neural Network to the data in data sequence by arriving first after
Processing order, the processed data of preceding preset number of the data.
In some embodiments, training obtains Recognition with Recurrent Neural Network as follows: obtaining training sample set, wherein
Training sample includes the data sequence and the corresponding processing result of data sequence for including at least one default mark;Utilize machine
The method of study, the data sequence in training sample that training sample is concentrated, will as the input of initial cycle neural network
Desired output of the processing result corresponding with the data sequence of input as initial cycle neural network, training obtain circulation nerve
Network.
In some embodiments, the data in data sequence are used to indicate the attribute value of target object;And circulation nerve
The attribute value or Recognition with Recurrent Neural Network that network is used to predict target object are for determining whether target object belongs to pre-set categories.
The third aspect, the embodiment of the present application provide a kind of electronic equipment, which includes: one or more processing
Device;Storage device, for storing one or more programs;When one or more programs are executed by one or more processors, make
Obtain method of the one or more processors realization as described in implementation any in first aspect.
Fourth aspect, the embodiment of the present application provide a kind of computer-readable medium, are stored thereon with computer program, should
The method as described in implementation any in first aspect is realized when computer program is executed by processor.
Method and apparatus provided by the embodiments of the present application for handling data, by obtain it is to be processed, include to
The data sequence of a few default mark, wherein the default missing values identified for indicating data sequence;For in data sequence
Data, execute following processing step: in response to determining that the data are default mark, determine it is preassigned, with the data pair
Data answering, in data sequence as the corresponding target data of the data, according to training in advance, for handling data sequence
Recognition with Recurrent Neural Network included by, the output of sub-neural network for processing target data is as a result, assign the data
Value, and determine sub-neural network that the data include as Recognition with Recurrent Neural Network by assigned value, for handling the data
Input, with obtain the sub-neural network for handling the data output as a result, so as in Recognition with Recurrent Neural Network to data
In the treatment process of sequence, assignment is carried out to the missing values in data sequence to handle and realize Recognition with Recurrent Neural Network to scarce
The data sequence of mistake value is effectively treated.
Detailed description of the invention
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, the application's is other
Feature, objects and advantages will become more apparent upon:
Fig. 1 is that one embodiment of the application can be applied to exemplary system architecture figure therein;
Fig. 2 is the flow chart according to one embodiment of the method for handling data of the application;
Fig. 3 is the adaptable Recognition with Recurrent Neural Network of one embodiment according to the method for handling data of the application
Network structure;
Fig. 4 is the flow chart according to another embodiment of the method for handling data of the application;
Fig. 5 is the schematic diagram according to an application scenarios of the method for handling data of the embodiment of the present application;
Fig. 6 is the structural schematic diagram according to one embodiment of the device for handling data of the application;
Fig. 7 is adapted for the structural schematic diagram for the computer system for realizing the electronic equipment of the embodiment of the present application.
Specific embodiment
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to
Convenient for description, part relevant to related invention is illustrated only in attached drawing.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
Fig. 1 is shown can be using the method for handling data of the application or the implementation of the device for handling data
The exemplary architecture 100 of example.
As shown in Figure 1, system architecture 100 may include terminal device 101,102,103, network 104 and server 105.
Network 104 between terminal device 101,102,103 and server 105 to provide the medium of communication link.Network 104 can be with
Including various connection types, such as wired, wireless communication link or fiber optic cables etc..
Terminal device 101,102,103 is interacted by network 104 with server 105, to receive or send message etc..Terminal
Various client applications can be installed in equipment 101,102,103.Such as web browser applications, searching class are applied, immediately
Means of communication etc..
Terminal device 101,102,103 can be hardware, be also possible to software.When terminal device 101,102,103 is hard
When part, it can be the various electronic equipments for supporting the storage and processing of data sequence, including but not limited to smart phone, plate electricity
Brain, E-book reader, pocket computer on knee and desktop computer etc..When terminal device 101,102,103 is software
When, it may be mounted in above-mentioned cited electronic equipment.Multiple softwares or software module may be implemented into (such as mentioning in it
For the multiple softwares or software module of Distributed Services), single software or software module also may be implemented into.It does not do herein specific
It limits.
Server 105 can be to provide the server of various services, for example, tool of the transmission of terminal device 101,102,103
The processing server for thering is the data sequence of missing values to be handled.Processing server can be to the received data with missing values
Missing values in sequence carry out the processing such as assignment, and generate processing result.
It should be noted that the above-mentioned data sequence with missing values can also be stored directly in the local of server 105,
Server 105 can directly extract the local data sequence with missing values stored and be handled, at this point it is possible to not deposit
In terminal device 101,102,103 and network 104.
It should be noted that the method provided by the embodiment of the present application for handling data is generally held by server 105
Row, correspondingly, the device for handling data is generally positioned in server 105.
