CN108764295A - A kind of soft-measuring modeling method based on semi-supervised integrated study - Google Patents
A kind of soft-measuring modeling method based on semi-supervised integrated study Download PDFInfo
- Publication number
- CN108764295A CN108764295A CN201810400942.7A CN201810400942A CN108764295A CN 108764295 A CN108764295 A CN 108764295A CN 201810400942 A CN201810400942 A CN 201810400942A CN 108764295 A CN108764295 A CN 108764295A
- Authority
- CN
- China
- Prior art keywords
- exemplar
- sample
- soft
- label
- unlabeled exemplars
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
Abstract
The invention discloses a kind of soft-measuring modeling methods based on semi-supervised integrated study, belong to complex industrial process modeling and hard measurement field.Chemical process for there is exemplar negligible amounts.This method is a kind of based on semi-supervised on-line prediction strategy.Unlabeled exemplars collection is divided by three sub- sample sets using Bagging algorithms, and using has exemplar to train three regression models；Then, its corresponding index value is calculated to unlabeled exemplars based on a kind of confidence indicator, the unlabeled exemplars that selection meets confidence level requirement are marked, and this sample after marking is added to corresponding has label subsample collection；Finally, there is label data collection to establish Gaussian process regression model three after expansion, result is merged using method of weighting.It can effectively utilize the unlabeled exemplars information in chemical process, realization accurately to predict key variables, to improve product quality, reduce production cost.
Description
Technical field
The present invention relates to a kind of soft-measuring modeling methods based on semi-supervised integrated study, belong to complex industrial process modeling
With hard measurement field.
Background technology
Some important quality variables in the industrial process such as chemical industry, metallurgy and fermentation, can not often be surveyed by in-line meter
Amount, and there are serious lag by way of the off-line analysis of laboratory.Soft-measuring modeling method based on data, without deep
The mechanism knowledge for having entered solution preocess has many advantages, such as that maintenance cost is low, low measurement delay, in recent years in industrial process modeling
Extensive use is arrived.Traditional soft-measuring modeling method only considers there is exemplar information in industrial process, has abandoned a large amount of
Unlabeled exemplars, however in real process, there is the quantity of exemplar to be far less than unlabeled exemplars, use at this time conventional
Soft-measuring modeling method is unable to reach the precision in ideal.Soft-measuring modeling method based on semi-supervised learning thought is using less
While having exemplar of amount, and can receive extensive attention and answer in conjunction with the implicit information for including in unlabeled exemplars
With.
The semi-supervised learning method of mainstream has production method, semi-supervised SVM methods, the method based on figure and is based at present
The method of disagreement.Three kinds of front semi-supervised learning method is all made of single learning machine and is utilized to unlabeled exemplars, and is based on
The method of disagreement utilizes multiple learning machines, and the utilization to unlabeled exemplars is realized by the difference between learning machine.Because half supervises
It superintends and directs integrated study and originates from the semi-supervised learning method based on disagreement, so it effectively combines semi-supervised learning and integrated study
The advantages of, not only had the advantages that semi-supervised learning expanded a small amount of marked sample using unlabeled exemplars, but also learn with integrated
Practise the advantages of otherness between enhancing grader promotes whole classifier performance.Wherein, Tri-training algorithms are answered extensively
For classification problem, Tri-training overcomes hypothesis requirement of the coorinated training algorithm to sample set redundant views, has
Better Generalization Capability；But Tri-training by using two learning machines (or grader) in three learning machines to not
The consistency of marker samples carries out selected marker to unmarked sample, therefore can have not side there are two confidence indicator generation
Just the problem of unmarked sample being selected.
The acquisition frequency of leading variable is thought well below auxiliary variable based on semi-supervised learning in actual industrial process at present
The soft-measuring modeling method thought has that model prediction performance is not high, it is inaccurate to predict key variables, so as to cause
The problem that product quality is low and production cost is high.
