CN109165244A - A kind of stream data processing method of the online multidimensional output based on correlation - Google Patents
A kind of stream data processing method of the online multidimensional output based on correlation Download PDFInfo
- Publication number
- CN109165244A CN109165244A CN201811035273.4A CN201811035273A CN109165244A CN 109165244 A CN109165244 A CN 109165244A CN 201811035273 A CN201811035273 A CN 201811035273A CN 109165244 A CN109165244 A CN 109165244A
- Authority
- CN
- China
- Prior art keywords
- multidimensional
- correlation
- support vector
- outcome variable
- regression model
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2216/00—Indexing scheme relating to additional aspects of information retrieval not explicitly covered by G06F16/00 and subgroups
- G06F2216/03—Data mining
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of stream data processing method of the online multidimensional output based on correlation, the relevance measure between the multidimensional outcome variable of multivariate regression estimation function is chosen first, relevance measure is learnt using stream data, obtain the correlation between the multidimensional outcome variable of multivariate regression estimation function, then multidimensional outcome variable correlation regularization model is established, and then obtain the multidimensional outcome variable correlation study system of multidimensional outcome variable, finally it is based on multidimensional outcome variable correlation study system, obtain multi output support vector regression model and its estimation function based on correlation, multi output support vector regression model is updated according to the new sample that multidimensional output training dataset is added simultaneously.
Description
Technical field
The present invention relates to being related to the fields such as big data correlation analysis and machine learning, it is especially a kind of based on correlation
The stream data processing method of line multidimensional output.
Background technique
Currently, all trades and professions have had accumulated a large amount of data resource, how various big datas are effectively analyzed,
And it is efficiently excavated from big data and is hidden in the useful information of its behind and then predicts that the development trend in things future is current
Information science field research hotspot.
Technology of the regression estimates as a kind of effective data mining and prediction in multidimensional output study is extensive
Applied to Prediction of Stock Price, the prediction of emergency worker's injures and deaths, fault diagnosis, channel estimation, air fare prediction, forecasting wind speed,
The fields such as computer vision.Multidimensional output regression is intended to the functional dependence between Mining Multidimensional input variable and multidimensional outcome variable
Relationship, and predict the following multidimensional output valve for inputting sample simultaneously using learned multivariate regression estimation function.Its basic assumption
It is between multidimensional outcome variable that there are correlations, and learn this correlation to obtain better estimated performance.
The research of existing multi output support vector regression model can be divided into two classes: the method and algorithm adaptability of problem conversion
Method.The method of problem conversion is that multi output problem is converted to multiple single output problems, exports supporting vector by means of single
The method of recurrence models different single output problems, is become by integrating multiple and different single output regression models and exporting to multidimensional
Amount is predicted.This method is convenient to carry out, but the calculation amount in training process is but very big, while having ignored output variable
Between potential correlation, independently different output variables are predicted, may cause and predict inaccurate problem.Algorithm is suitable
The method of answering property is to could be adjusted to processing multi output problem by the algorithm to single output problem, and this method can not only be right simultaneously
Multidimensional outcome variable is predicted, also can sufficiently be constructed and be explained the correlation between output variable.With the method for problem conversion
It compares, the method for algorithm adaptability can be easier to explain multi-output regression model, while also ensure that acquisition one is more accurate
Multidimensional exports prediction model.
However existing multi output support vector regression model is only realized and is carried out from the correlation between output variable
Preliminary exploration, however the problem of some keys still has to be solved, is mainly reflected in:
Relativity problem.The correlation analysis of big data can for big data depth analysis and excavate so find things in
Rule provides " navigation " function, and the correlation analysis for incorporating big data can sufficiently excavate correlation between output variable, while
Help to improve the estimated performance of multi output learning system.However, existing multi output support vector regression does not merge big data
Correlation analysis is modeled.
Lack the stability and generalization of model.Generalization ability is the important indicator for evaluating a model superiority and inferiority, is also
It practises algorithm and theoretical guarantee is provided.The extensive error of single output support vector regression has been widely studied, and has more been stepped up
The upper error to gather, however the research of existing multi output support vector regression is also not directed to the research of stability and generalization, it
Cannot theoretically guarantee the good estimated performance of model.