It may also be noted that terminal device 101,102,103 can also be to the data sequence with missing values at
Reason, at this point, the method for handling data can also be executed by terminal device 101,102,103, correspondingly, for handling data
Device also can be set in terminal device 101,102,103.At this point, exemplary system architecture 100 can there is no services
Device 105 and network 104.
It should be noted that server can be hardware, it is also possible to software.When server is hardware, may be implemented
At the distributed server cluster that multiple servers form, individual server also may be implemented into.It, can when server is software
It, can also be with to be implemented as multiple softwares or software module (such as providing multiple softwares of Distributed Services or software module)
It is implemented as single software or software module.It is not specifically limited herein.
It should be understood that the number of terminal device, network and server in Fig. 1 is only schematical.According to realization need
It wants, can have any number of terminal device, network and server.
With continued reference to Fig. 2, it illustrates the processes according to one embodiment of the method for handling data of the application
200.This be used for handle data method the following steps are included:
Step 201, data sequence that is to be processed, including at least one default mark is obtained.
In the present embodiment, the above-mentioned executing subject (server 105 as shown in Figure 1) for handling data can be first
By way of wired connection or wireless connection from local or other data storage devices obtain it is to be processed, include at least one
The data sequence of a default mark.Wherein, default to identify the missing values that can be used to indicate that data sequence.
Data sequence can refer to the set of data with certain sequence.For example, data sequence can be the various times
Sequence data.At this point, data sequence can be the data being collected into sequentially in time.In another example data sequence can be from
The data obtained in specified data library or data set, and in data sequence according to specified sequence (as in the database
Storage order etc.) arrangement.
Missing values can refer to according to the corresponding sequence of data sequence, the data for not acquiring or having not been obtained.That is data sequence
In data be incomplete.For example, for time series data, when missing values can refer in data acquisition certain section
Between fail the data being collected into.
Default mark can be by the pre-set any form of mark of technical staff.For example, special symbol can be used
Number, the mark as missing values such as specified letter and the combination of number etc..
Step 202, for the data in data sequence, following processing step is executed: in response to determining that the data are default
Mark determines data corresponding with the data, in data sequence as the corresponding target data of the data, according to preparatory training
, included by Recognition with Recurrent Neural Network for handling data sequence, the output of sub-neural network for processing target data
As a result, to the data carry out assignment, and determine the data include as Recognition with Recurrent Neural Network by assigned value, for handling this
The input of the sub-neural network of data, to obtain the output result of the sub-neural network for handling the data.
In the present embodiment, it can use Recognition with Recurrent Neural Network train in advance, for handling data sequence to data
Data in sequence are handled.Specifically, for the data in data sequence, however, it is determined that the data are default mark, can be with
The data are handled as follows, with the processing of the missing values of complete paired data sequence:
Step 1 determines data corresponding with the data, in data sequence as the corresponding target data of the data.
Wherein, mark default at least one of data sequence, can preassign these by technical staff and preset
Identify corresponding target data.For example, distance in data sequence this can be preset mark most for any default mark
The data of close, non-default mark are as the corresponding target data of the default mark.
Wherein it is possible to be determined between the data in data sequence according to the corresponding sequence of data sequence (such as time sequencing)
The distance of distance.For example, with the data of the data direct neighbor and compared with the data interphase is separated with the data of other data,
Distance with the data of the data direct neighbor apart from the data is closer.For example, a pair of of data that the corresponding time is closer, then
The distance of this pair of data is also closer.
It is multiple that each the corresponding target data of default mark, which can specify,.For example, can be incited somebody to action for any default mark
The distance default mark is nearest in data sequence, data of two or more non-default marks are as the default mark
Corresponding target data.
Step 2, according to included by Recognition with Recurrent Neural Network training in advance, for handling data sequence, for handling
The output of the sub-neural network of target data is as a result, carry out assignment to the data.
Wherein, for the data in data sequence, Recognition with Recurrent Neural Network may include the son mind handled the data
Through network.Generally, the sub-neural network handled to the data is using the data as input.
As an example, Fig. 3 is followed according to one embodiment of the method for handling data of the application is adaptable
The network structure of ring neural network.As shown in figure 3, with data sequence (T1, T2 ..., TN) 301 as an example, data sequence
Data in 301 are acquired to obtain sequentially in time.For any data in data sequence 301, circulation nerve
Network may include the sub-neural network handled the data.As shown in the figure, Recognition with Recurrent Neural Network includes to data T1
The sub-neural network 3021 of processing, to data T2 processing sub-neural network 3022, and so on, also include to data TN processing
Sub-neural network 302N.
The input that each data in data sequence 301 can be used as corresponding sub-neural network (is expressed as X in figure
(T1),X(T2),……,XTN).With data T1 as an example, the sub-neural network 3021 for handling data T1 is with data T1
As input, it is expressed as X (T1), obtains sub-neural network 3021 to the output of data T1 as a result, being expressed as Y (T1).