Invention content
In order to solve presently, there are due to leading variable in actual industrial process acquisition frequency well below auxiliary become
Amount, the soft-measuring modeling method based on semi-supervised learning thought are predicted key variables inadequate there are model prediction performance is not high
Accurate problem, the present invention provides the soft-measuring modeling method based on semi-supervised integrated study, the technical solution is as follows：
A kind of soft-measuring modeling method based on semi-supervised integrated study, the method includes：
Step 1：Gatherer process has exemplar collection L={ XL,YL, L indicates label；With unlabeled exemplars collection U=
{XU, U indicates no label, and three unlabeled exemplars subset U are generated using Bagging algorithms to unlabeled exemplars collection U1、U2、U3；
Step 2：Initial soft-sensing model, f are established using there is exemplar subseti=Learn (Li), Learn is hard measurement
Modeling method initially has exemplar subset Li=L, i=1,2,3；
Step 3：For U1In each sample xu, using nearest neighbour method respectively from L2And L3In to select num distance close
Sample, obtain neighbour's sample set Ω2And Ω3；
Step 4：As formula (1) uses f2And f3To xuIt is predicted, obtains xuPseudo- label yu,2And yu,3, { xu,yu,jClaimed
For pseudo- marker samples, to thering is exemplar subset to establish the soft-sensing model as shown in formula (2) after the pseudo- marker samples of addition；
yu,j=fj(xu), j=2,3 (1)
Step 5：X is calculated according to formula (3)uIn learning machine f2And f3Under corresponding confidence indicator, according to formula (4) to two
A confidence level is merged to obtain final confidence indicator Threshold；
Wherein ΩjFor unlabeled exemplars xuRespectively from L2And L3In neighbour's sample set for selecting, yiFor num neighbour's sample
True tag value, fj(xi) it is xiCorresponding initial model prediction result, f 'j(xi) it is the pseudo- marker samples x of additionuIt establishes later
Soft-sensing model prediction result；
Step 6：Add according to the Threshold unlabeled exemplars for selecting confidence level high and to it label, corresponding label
Value isAnd the pseudo- marker samples { x after marking additionu,yuIt is added to L1There is exemplar concentration,
Threshold values are smaller, and to represent confidence level higher；
Step 7：Similarly using U2And U3To L2And L3Carry out sample expansion；
Step 8：The 2- steps that repeat the above steps 7T times, until reaching maximum iteration or L1、L2、L3No longer change
Become；
Step 9：There is exemplar collection L using updated1、L2、L3, hard measurement regression model is established, f is obtainedi=
Learn(Li)；
Step 10：For sample of newly arriving, model f is respectively adoptedi, i=1,2,3 are predicted, using mean value amalgamation mode,
The predicted value of each model is merged, final prediction output valve is obtained.
Optionally, the leading variable value that missing is estimated using Tri-training regression algorithms, establishes the single of fusion
Confidence calculations mode carries out selected marker to no label, and the model of otherness is established using the thought of integrated study.
The soft-measuring modeling method based on semi-supervised integrated study can be applied to the industrial mistake such as chemical industry, metallurgy and fermentation
Cheng Zhong.
The advantageous effect that technical solution provided by the invention is brought is：
It, will using Bagging algorithms by using the leading variable sample of Tri-training GPR algorithms estimation missing
Unlabeled exemplars collection is divided into three unlabeled exemplars subsets, to there is three learning machines of exemplar collection training；Using trained
Unlabeled exemplars are marked in learning machine, are set for two in a kind of new confidence indicator substitution tradition Tri-training of proposition
This puppet marker samples is added to learning machine by the calculating of reliability index if the confidence level of pseudo- marker samples meets threshold requirement
There is exemplar concentration, expanding three has exemplar subset, has exemplar subset according to after expansion, establishes corresponding
GPR models predict the query sample newly to arrive, are melted using the prediction output of three models of mean value amalgamation mode pair
It closes, obtains final prediction output.Key variables are accurately predicted to realize, it is low to have reached raising product quality
And reduce the purpose of production cost.
Description of the drawings
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, other are can also be obtained according to these attached drawings
Attached drawing.
Fig. 1 is SSEGPR modeling procedures method flow diagram provided in an embodiment of the present invention；
Fig. 2 is the standard error (root-mean- of various method predictions under different label rates provided in an embodiment of the present invention
square error,RMSE)；
Fig. 3 is the prediction result scatter plot of the various methods under 25% label rate.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached drawing to embodiment party of the present invention
Formula is described in further detail.