The online adaptive capability problems of streaming big data.Existing multi-output regression model is in batch mode of learning mostly
Lower design, i.e., hypothesis possesses whole training examples before learning and can be by once learning to obtain final estimation function.But
If in practical application data it is excessive cannot disposably handle or simply because the arrival of new data and to abandon learning outcome, again
Total data is learnt, it will take a substantial amount of time and space resources.
Therefore, a technical problem for needing urgently to solve instantly is exactly: to adapt to big data background downflow system data
Feature further increases the estimated performance of multi-output regression, and that how to innovate proposes a kind of effective measures, existing to solve
The problem of with the presence of technology, meets the greater demand of practical application.
Summary of the invention
Technical problem solved by the present invention is having overcome the deficiencies of the prior art and provide a kind of based on the online of correlation
The stream data processing method of multidimensional output, effectively manages streaming big data, completes the analysis of streaming data, realizes information
The real-time prediction of all kinds of states and real-time fault diagnosis in scientific domain.
The technical solution of the invention is as follows: a kind of stream data processing side of the online multidimensional output based on correlation
Method includes the following steps:
Step S101 chooses the relevance measure between the multidimensional outcome variable of multivariate regression estimation function, then using stream
Formula data learn the relevance measure between multidimensional outcome variable, and the multidimensional output for obtaining multivariate regression estimation function becomes
Correlation between amount；
Step S102 establishes multidimensional outcome variable correlation regularization model, constructs the multidimensional output of multidimensional outcome variable
Correlation of variables learning system；
Step S103 is based on multidimensional outcome variable correlation study system, constructs and solve the multi output based on correlation
Support vector regression model；
Step S104, according to the new addition sample of error prediction in the estimation function of multi output support vector regression model
Each weight vectors are updated；
Step S105 calculates the extensive upper error of multi output support vector regression model, the degree upper bound of regretting；
Step S106, according to the multidimensional output training dataset changed with stream data to multi output support vector regression mould
Type and its estimation function are updated.
Using the form of covariance matrix to the multidimensional outcome variable of multivariate regression estimation function in the step S101
Correlation be indicated.
It is based on multidimensional outcome variable correlation study system in the step S103, constructs and solves based on correlation
The method of multi output support vector regression model are as follows:
(1) damage of learning training is carried out using the relevance measure between hypersphere loss function measurement multidimensional outcome variable
It loses；
(2) multi output is obtained using the Frobenius- norm of the constituted matrix of the corresponding weight vector of each multidimensional outcome variable
The complexity of support vector regression model；
(3) it obtains and exports training dataset with the multidimensional that stream data changes, it is minimum on multidimensional output training dataset
Change the hypersphere loss of all output variables, the correlation regularization term of model complexity and output variable, construction is based on phase
The multi output support vector regression model of closing property；
(4) multi output support vector regression model is solved using interior point method or second order cone optimization, how defeated is obtained
The estimation function of support vector regression model out.
Multi output support vector regression model is estimated according to the new addition sample of error prediction in the step S104
The method that each weight vectors are updated in meter function are as follows:
(1) the stream data sample being newly added is carried out using the estimation function of multi output support vector regression model pre-
It surveys, judges whether the stream data sample being newly added is correct output valve, be then calculated according to obtained correct output valve
The hypersphere of the new stream data sample loses；
(2) if hypersphere loss is 0, not to each weight vectors in the estimation function of multi output support vector regression model
It is updated, if hypersphere loss is not 0, training dataset is exported according to the multidimensional after the new stream data sample of addition, more
Each weight vectors in the estimation function of new multi output support vector regression model.
A kind of computer readable storage medium, the computer-readable recording medium storage has computer program, described
Computer program the step of the method as any such as claim 1- claim 4 is realized when being executed by processor
A kind of stream data processing system of the online multidimensional output of correlation, including it is learning system constructing module, how defeated
Support vector regression model generates update module out, in which:
Learning system constructing module chooses the relevance measure between the multidimensional outcome variable of multivariate regression estimation function, so
The relevance measure between multidimensional outcome variable is learnt using stream data afterwards, obtains the more of multivariate regression estimation function
Tie up the correlation between output variable；
Multi output support vector regression model generates update module, establishes multidimensional outcome variable correlation regularization model,
The multidimensional outcome variable correlation study system of multidimensional outcome variable is constructed, and solves the multi output supporting vector based on correlation
Regression model, estimation function, extensive upper error, the sorry degree upper bound；According to the new addition sample of error prediction to multi output branch
Each weight vectors in the estimation function of vector regression model are held to be updated.