It is alternatively possible to using the output result of sub-neural network for processing target data as the value of the data.
It is alternatively possible to based on the output result of sub-neural network for processing target data, to output result
Default processing operation is carried out, and using the processing result to output result as the value of the data.Wherein, processing operation is preset
It can be by the preassigned processing operation of technical staff.For example, default processing operation can be will output result add deduct or
Multiply or except specified adjustment factor.
The corresponding target data of the data there are two or it is more than two when, can will corresponding each target data difference
Value of the average value of corresponding output result as the data.It is of course also possible to based on obtained average value, to average
Value carries out default processing operation, and using the processing result to average value as the value of the data.
Step 3 determines son mind that the data include as Recognition with Recurrent Neural Network by assigned value, for handling the data
Input through network, to obtain the output result of the sub-neural network for handling the data.
In some optional implementations of the present embodiment, for the data in data sequence, however, it is determined that the data are not
For default mark, the data can be determined as the sub-neural network included by Recognition with Recurrent Neural Network, for handling the data
Input, to obtain the output result of the sub-neural network for handling the data.
In some optional implementations of the present embodiment, the corresponding output knot of data in data sequence can be determined
Fruit is as the sub-neural network included by Recognition with Recurrent Neural Network, for handling the corresponding output result of the data in data sequence
Input, and will be used to handle the data in data sequence it is corresponding output result sub-neural network output result determine
For the corresponding processing result of data sequence.
Wherein, the corresponding output result of data in data sequence can refer to it is that Recognition with Recurrent Neural Network includes, be respectively used to
Handle the output result of the sub-neural network of each data in data sequence.Recognition with Recurrent Neural Network can be in data sequence
The corresponding output result of data is further processed, to obtain Recognition with Recurrent Neural Network to the final process result of data sequence.
As an example, as shown in the figure, Recognition with Recurrent Neural Network can also be in 301 in data sequence with continued reference to Fig. 3
The corresponding output result of each data (be expressed as in figure Y (T1), Y (T2) ..., Y (TN)) be further processed.Specifically
Ground, Recognition with Recurrent Neural Network can also include for handling the corresponding output knot of each data in 301 in data sequence
The sub-neural network 303 of fruit, and the output result 304 of sub-neural network 303 can be determined as Recognition with Recurrent Neural Network logarithm
According to the processing result of sequence 301.
In practice, according to different application demand or application scenarios, what Recognition with Recurrent Neural Network included is used to handle data sequence
The sub-neural network 303 of the corresponding output result of each data in 301 in column, specific network structure can not
Together.
In practice, the effect of the Recognition with Recurrent Neural Network under different application scene is various, to data sequence reality
Existing different processing, to solve the problems, such as corresponding under different application scene.
In some optional implementations of the present embodiment, the data in data sequence can be used to indicate that target object
Attribute value.At this point, Recognition with Recurrent Neural Network can be used for predicting that the attribute value of target object or Recognition with Recurrent Neural Network can be used
In determining whether target object belongs to pre-set categories.
Wherein, object can refer to arbitrary object.Target object can be the object specified under concrete application scene.As
Example, target object can be weather.Attribute value can be used to indicate that weather particular situation.Such as temperature.In data sequence
Data can be when temperature daily in the previous moon.At this point, Recognition with Recurrent Neural Network can be designed for prediction next month
Interior daily temperature.Alternatively, Recognition with Recurrent Neural Network can be designed for determining that whether belonging to sleet sky within the previous moon compares
More one month.
The specific training method and training process of Recognition with Recurrent Neural Network can be according to the network structures of Recognition with Recurrent Neural Network, benefit
Recognition with Recurrent Neural Network is obtained with the training method training of existing various Recognition with Recurrent Neural Network.
In some optional implementations of the present embodiment, it can train as follows and obtain circulation nerve net
Network: training sample set is obtained.Wherein, each training sample may include include at least one default mark data sequence and
The corresponding processing result of data sequence.Using the method for machine learning, by the data sequence in the training sample of training sample concentration
The input as initial cycle neural network is arranged, using processing result corresponding with the data sequence of input as initial cycle nerve
The desired output of network, training obtain Recognition with Recurrent Neural Network.
Wherein, initial cycle neural network can design specific network structure, Yi Jili according to specific application scenarios
It builds to obtain with existing various deep learning frames.
It should be noted that the above-mentioned training about Recognition with Recurrent Neural Network is the known skill studied and applied extensively at present
Art, details are not described herein.
It,, can be with after being handled using Recognition with Recurrent Neural Network data sequence according to specific application scenarios in practice
Processing result is saved, can also show processing result to user.It is of course also possible to save some output results in treatment process
(the output result of such as each sub-neural network).