Embodiment one：
The present embodiment is in conjunction with common chemical process --- for debutanizing tower process, referring to Fig. 1, experimental data comes from
Debutanizing tower E processes, with the soft-measuring modeling method provided by the invention based on semi-supervised integrated study to butane concentration into
Row prediction：
Step 1：Gatherer process has exemplar collection L={ XL,YL, L indicates label；With unlabeled exemplars collection U=
{XU, U indicates no label, and three unlabeled exemplars subset U are generated using Bagging algorithms to unlabeled exemplars collection U1、U2、U3。
Step 2：Initial GPR models, f are established using there is exemplar subseti=GPR (Li), initially there is exemplar subset
Li=L, i=1,2,3.
Step 3：For U1In each sample xu, using nearest neighbour method respectively from L2And L3In to select num distance close
Sample, obtain neighbour's sample set Ω2And Ω3。
Step 4：As formula (1) uses f2And f3To xuIt is predicted, obtains xuPseudo- label yu,2And yu,3, { xu,yu,jClaimed
For pseudo- marker samples, to thering is exemplar subset to establish the soft-sensing model as shown in formula (2) after the pseudo- marker samples of addition；
yu,j=fj(xu), j=2,3 (1)
Step 5：X is calculated according to formula (3)uIn learning machine f2And f3Under corresponding confidence indicator, according to formula (4) to two
A confidence level is merged to obtain final confidence indicator Threshold；
Wherein ΩjFor unlabeled exemplars xuRespectively from L2And L3In neighbour's sample set for selecting, yiFor num neighbour's sample
True tag value, fj(xi) it is xiCorresponding initial model prediction result, f 'j(xi) it is the pseudo- marker samples x of additionuIt establishes later
Soft-sensing model prediction result.
Step 6：Threshold values are smaller to represent that confidence level is higher, and the unlabeled exemplars for selecting confidence level high add it
Label, corresponding mark value areAnd the pseudo- marker samples { x after marking additionu,yuIt is added to L1There is mark
This concentration of signed-off sample.
Step 7：Similarly using U2And U3To L2And L3Carry out sample expansion.
Step 8：The 2- steps that repeat the above steps 7T times, until reaching maximum iteration or L1、L2、L3No longer change
Become.
Step 9：There is exemplar collection L using updated1、L2、L3, GPR models are established, f is obtainedi=GPR (Li)。
Data-oriented collection { X, y }, wherein X ∈ Rn×m, y ∈ Rn×1, n sample points, m is sample dimension.Between input and output
Meet shown in formula (5)
Y=f (x)+ε (5)
It is 0 that ε, which is mean value, in formula, and variance isGaussian noise, f is unknown functional form.GPR assumes regression function y
There is=f (x) Gaussian prior of zero-mean to be distributed, and describe such as formula (6)
Y~N (0, C) (6)
C is the covariance matrix of n × n in formula, and i row j column elements are defined as Cij=C (xi,xj；θ), covariance matrix is logical
It crosses kernel function to be calculated, square index covariance kernel function is chosen in text, define as shown in formula (7)
δ in formulaij=1 only sets up in i=j, otherwise δij=0, l are variance measure,For signal variance,For noise
Variance,Selection for the hyper parameter of GPR, hyper parameter has a significant impact to model performance, using maximum likelihood
The estimation technique can obtain optimal hyper parameter.
For new sample xq, shown in corresponding GPR models output mean value and variance such as formula (8) and (9)
Wherein c (xq)=[C (xq,x1),...,C(xq,xn)] it is the covariance matrix newly arrived between sample and training sample, C
It is the covariance matrix between training sample, C (xq,xq) it is the auto-covariance of sample of newly arriving.
Step 10：For sample of newly arriving, model f is respectively adoptedi, i=1,2,3 are predicted, using mean value amalgamation mode,
The predicted value of each model is merged, final prediction output valve is obtained.
Fig. 2 is the standard variance RMSE of the various method predictions under different label rates；Fig. 3 is not Tongfang under 25% label rate
The prediction result scatter plot of method；As seen from the figure, the soft-measuring modeling method based on semi-supervised integrated study can be utilized effectively
Unlabeled exemplars information, being capable of accurate prediction butane concentration.