The advantages of the present invention over the prior art are that:
(1) compared with prior art, the present invention being established by using big data correlation analysis and machine Learning Theory online
Multi output support vector regression algorithm, using the correlation of big data as starting point, by defining relevance measure, relational learning etc.
Correlation between basic conception Mining Multidimensional output variable, establishes the theoretical frame of output variable relational learning system, herein
The multi output support vector regression model based on correlation is constructed under frame, and there is good use value；
(2) present invention proposes the definition of the uniform and stable property of multi output support vector regression model, studies its analysis of stability
Analysis carries out theory analysis and estimation to the extensive upper error of model based on this, and then obtains the online multi output branch of two-way update
Vector regression algorithm is held, the extensive upper error and the sorry degree upper bound to the on-line Algorithm carry out theory analysis and estimation, improve
The estimated performance of multi-output regression.
Detailed description of the invention
Fig. 1 is a kind of stream data processing method flow chart of online multidimensional output based on correlation.
Specific embodiment
The present invention is by the correlation analysis of integrated use big data, Statistical Learning Theory, convex optimum theory, probability theory, engineering
Practise the basic skills and newest research results with the related disciplines such as data mining, from the high coupling of big data, magnanimity and
The features such as real-time, sets out, the method combined using theory deduction with experimental analysis, proposes a kind of based on the online of correlation
Multidimensional exports support vector regression algorithm, and the method achieve the necks such as the real-time online prediction that will be applied to multidimensional output system
Domain can further increase the estimated performance of multi-output regression.
The present invention is based on online multidimensional to export support vector regression method, including realizes process as follows:
To between the multidimensional weight vector of multivariate regression estimation function relevance measure and relational learning be defined；
Multidimensional outcome variable correlation regularization model is established, the relational learning system of multidimensional outcome variable is constructed；
Based on above-mentioned relational learning system, constructs and solve the multi output support vector regression model based on correlation；
The uniform and stable property of multi output support vector regression model is defined；
Weight in each output dimension of new addition sample and existing supporting vector to error prediction is updated；
The online multi output support vector regression algorithm of two-way update is constructed, in the extensive error for estimating proposed on-line Algorithm
Boundary and the sorry degree upper bound；
The stream data in big data is chosen, experimental analysis is effective with the online multi output support vector regression model of verifying
Property and high efficiency.
Multi output support vector regression model includes relevance measure, relational learning and the correlation between multidimensional outcome variable
The theoretical frame of property regularization.Multi output support vector regression model includes the definition of uniform and stable property, extensive upper error
Assessment.Online multi output support vector regression is related to each output dimension of the new addition sample to error prediction, existing supporting vector
On weight update.On-line learning algorithm includes the online multi output support vector regression algorithm of two-way update.Online multi output
Support vector regression algorithm includes extensive upper error, the sorry feasibility and high efficiency for spending the upper bound, on-line Algorithm.Extensive error
The upper bound is shown using VC dimension table.The sorry degree upper bound is indicated using the upper bound of loss function gradient and the number of update.On-line Algorithm
Feasibility and high efficiency be by carrying out experimental analysis in the streaming big data that multidimensional exports with compared with, when consideration multidimensional is defeated
Out when correlation between variable, precision of prediction is more much higher than the method for not considering correlation, while in the processing big number of streaming
According to when timeliness with higher.Further explanation and illustration is carried out to the method for the present invention with reference to the accompanying drawing.