The application it is provided by the above embodiment for handle the methods of data by Recognition with Recurrent Neural Network to data sequence
In the treatment process of column, assignment is carried out to missing to missing values using the corresponding target data of missing values corresponding output result
Value is handled, to realize Recognition with Recurrent Neural Network being effectively treated to the data sequence with missing values, and is right
In the treatment process of data sequence, assignment is carried out to missing values in processing, to further help in promotion to missing
The processing speed of the data sequence of value.
With further reference to Fig. 4, it illustrates the processes 400 of another embodiment of the method for handling data.The use
In the process 400 of the method for processing data, comprising the following steps:
Step 401, data sequence that is to be processed, including at least one default mark is obtained.
The specific implementation procedure of this step 401 can refer to the related description of the step 201 in Fig. 2 corresponding embodiment,
This is repeated no more.
Step 402, for the data in data sequence, execute following 4021 and 4022 shown in processing step:
Step 4021, in response to determining that the data are default mark, number corresponding with the data, in data sequence is determined
According to as the corresponding target data of the data, according to training in advance, wrapped for handling the Recognition with Recurrent Neural Network of data sequence
The output of sub-neural network including, for processing target data is as a result, carry out assignment to the data, and determine the data quilt
The input of sub-neural network that assigned value includes as Recognition with Recurrent Neural Network, for handling the data, to obtain for handling
The output result of the sub-neural network of the data, wherein target data is according to Recognition with Recurrent Neural Network to the number in data sequence
According to by the processing order after arriving first, the processed data of preceding preset number of the data.
In the present embodiment, Recognition with Recurrent Neural Network is usually to have processing sequence when handling data sequence.For example, right
It when time series data is handled, is successively handled according to the corresponding time order and function of data.Still referring to FIG. 3, as in Fig. 3
Shown, the processing sequence to data sequence 301 is first to handle T1, then handle T2, and so on, finally handle TN.
Optionally, preset number can be one.At this point, target data is the previous processed data of the data.
Step 4022, it is default mark in response to determining the data not, determines that the data are wrapped as Recognition with Recurrent Neural Network
The input of sub-neural network include, for handling the data, to obtain the output of the sub-neural network for handling the data
As a result.
The specific implementation procedure of this step 402 can refer to the related description of the step 202 in Fig. 2 corresponding embodiment,
This is repeated no more.
Step 403, determine the corresponding output result of data in data sequence as included by Recognition with Recurrent Neural Network, use
The input of the sub-neural network of the corresponding output result of data in processing data sequence, and will be used to handle data sequence
In the output result of sub-neural network of the corresponding output result of data be determined as the corresponding processing result of data sequence.
The specific implementation procedure of this step 403 can refer to the related description of the step 202 in Fig. 2 corresponding embodiment, herein
It is not repeating.
With continued reference to the signal that Fig. 5, Fig. 5 are according to the application scenarios of the method for handling data of the present embodiment
Figure 50 0.It include two missing values in data sequence 501, and indicated using mark " M " in the application scenarios of Fig. 5.Data sequence
Further include in column 501 data T1, T4 ..., TN.
Recognition with Recurrent Neural Network successively handles the data in data sequence 501.Firstly, using data T1 as being used to locate
The input for managing the sub-neural network 5021 of data T1, is expressed as X (T1), to obtain the output result Y of sub-neural network 5021
(T1)。
It later, can be corresponding by the previous processed data T1 of first missing values for first missing values
Output result Y (T1) is assigned to first missing values.It is used at this point it is possible to which first missing values is used as by assigned value Y (T1)
The input for handling the sub-neural network 5022 of first missing values obtains the corresponding output result Y (T2) of first missing values.
Later, can be by the previous processed data of second missing values for second missing values, i.e., first
The corresponding output result Y (T2) of missing values is assigned to the second missing values.At this point it is possible to by second missing values by assigned value Y
(T2) input as the sub-neural network 5023 for handling second missing values, obtains the corresponding output of second missing values
As a result Y (T3).
Later, to fourth data T4, using data T4 as the input for the sub-neural network 5024 for being used to handle data T4
It indicates are as follows: X (T4) obtains the output result Y (T4) of sub-neural network 5024.
Similarly, each data between data T5 to the TN in data sequence 501 can be handled, is obtained later
The corresponding output result of each data.
Later, it is right respectively that each data for handling in data sequence 501 that Recognition with Recurrent Neural Network includes be can use
The sub-neural network 503 for the output result answered output result corresponding to each data (as illustrated in the drawing from Y (T1) to
Y (TN)) it is further processed, to obtain processing knot of the output result 504 as Recognition with Recurrent Neural Network to data sequence 501
Fruit.