The embodiment of the present invention is used by using the leading variable sample of Tri-training GPR algorithms estimation missing
Unlabeled exemplars collection is divided into three unlabeled exemplars subsets by Bagging algorithms, to there is three study of exemplar collection training
Machine；Unlabeled exemplars are marked using trained learning machine, propose a kind of new confidence indicator substitution tradition Tri-
The calculating of two confidence indicators in training, if the confidence level of pseudo- marker samples meets threshold requirement, by this puppet label sample
Originally be added to learning machine has exemplar concentration, and expanding three has exemplar subset, has exemplar according to after expansion
Subset is established corresponding GPR models, is predicted the query sample newly to arrive, using three models of mean value amalgamation mode pair
Prediction output merged, obtain final prediction output.Key variables are accurately predicted to realize, are reached
It is low and reduce the purpose of production cost to improve product quality.
Part steps in the embodiment of the present invention can utilize software realization, and corresponding software program can be stored in can
In the storage medium of reading, such as CD or hard disk.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
Claims (2)
1. a kind of soft-measuring modeling method based on semi-supervised integrated study, which is characterized in that the method includes：
Step 1：Gatherer process has exemplar collection L={ XL,YL, L indicates label；With unlabeled exemplars collection U={ XU, U tables
Show no label, three unlabeled exemplars subset U are generated using Bagging algorithms to unlabeled exemplars collection U1、U2、U3；
Step 2：Initial soft-sensing model, f are established using there is exemplar subseti=Learn (Li), Learn is soft sensor modeling
Method initially has exemplar subset Li=L, i=1,2,3；
Step 3：For U1In each sample xu, using nearest neighbour method respectively from L2And L3In select num apart from close sample
This, obtains neighbour's sample set Ω2And Ω3；
Step 4：As formula (1) uses f2And f3To xuIt is predicted, obtains xuPseudo- label yu,2And yu,3, { xu,yu,jIt is referred to as puppet
Marker samples, to thering is exemplar subset to establish the soft-sensing model as shown in formula (2) after the pseudo- marker samples of addition；
yu,j=fj(xu), j=2,3 (1)
f′j=Learn (Lj∪{xu,yu,j}) (2)
Step 5：X is calculated according to formula (3)uIn learning machine f2And f3Under corresponding confidence indicatorTwo are set according to formula (4)
Reliability is merged to obtain final confidence indicator Threshold；
Wherein ΩjFor unlabeled exemplars xuRespectively from L2And L3In neighbour's sample set for selecting, yiFor the true of num neighbour's sample
Real label value, fj(xi) it is xiCorresponding initial model prediction result, f 'j(xi) it is the pseudo- marker samples x of additionuThat establishes later is soft
The prediction result of measurement model；
Step 6：Label is added according to the Threshold unlabeled exemplars for selecting confidence level high and to it, corresponding mark value is yu
=yu,2+yu,3/ 2, and the pseudo- marker samples { x after marking will be addedu,yuIt is added to L1There are exemplar concentration, Threshold
It is worth that smaller to represent confidence level higher；
Step 7：Similarly using U2And U3To L2And L3Carry out sample expansion；
Step 8：The 2- steps that repeat the above steps 7T times, until reaching maximum iteration or L1、L2、L3No longer change；
Step 9：There is exemplar collection L using updated1、L2、L3, hard measurement regression model is established, f is obtainedi=Learn
(Li)；
Step 10：For sample of newly arriving, model f is respectively adoptedi, i=1,2,3 are predicted, using mean value amalgamation mode, to each
The predicted value of model is merged, and final prediction output valve is obtained.