The stream data processing method flow chart of online multidimensional output to be a kind of based on correlation as shown in Figure 1, this hair
It is bright that the flow diagram of online multidimensional output support vector regression method is provided, comprising:
Step S101 chooses the relevance measure between the multidimensional outcome variable of multivariate regression estimation function, then uses
Stream data learns the relevance measure between multidimensional outcome variable, obtains the multidimensional output of multivariate regression estimation function
Correlation between variable；
For the correlation from profound level between Mining Multidimensional output variable, matrix variables normal state in probability theory point is utilized
The concept of cloth, with the formal definition multidimensional outcome variable of the covariance matrix between the output variable of multivariate regression estimation function it
Between relevance measure and the basic conceptions such as relational learning；
Step S102 establishes multidimensional outcome variable correlation regularization model, constructs the multidimensional output of multidimensional outcome variable
Correlation of variables learning system；
Based on the theoretical frame of this correlation regularization for proposing multidimensional outcome variable, to construct multidimensional outcome variable
Relational learning system；
Step S103 is based on multidimensional outcome variable correlation study system, constructs and solve the multi output based on correlation
Support vector regression model；
In order to construct multi output support vector regression model, measured using hypersphere loss function multidimensional outcome variable it
Between relevance measure carry out learning training loss, with the constituted matrix of the corresponding weight vector of each multidimensional outcome variable
The complexity of Frobenius- norm multi output support vector regression model is obtained as stream data is continually changing more
Dimension output training dataset, the hypersphere loss of all output variables of minimization, model are multiple on multidimensional output training dataset
The correlation regularization term of miscellaneous degree and output variable, constructs the multi output support vector regression model based on correlation, utilizes
The multi output support vector regression model that the theoretical proof of convex optimization is proposed is convex Optimized model, solves institute based on duality theory
It is proposed model dual problem, using in convex optimization interior point method or second order cone optimization model is solved, to obtain
The estimation function of multi output support vector regression model.
Step S104 carries out the weight in each output dimension of existing supporting vector according to the new addition sample of error prediction
It updates；
The case where being continuously added training dataset at any time for stream data, in each round, all using current how defeated
Regression estimates function predicts the stream data sample being newly added out, judges whether the stream data sample being newly added is positive
True output valve, the hypersphere for obtaining the new stream data sample thereafter according to correct output valve obtained loses, if it loses
It is 0, then each weight vectors of current regression estimates function is not updated；Otherwise according to new stream data sample is added after
Multidimensional exports training dataset, updates each weight vectors in the estimation function of multi output support vector regression model；
Step S105 constructs the extensive upper error of multi output support vector regression model and the degree upper bound of regretting；
Theory analysis is carried out to above-mentioned proposed multi output support vector regression model, it is how defeated online to obtain two-way update
Support vector regression algorithm out provides the pseudocode of the algorithm；
In order to theoretically guarantee that the online multi output support vector regression algorithm of the two-way update proposed (i.e. support by multi output
Vector regression model) validity, use for reference the sorry degree of support vector regression model, property by analyzing the sorry degree is estimated
The multi output support vector regression model upper bound, estimates the extensive upper error of the on-line Algorithm；
Step S106 chooses the stream data in big data, experimental analysis and the online multi output support vector regression of verifying
The validity and high efficiency of model.
By carrying out experimental analysis in the streaming big data that multidimensional exports compared with, the online multi output proposed is verified
The feasibility and high efficiency of support vector regression algorithm, when considering the correlation between multidimensional outcome variable, precision of prediction ratio
The method for not considering correlation is much higher, while the timeliness with higher when handling streaming big data, obstructed when verifying
It crosses (when obtaining the hypersphere loss of the new stream data sample for 0 according to correct output valve obtained), is transferred to step
S104 updates each weight vectors in the estimation function of multi output support vector regression model, until online multi output supporting vector
Regression model is verified.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.
Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, of the invention
It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one
The widest scope of cause.
The content that description in the present invention is not described in detail belongs to the well-known technique of those skilled in the art.
Claims (6)
1. a kind of stream data processing method of the online multidimensional output based on correlation, it is characterised in that include the following steps:
Step S101 chooses the relevance measure between the multidimensional outcome variable of multivariate regression estimation function, then uses streaming number
Learn according to the relevance measure between multidimensional outcome variable, obtain multivariate regression estimation function multidimensional outcome variable it
Between correlation；
Step S102 establishes multidimensional outcome variable correlation regularization model, constructs the multidimensional outcome variable of multidimensional outcome variable
Correlation study system；
Step S103 is based on multidimensional outcome variable correlation study system, constructs and solves the multi output based on correlation and supports
Vector regression model；
Step S104, according to the new addition sample of error prediction to respectively being weighed in the estimation function of multi output support vector regression model
Weight vector is updated；
Step S105 calculates the extensive upper error of multi output support vector regression model, the degree upper bound of regretting；
Step S106, according to the multidimensional output training dataset changed with stream data to multi output support vector regression model and
Its estimation function is updated.