Figure 4, it is seen that the method for handling data compared with the corresponding embodiment of Fig. 2, in the present embodiment
Process 400 highlight for any missing values, can according in data sequence before the missing values processed data pair
The step of output result answered carries out assignment to the missing values, thus realizing Recognition with Recurrent Neural Network to the data with missing values
While being effectively treated of sequence helps to ensure that the accuracy to the processing result of data sequence.
With further reference to Fig. 6, as the realization to method shown in above-mentioned each figure, this application provides for handling data
One embodiment of device, the Installation practice is corresponding with embodiment of the method shown in Fig. 2, which specifically can be applied to
In various electronic equipments.
As shown in fig. 6, the device 600 provided in this embodiment for handling data includes that acquiring unit 601 and processing are single
Member 602.Wherein, acquiring unit 601 is configured to obtain data sequence that is to be processed, including at least one default mark,
Wherein, the missing values identified for indicating data sequence are preset.Processing unit 602 is configured to for the number in data sequence
According to executing following processing step: in response to determining that the data are default mark, determine it is preassigned, corresponding with the data,
Data in data sequence as the corresponding target data of the data, according to training in advance, for handling following for data sequence
Included by ring neural network, the output of sub-neural network for processing target data as a result, to the data carry out assignment, with
And determine the input of sub-neural network that the data include as Recognition with Recurrent Neural Network by assigned value, for handling the data,
To obtain the output result of the sub-neural network for handling the data.
In the present embodiment, in the device 600 for handling data: the specific place of acquiring unit 601 and processing unit 602
Reason and its brought technical effect can refer to the related description of step 201 and step 202 in Fig. 2 corresponding embodiment respectively,
Details are not described herein.
In some optional implementations of the present embodiment, processing unit 602 is further configured in response to determination
The data are not default mark, determine the data as the son nerve included by Recognition with Recurrent Neural Network, for handling the data
The input of network, to obtain the output result of the sub-neural network for handling the data.
In some optional implementations of the present embodiment, for handling the device 600 of data further include: determination unit
(not shown) is configured to determine the corresponding output result of the data in data sequence as included by Recognition with Recurrent Neural Network
, the input of sub-neural network for handling the corresponding output result of the data in data sequence, and will be used to handle number
It is determined as the corresponding processing knot of data sequence according to the output result of the sub-neural network of the corresponding output result of the data in sequence
Fruit.
In some optional implementations of the present embodiment, target data is according to Recognition with Recurrent Neural Network to data sequence
In data by the processing order after arriving first, the processed data of preceding preset number of the data.
In some optional implementations of the present embodiment, training obtains Recognition with Recurrent Neural Network as follows: obtaining
Take training sample set, wherein training sample includes that the data sequence for including at least one default mark and data sequence correspond to
Processing result;Using the method for machine learning, using the data sequence in the training sample that training sample is concentrated as initially following
The input of ring neural network, processing result corresponding with the data sequence of input is defeated as the expectation of initial cycle neural network
Out, training obtains Recognition with Recurrent Neural Network.
In some optional implementations of the present embodiment, the data in data sequence are used to indicate the category of target object
Property value;And Recognition with Recurrent Neural Network is used to predict that the attribute value of target object or Recognition with Recurrent Neural Network to be used to determine target object
Whether pre-set categories are belonged to.
The device provided by the above embodiment of the application, first acquiring unit obtain it is to be processed, include at least one
The data sequence of default mark, wherein the default missing values identified for indicating data sequence.Then processing unit is for data
Data in sequence execute following processing step: being default mark in response to the determining data, determine preassigned and be somebody's turn to do
Data are corresponding, the data in data sequence are as the corresponding target data of the data, according to training in advance, for handling number
Included by Recognition with Recurrent Neural Network according to sequence, the output of sub-neural network for processing target data is as a result, to the data
Assignment is carried out, and determines son nerve that the data include as Recognition with Recurrent Neural Network by assigned value, for handling the data
The input of network, to obtain the output of the sub-neural network for handling the data as a result, so as in Recognition with Recurrent Neural Network
To in the treatment process of data sequence, carrying out assignment to the missing values in data sequence realizes Recognition with Recurrent Neural Network pair to handle
Data sequence with missing values is effectively treated.
Below with reference to Fig. 7, it illustrates the computer systems 700 for the electronic equipment for being suitable for being used to realize the embodiment of the present application
Structural schematic diagram.Electronic equipment shown in Fig. 7 is only an example, function to the embodiment of the present application and should not use model
Shroud carrys out any restrictions.
As shown in fig. 7, computer system 700 includes central processing unit (CPU) 701, it can be read-only according to being stored in
Program in memory (ROM) 702 or be loaded into the program in random access storage device (RAM) 703 from storage section 708 and
Execute various movements appropriate and processing.In RAM 703, also it is stored with system 700 and operates required various programs and data.
CPU 701, ROM 702 and RAM 703 are connected with each other by bus 704.Input/output (I/O) interface 705 is also connected to always
Line 704.