2. according to the method described in claim 1, it is characterized in that, estimating missing using Tri-training regression algorithms
Leading variable value, the single confidence calculations mode for establishing fusion carries out selected marker to no label, using the think of of integrated study
Want to establish the model of otherness.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810400942.7A CN108764295B (en) | 2018-04-28 | 2018-04-28 | Method for predicting concentration of butane at bottom of debutanizer tower based on soft measurement modeling of semi-supervised ensemble learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810400942.7A CN108764295B (en) | 2018-04-28 | 2018-04-28 | Method for predicting concentration of butane at bottom of debutanizer tower based on soft measurement modeling of semi-supervised ensemble learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108764295A true CN108764295A (en) | 2018-11-06 |
| CN108764295B CN108764295B (en) | 2021-05-28 |
Family
ID=64012455
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810400942.7A Active CN108764295B (en) | 2018-04-28 | 2018-04-28 | Method for predicting concentration of butane at bottom of debutanizer tower based on soft measurement modeling of semi-supervised ensemble learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108764295B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109829497A (en) * | 2019-01-31 | 2019-05-31 | 清华四川能源互联网研究院 | A kind of area user identification zone and method of discrimination based on supervised learning |
| CN111860863A (en) * | 2020-07-17 | 2020-10-30 | 江南大学 | Active learning method based on covariance matrix |
| CN111914492A (en) * | 2020-04-28 | 2020-11-10 | 昆明理工大学 | Evolution optimization-based semi-supervised learning industrial process soft measurement modeling method |
| CN111985103A (en) * | 2020-08-20 | 2020-11-24 | 江南大学 | Sample processing method and system based on kernel limit learning machine |
| CN112650063A (en) * | 2020-12-30 | 2021-04-13 | 浙江大学 | Self-adaptive soft measurement method based on semi-supervised incremental Gaussian mixture regression |
| CN112749841A (en) * | 2020-12-30 | 2021-05-04 | 科大国创云网科技有限公司 | User public praise prediction method and system based on self-training learning |
| CN113158473A (en) * | 2021-04-27 | 2021-07-23 | 昆明理工大学 | Semi-supervised integrated instant learning industrial rubber compound Mooney viscosity soft measurement method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102693452A (en) * | 2012-05-11 | 2012-09-26 | 上海交通大学 | Multiple-model soft-measuring method based on semi-supervised regression learning |
| CN102708294A (en) * | 2012-05-11 | 2012-10-03 | 上海交通大学 | Self-adaptive parameter soft measuring method on basis of semi-supervised local linear regression |
| CN103279030A (en) * | 2013-03-07 | 2013-09-04 | 清华大学 | Bayesian framework-based dynamic soft measurement modeling method and device |
| CN104699894A (en) * | 2015-01-26 | 2015-06-10 | 江南大学 | JITL (just-in-time learning) based multi-model fusion modeling method adopting GPR (Gaussian process regression) |
| CN104914723A (en) * | 2015-05-22 | 2015-09-16 | 浙江大学 | Industrial process soft measurement modeling method based on cooperative training partial least squares model |
| CN107451102A (en) * | 2017-07-28 | 2017-12-08 | 江南大学 | A kind of semi-supervised Gaussian process for improving self-training algorithm returns soft-measuring modeling method |
-
2018
- 2018-04-28 CN CN201810400942.7A patent/CN108764295B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102693452A (en) * | 2012-05-11 | 2012-09-26 | 上海交通大学 | Multiple-model soft-measuring method based on semi-supervised regression learning |
| CN102708294A (en) * | 2012-05-11 | 2012-10-03 | 上海交通大学 | Self-adaptive parameter soft measuring method on basis of semi-supervised local linear regression |
| CN103279030A (en) * | 2013-03-07 | 2013-09-04 | 清华大学 | Bayesian framework-based dynamic soft measurement modeling method and device |
| CN104699894A (en) * | 2015-01-26 | 2015-06-10 | 江南大学 | JITL (just-in-time learning) based multi-model fusion modeling method adopting GPR (Gaussian process regression) |
| CN104914723A (en) * | 2015-05-22 | 2015-09-16 | 浙江大学 | Industrial process soft measurement modeling method based on cooperative training partial least squares model |
| CN107451102A (en) * | 2017-07-28 | 2017-12-08 | 江南大学 | A kind of semi-supervised Gaussian process for improving self-training algorithm returns soft-measuring modeling method |
Non-Patent Citations (1)
| Title |
|---|