2. a kind of stream data processing method of online multidimensional output based on correlation according to claim 1, special
Sign is: using the form of covariance matrix to the multidimensional outcome variable of multivariate regression estimation function in the step S101
Correlation is indicated.
3. a kind of stream data processing method of online multidimensional output based on correlation according to claim 1 or 2,
It is characterized in that: based on multidimensional outcome variable correlation study system in the step S103, constructing and solve based on correlation
Multi output support vector regression model method are as follows:
(1) loss of learning training is carried out using the relevance measure between hypersphere loss function measurement multidimensional outcome variable；
(2) multi output is obtained using the Frobenius- norm of the constituted matrix of the corresponding weight vector of each multidimensional outcome variable to support
The complexity of vector regression model；
(3) it obtains and exports training dataset with the multidimensional that stream data changes, the minimization institute on multidimensional output training dataset
There are the hypersphere loss of output variable, the correlation regularization term of model complexity and output variable, construction is based on correlation
Multi output support vector regression model；
(4) multi output support vector regression model is solved using interior point method or second order cone optimization, obtains multi output branch
Hold the estimation function of vector regression model.
4. a kind of stream data processing method of online multidimensional output based on correlation according to claim 1 or 2,
It is characterized in that: multi output support vector regression model being estimated according to the new addition sample of error prediction in the step S104
The method that each weight vectors are updated in meter function are as follows:
(1) the stream data sample being newly added is predicted using the estimation function of multi output support vector regression model, is sentenced
Whether the disconnected stream data sample being newly added is correct output valve, and the new stream then is calculated according to obtained correct output valve
The hypersphere of formula data sample loses；
(2) if hypersphere loss is 0, weight vectors each in the estimation function of multi output support vector regression model are not carried out
It updates, if hypersphere loss is not 0, training dataset is exported according to the multidimensional after the new stream data sample of addition, is updated more
Export each weight vectors in the estimation function of support vector regression model.
5. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, feature
It is, the step such as any the method for claim 1- claim 4 is realized when the computer program is executed by processor
Suddenly.
6. the stream data processing system that a kind of online multidimensional of correlation exports, it is characterised in that construct mould including learning system
Block, multi output support vector regression model generate update module, in which:
Learning system constructing module is chosen the relevance measure between the multidimensional outcome variable of multivariate regression estimation function, is then made
The relevance measure between multidimensional outcome variable is learnt with stream data, the multidimensional for obtaining multivariate regression estimation function is defeated
Correlation between variable out；
Multi output support vector regression model generates update module, establishes multidimensional outcome variable correlation regularization model, constructs
The multidimensional outcome variable correlation study system of multidimensional outcome variable, and solve the multi output support vector regression based on correlation
Model, estimation function, extensive upper error, the sorry degree upper bound；According to the new addition sample of error prediction to multi output support to
Each weight vectors in the estimation function of regression model are measured to be updated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811035273.4A CN109165244A (en) | 2018-09-06 | 2018-09-06 | A kind of stream data processing method of the online multidimensional output based on correlation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811035273.4A CN109165244A (en) | 2018-09-06 | 2018-09-06 | A kind of stream data processing method of the online multidimensional output based on correlation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109165244A true CN109165244A (en) | 2019-01-08 |
Family
ID=64894244