I/O interface 705 is connected to lower component: the importation 706 including keyboard, mouse etc.;It is penetrated including such as cathode
The output par, c 707 of spool (CRT), liquid crystal display (LCD) etc. and loudspeaker etc.;Storage section 708 including hard disk etc.;
And the communications portion 709 of the network interface card including LAN card, modem etc..Communications portion 709 via such as because
The network of spy's net executes communication process.Driver 710 is also connected to I/O interface 705 as needed.Detachable media 711, such as
Disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on as needed on driver 710, in order to read from thereon
Computer program be mounted into storage section 708 as needed.
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be carried on computer-readable medium
On computer program, which includes the program code for method shown in execution flow chart.In such reality
It applies in example, which can be downloaded and installed from network by communications portion 709, and/or from detachable media
711 are mounted.When the computer program is executed by central processing unit (CPU) 701, limited in execution the present processes
Above-mentioned function.
It should be noted that the computer-readable medium of the application can be computer-readable signal media or computer
Readable storage medium storing program for executing either the two any combination.Computer readable storage medium for example can be --- but it is unlimited
In system, device or the device of --- electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, or any above combination.It calculates
The more specific example of machine readable storage medium storing program for executing can include but is not limited to: have the electrical connection, portable of one or more conducting wires
Formula computer disk, hard disk, random access storage device (RAM), read-only memory (ROM), erasable programmable read only memory
(EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device or
The above-mentioned any appropriate combination of person.In this application, computer readable storage medium can be it is any include or storage program
Tangible medium, which can be commanded execution system, device or device use or in connection.And in this Shen
Please in, computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal,
In carry computer-readable program code.The data-signal of this propagation can take various forms, including but not limited to
Electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be computer-readable
Any computer-readable medium other than storage medium, the computer-readable medium can send, propagate or transmit for by
Instruction execution system, device or device use or program in connection.The journey for including on computer-readable medium
Sequence code can transmit with any suitable medium, including but not limited to: wireless, electric wire, optical cable, RF etc. are above-mentioned
Any appropriate combination.
Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the application, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, program segment or code of table, a part of the module, program segment or code include one or more use
The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box
The function of note can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are actually
It can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it to infuse
Meaning, the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart can be with holding
The dedicated hardware based system of functions or operations as defined in row is realized, or can use specialized hardware and computer instruction
Combination realize.
Being described in unit involved in the embodiment of the present application can be realized by way of software, can also be by hard
The mode of part is realized.Described unit also can be set in the processor, for example, can be described as: a kind of processor, packet
Include acquiring unit and processing unit.Wherein, the title of these units does not constitute the limit to the unit itself under certain conditions
It is fixed, for example, acquiring unit is also described as " obtaining data sequence that is to be processed, including at least one default mark
Unit ".
As on the other hand, present invention also provides a kind of computer-readable medium, which be can be
Included in electronic equipment described in above-described embodiment;It is also possible to individualism, and without in the supplying electronic equipment.
Above-mentioned computer-readable medium carries one or more program, when said one or multiple programs are held by the electronic equipment
When row, so that the electronic equipment: obtaining data sequence that is to be processed, including at least one default mark, wherein pre- bidding
Know the missing values for indicating data sequence;For the data in data sequence, following processing step is executed: should in response to determining
Data are default mark, determine that preassigned, corresponding with the data, in data sequence data are corresponding as the data
Target data according to included by Recognition with Recurrent Neural Network training in advance, for handling data sequence, is used for processing target number
According to sub-neural network output as a result, to the data carry out assignment, and determine the data by assigned value as circulation nerve
The input of sub-neural network that network includes, for handling the data, to obtain the sub-neural network for handling the data
Output result.
Above description is only the preferred embodiment of the application and the explanation to institute's application technology principle.Those skilled in the art
Member is it should be appreciated that invention scope involved in the application, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from foregoing invention design, it is carried out by above-mentioned technical characteristic or its equivalent feature
Any combination and the other technical solutions formed.Such as features described above has similar function with (but being not limited to) disclosed herein
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (14)
1. a kind of method for handling data, comprising:
Obtain data sequence that is to be processed, including at least one default mark, wherein default mark is for indicating the number
According to the missing values of sequence;
For the data in the data sequence, following processing step is executed: in response to determining that the data are default mark, determining
Data in data sequence corresponding with the data, described are as the corresponding target data of the data, according to training in advance, use
Sub-neural network included by the Recognition with Recurrent Neural Network for handling the data sequence, for handling the target data it is defeated
Out as a result, to the data carry out assignment, and determine the data include as the Recognition with Recurrent Neural Network by assigned value, be used for
The input of the sub-neural network of the data is handled, to obtain the output result of the sub-neural network for handling the data.