| WEI ZHANG ET AL.: "Adaptive Soft Sensor for Online Prediction Based on", 《2015 INTERNATIONAL CONFERENCE ON CONTROL, AUTOMATION AND INFORMATION SCIENCES (ICCAIS)》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109829497A (en) * | 2019-01-31 | 2019-05-31 | 清华四川能源互联网研究院 | A kind of area user identification zone and method of discrimination based on supervised learning |
| CN111914492A (en) * | 2020-04-28 | 2020-11-10 | 昆明理工大学 | Evolution optimization-based semi-supervised learning industrial process soft measurement modeling method |
| CN111914492B (en) * | 2020-04-28 | 2022-09-13 | 昆明理工大学 | Evolution optimization-based semi-supervised learning industrial process soft measurement modeling method |
| CN111860863A (en) * | 2020-07-17 | 2020-10-30 | 江南大学 | Active learning method based on covariance matrix |
| CN111860863B (en) * | 2020-07-17 | 2023-04-07 | 江南大学 | Active learning method based on covariance matrix |
| CN111985103A (en) * | 2020-08-20 | 2020-11-24 | 江南大学 | Sample processing method and system based on kernel limit learning machine |
| CN112650063A (en) * | 2020-12-30 | 2021-04-13 | 浙江大学 | Self-adaptive soft measurement method based on semi-supervised incremental Gaussian mixture regression |
| CN112749841A (en) * | 2020-12-30 | 2021-05-04 | 科大国创云网科技有限公司 | User public praise prediction method and system based on self-training learning |
| CN113158473A (en) * | 2021-04-27 | 2021-07-23 | 昆明理工大学 | Semi-supervised integrated instant learning industrial rubber compound Mooney viscosity soft measurement method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108764295B (en) | 2021-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108764295A (en) | A kind of soft-measuring modeling method based on semi-supervised integrated study | |
| Zeng et al. | Deep-reinforcement-learning-based images segmentation for quantitative analysis of gold immunochromatographic strip | |
| Galante et al. | The challenge of modeling niches and distributions for data‐poor species: a comprehensive approach to model complexity | |
| CN107273490B (en) | Combined wrong question recommendation method based on knowledge graph | |
| CN107134144B (en) | A kind of vehicle checking method for traffic monitoring | |
| CN103577843B (en) | A kind of aerial hand-written character string recognition methods | |
| CN107563494A (en) | A kind of the first visual angle Fingertip Detection based on convolutional neural networks and thermal map | |
| CN107437100A (en) | A kind of picture position Forecasting Methodology based on the association study of cross-module state | |
| CN107451102A (en) | A kind of semi-supervised Gaussian process for improving self-training algorithm returns soft-measuring modeling method | |
| CN102693452A (en) | Multiple-model soft-measuring method based on semi-supervised regression learning | |
| CN106528417A (en) | Intelligent detection method and system of software defects | |
| CN105260805B (en) | A kind of antimony ore grade flexible measurement method selectively merged based on isomery grader | |
| CN111368920A (en) | Quantum twin neural network-based binary classification method and face recognition method thereof | |
| CN106933105A (en) | Track under confined condition updates integrated forecasting Iterative Learning Control Algorithm | |
| CN113807171B (en) | Text classification method based on semi-supervised transfer learning | |
| CN106203296B (en) | The video actions recognition methods of one attribute auxiliary | |
| CN109543731A (en) | A kind of three preferred Semi-Supervised Regression algorithms under self-training frame | |
| Matić et al. | Oscillating Adriatic temperature and salinity regimes mapped using the Self-Organizing Maps method | |
| CN115270007B (en) | POI recommendation method and system based on mixed graph neural network | |
| CN109447110A (en) | The method of the multi-tag classification of comprehensive neighbours' label correlative character and sample characteristics | |
| CN111695737A (en) | Group target advancing trend prediction method based on LSTM neural network | |
| CN104850868A (en) | Customer segmentation method based on k-means and neural network cluster | |
| CN110648215A (en) | Distributed scoring card model building method | |
| CN113516228A (en) | Network anomaly detection method based on deep neural network | |
| CN103235894A (en) | Self-adapting preferred fuzzy kernel clustering based naphtha attribute clustering method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221101 Address after: 230000 B-1015, wo Yuan Garden, 81 Ganquan Road, Shushan District, Hefei, Anhui. Patentee after: HEFEI MINGLONG ELECTRONIC TECHNOLOGY Co.,Ltd. Address before: No. 1800 Lihu Avenue, Wuxi City, Jiangsu Province Patentee before: Jiangnan University |