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811035273.4A Pending CN109165244A (en) | 2018-09-06 | 2018-09-06 | A kind of stream data processing method of the online multidimensional output based on correlation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109165244A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101498458A (en) * | 2009-03-02 | 2009-08-05 | 杭州电子科技大学 | Model updating method for on-line boiler combustion optimization |
| CN104715292A (en) * | 2015-03-27 | 2015-06-17 | 上海交通大学 | City short-term water consumption prediction method based on least square support vector machine model |
| CN104731778A (en) * | 2013-12-18 | 2015-06-24 | 中国电子信息产业集团有限公司第六研究所 | Active danger prevention method based on online time sequence |
-
2018
- 2018-09-06 CN CN201811035273.4A patent/CN109165244A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101498458A (en) * | 2009-03-02 | 2009-08-05 | 杭州电子科技大学 | Model updating method for on-line boiler combustion optimization |
| CN104731778A (en) * | 2013-12-18 | 2015-06-24 | 中国电子信息产业集团有限公司第六研究所 | Active danger prevention method based on online time sequence |
| CN104715292A (en) * | 2015-03-27 | 2015-06-17 | 上海交通大学 | City short-term water consumption prediction method based on least square support vector machine model |
Non-Patent Citations (3)
| Title |
|---|
| WHITENIGHTWU: "机器学习基础-各种学习方式（27）-在线学习", 《HTTPS://BLOG.CSDN.NET/WYDBYXR/ARTICLE/DETAILS/81261847》 * |
| 崔庆安，何桢，车建国: "一种基于支持向量机的非参数双响应曲面法", 《天津大学学报》 * |
| 李建伟，杨晓光，汪友华: "多维输出 SVR的ECT两相流图像重建方法", 《计算机工程与应用》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fei et al. | DS-VIKOR: A new multi-criteria decision-making method for supplier selection | |
| Sun et al. | Using Bayesian deep learning to capture uncertainty for residential net load forecasting | |
| WO2022262500A1 (en) | Steof-lstm-based method for predicting marine environmental elements | |
| CN110619420A (en) | Attention-GRU-based short-term residential load prediction method | |
| Ringsquandl et al. | Semantic-guided feature selection for industrial automation systems | |
| CN109325880A (en) | A kind of Mid-long term load forecasting method based on Verhulst-SVM | |
| CN105893669A (en) | Global simulation performance predication method based on data digging | |
| Li et al. | Grey relational decision making model of three-parameter interval grey number based on AHP and DEA | |
| Yang et al. | Physics-informed multi-state temporal frequency network for RUL prediction of rolling bearings | |
| Darudi et al. | Partial mutual information based algorithm for input variable selection for time series forecasting | |
| Rodrigues et al. | A system for analysis and prediction of electricity-load streams | |
| CN109165244A (en) | A kind of stream data processing method of the online multidimensional output based on correlation | |
| Şişman | A comparison of ARIMA and grey models for electricity consumption demand forecasting: The case of Turkey | |
| Xie et al. | Optimal sampling designs for multidimensional streaming time series with application to power grid sensor data | |
| Li et al. | Research on population development trend in Huizhou of China forecast based on optimal weighted combination method and fractional grey model | |
| Niu et al. | A distributed stochastic proximal-gradient algorithm for composite optimization | |
| Hassan | The interplay between the Bayesian and frequentist approaches: a general nesting spatial panel data model | |
| Liu et al. | Probabilistic double hierarchy linguistic risk analysis based on failure mode and effects analysis and S-ARAS method | |
| CN105491157A (en) | User collaborative regularization-based personalized Web service composition method | |
| Dette et al. | Identifying shifts between two regression curves | |
| XU et al. | Maximal clique searching algorithm for hypergraphs | |
| Yaodong et al. | Probabilistic short-term wind speed forecasting using a novel ensemble QRNN | |
| Zhu et al. | Time Granularity Setting Principle for Short-Term Passenger Flow Prediction in Urban Rail Transit | |
| Yang et al. | Applying support vector machine method to forecast electricity consumption | |
| Musa | A Hybrid Modern and Classical Algorithm for Indonesian Electricity Demand Forecasting |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200108 Address after: 266000 No. 1 Wenhai Road, Aoshan Weizhen, Jimo District, Qingdao City, Shandong Province Applicant after: Qingdao Marine Science and technology pilot National Laboratory Development Center Applicant after: BEIJIGN INSTITUTE OF AEROSPACE CONTROL DEVICES Address before: 142 box 403, box 100854, Beijing, Beijing, Haidian District Applicant before: BEIJIGN INSTITUTE OF AEROSPACE CONTROL DEVICES |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190108 |