2. according to the method described in claim 1, wherein, the processing step further include:
Be not default mark in response to determining the data, determine the data as included by the Recognition with Recurrent Neural Network, be used for
The input of the sub-neural network of the data is handled, to obtain the output result of the sub-neural network for handling the data.
3. according to the method described in claim 1, wherein, the method also includes:
Determine the corresponding output result of data in the data sequence as included by the Recognition with Recurrent Neural Network, for locating
The input of the sub-neural network of the corresponding output result of data in the data sequence is managed, and will be used to handle the data
The output result of the sub-neural network of the corresponding output result of data in sequence is determined as the corresponding processing of the data sequence
As a result.
4. according to the method described in claim 1, wherein, the target data is according to the Recognition with Recurrent Neural Network to the number
According to the data in sequence by the processing order after arriving first, the processed data of preceding preset number of the data.
5. according to the method described in claim 1, wherein, training obtains the Recognition with Recurrent Neural Network as follows:
Obtain training sample set, wherein training sample includes the data sequence and data sequence for including at least one default mark
Arrange corresponding processing result;
Using the method for machine learning, the data sequence in training sample that the training sample is concentrated is as initial cycle mind
Input through network, using processing result corresponding with the data sequence of input as the desired output of initial cycle neural network,
Training obtains the Recognition with Recurrent Neural Network.
6. method described in one of -5 according to claim 1, wherein the data in the data sequence are for indicating target object
Attribute value;And
The attribute value or the Recognition with Recurrent Neural Network that the Recognition with Recurrent Neural Network is used to predict the target object are for determining institute
State whether target object belongs to pre-set categories.
7. a kind of for handling the device of data, comprising:
Acquiring unit is configured to obtain data sequence that is to be processed, including at least one default mark, wherein default
Identify the missing values for indicating the data sequence;
Processing unit is configured to execute following processing step for the data in the data sequence: in response to determining the number
According to the data for default mark, determined in data sequence corresponding with the data, described as the corresponding target data of the data,
According to included by Recognition with Recurrent Neural Network training in advance, for handling the data sequence, for handling the number of targets
According to sub-neural network output as a result, to the data carry out assignment, and determine the data by assigned value as the circulation
The input of sub-neural network that neural network includes, for handling the data, to obtain the son nerve for handling the data
The output result of network.
8. device according to claim 7, wherein the processing unit is further configured to:
Be not default mark in response to determining the data, determine the data as included by the Recognition with Recurrent Neural Network, be used for
The input of the sub-neural network of the data is handled, to obtain the output result of the sub-neural network for handling the data.
9. device according to claim 7, wherein described device further include:
Determination unit, the corresponding output result of the data being configured to determine in the data sequence is as the circulation nerve net
The input of sub-neural network included by network, for handling the corresponding output result of the data in the data sequence, and
The output result for being used to handle the sub-neural network of the corresponding output result of data in the data sequence is determined as described
The corresponding processing result of data sequence.
10. device according to claim 7, wherein the target data is according to the Recognition with Recurrent Neural Network to described
Data in data sequence are by the processing order after arriving first, the processed data of preceding preset number of the data.
11. device according to claim 7, wherein training obtains the Recognition with Recurrent Neural Network as follows:
Obtain training sample set, wherein training sample includes the data sequence and data sequence for including at least one default mark
Arrange corresponding processing result;
Using the method for machine learning, the data sequence in training sample that the training sample is concentrated is as initial cycle mind
Input through network, using processing result corresponding with the data sequence of input as the desired output of initial cycle neural network,
Training obtains the Recognition with Recurrent Neural Network.
12. the device according to one of claim 7-11, wherein the data in the data sequence are for indicating target pair
The attribute value of elephant;And
The attribute value or the Recognition with Recurrent Neural Network that the Recognition with Recurrent Neural Network is used to predict the target object are for determining institute
State whether target object belongs to pre-set categories.
13. a kind of electronic equipment, comprising:
One or more processors;
Storage device is stored thereon with one or more programs;
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
Now such as method as claimed in any one of claims 1 to 6.
14. a kind of computer-readable medium, is stored thereon with computer program, wherein the realization when program is executed by processor
Such as method as claimed in any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811384897.7A CN109635923A (en) | 2018-11-20 | 2018-11-20 | Method and apparatus for handling data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811384897.7A CN109635923A (en) | 2018-11-20 | 2018-11-20 | Method and apparatus for handling data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109635923A true CN109635923A (en) | 2019-04-16 |
Family
ID=66068552
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811384897.7A Pending CN109635923A (en) | 2018-11-20 | 2018-11-20 | Method and apparatus for handling data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109635923A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111046027A (en) * | 2019-11-25 | 2020-04-21 | 北京百度网讯科技有限公司 | Missing value filling method and device for time series data |
CN111709583A (en) * | 2020-06-18 | 2020-09-25 | 北京字节跳动网络技术有限公司 | User retention time generation method and device, electronic equipment and medium |
CN111767987A (en) * | 2020-06-28 | 2020-10-13 | 北京百度网讯科技有限公司 | Data processing method, device and equipment based on recurrent neural network |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180063168A1 (en) * | 2016-08-31 | 2018-03-01 | Cisco Technology, Inc. | Automatic detection of network threats based on modeling sequential behavior in network traffic |
CN108197736A (en) * | 2017-12-29 | 2018-06-22 | 北京工业大学 | A kind of Air Quality Forecast method based on variation self-encoding encoder and extreme learning machine |
CN108460481A (en) * | 2018-01-30 | 2018-08-28 | 中国航天电子技术研究院 | Unmanned plane spot development law prediction technique based on Recognition with Recurrent Neural Network |
CN108615096A (en) * | 2018-05-10 | 2018-10-02 | 平安科技(深圳)有限公司 | Server, the processing method of Financial Time Series and storage medium |
-
2018
- 2018-11-20 CN CN201811384897.7A patent/CN109635923A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180063168A1 (en) * | 2016-08-31 | 2018-03-01 | Cisco Technology, Inc. | Automatic detection of network threats based on modeling sequential behavior in network traffic |
CN108197736A (en) * | 2017-12-29 | 2018-06-22 | 北京工业大学 | A kind of Air Quality Forecast method based on variation self-encoding encoder and extreme learning machine |
CN108460481A (en) * | 2018-01-30 | 2018-08-28 | 中国航天电子技术研究院 | Unmanned plane spot development law prediction technique based on Recognition with Recurrent Neural Network |
CN108615096A (en) * | 2018-05-10 | 2018-10-02 | 平安科技(深圳)有限公司 | Server, the processing method of Financial Time Series and storage medium |
Non-Patent Citations (4)
Title |
---|
MINH NGUYEN ET AL: "Modeling Alzheimer’s disease progression using deep recurrent neural networks", 《2018 INTERNATIONAL WORKSHOP ON PATTERN RECOGNITION IN NEUROIMAGING (PRNI)》 * |
WEI CAO ET AL: "BRITS: Bidirectional Recurrent Imputation for Time Series", 《ARXIV:1805.10572V1[CS.LG]》 * |
胡航: "《语音信号处理》", 31 July 2009 * |
赵晶夫: "《奋斗的历程 南京市政规划管理工作联动计划 2006》", 31 December 2006, 南京:东南大学出版社 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111046027A (en) * | 2019-11-25 | 2020-04-21 | 北京百度网讯科技有限公司 | Missing value filling method and device for time series data |
CN111709583A (en) * | 2020-06-18 | 2020-09-25 | 北京字节跳动网络技术有限公司 | User retention time generation method and device, electronic equipment and medium |
CN111709583B (en) * | 2020-06-18 | 2023-05-23 | 抖音视界有限公司 | User retention time generation method, device, electronic equipment and medium |
CN111767987A (en) * | 2020-06-28 | 2020-10-13 | 北京百度网讯科技有限公司 | Data processing method, device and equipment based on recurrent neural network |
CN111767987B (en) * | 2020-06-28 | 2024-02-20 | 北京百度网讯科技有限公司 | Data processing method, device and equipment based on cyclic neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20190012575A1 (en) | Method, apparatus and system for updating deep learning model | |
CN108734293A (en) | Task management system, method and apparatus | |
CN110288049A (en) | Method and apparatus for generating image recognition model | |
CN107919129A (en) | Method and apparatus for controlling the page | |
CN108287927B (en) | For obtaining the method and device of information | |
CN109410253B (en) | For generating method, apparatus, electronic equipment and the computer-readable medium of information | |
CN109635923A (en) | Method and apparatus for handling data | |
CN109976997A (en) | Test method and device | |
CN108958992A (en) | test method and device | |
CN109634767A (en) | Method and apparatus for detection information | |
CN110321738A (en) | Information processing method and device | |
CN107590252A (en) | Method and device for information exchange | |
CN109862100A (en) | Method and apparatus for pushed information | |
CN109902446A (en) | Method and apparatus for generating information prediction model | |
CN108960110A (en) | Method and apparatus for generating information | |
CN108933695A (en) | Method and apparatus for handling information | |
CN107330091A (en) | Information processing method and device | |
CN109688086A (en) | Authority control method and device for terminal device | |
CN107968743A (en) | The method and apparatus of pushed information | |
CN109614549B (en) | Method and apparatus for pushed information | |
CN109446379A (en) | Method and apparatus for handling information | |
CN109597912A (en) | Method for handling picture | |
CN109492687A (en) | Method and apparatus for handling information | |
CN109614327A (en) | Method and apparatus for output information | |
CN109101956A (en) | Method and apparatus for handling image |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190416 |