CN114330468A - Classifier screening method and system based on dynamic programming and computer equipment - Google Patents
Classifier screening method and system based on dynamic programming and computer equipment Download PDFInfo
- Publication number
- CN114330468A CN114330468A CN202110797631.0A CN202110797631A CN114330468A CN 114330468 A CN114330468 A CN 114330468A CN 202110797631 A CN202110797631 A CN 202110797631A CN 114330468 A CN114330468 A CN 114330468A
- Authority
- CN
- China
- Prior art keywords
- vus
- classifier
- dynamic programming
- variance
- vector matrix
- 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
- 238000012216 screening Methods 0.000 title claims abstract description 43
- 238000000034 method Methods 0.000 title claims abstract description 38
- 239000013598 vector Substances 0.000 claims abstract description 45
- 239000011159 matrix material Substances 0.000 claims abstract description 42
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 26
- 238000012163 sequencing technique Methods 0.000 claims abstract description 6
- 238000004364 calculation method Methods 0.000 claims description 30
- 238000004590 computer program Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 4
- 230000001174 ascending effect Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims 2
- 238000010801 machine learning Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 206010020772 Hypertension Diseases 0.000 description 1
- 208000001953 Hypotension Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000036772 blood pressure Effects 0.000 description 1
- 238000009530 blood pressure measurement Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000036543 hypotension Effects 0.000 description 1
- 229940126671 regional medicine Drugs 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a classifier screening method, a system and computer equipment based on dynamic programming, which comprises the following steps: sequencing the sample data, and converting the sample data into a vector matrix; obtaining VUS estimated values under three types of classifier ROC curved surfaces by using a dynamic programming algorithm according to the vector matrix; substituting the VUS estimated value and the vector matrix into the VUS variance expression to solve the VUS variance value; and comparing according to the VUS variance value to obtain a VUS maximum estimation value, and taking the classifier of the maximum estimation value as an optimal classifier. Sequencing the existing data, and converting the existing data into a vector matrix; obtaining VUS estimated values under three types of classifier ROC curved surfaces and corresponding variance values by using a dynamic programming algorithm according to the vector matrix; comparing VUS maximum estimation values obtained by a dynamic programming algorithm, and taking a classifier of the maximum estimation values as an optimal classifier, wherein the method is based on linear time and is faster than quadratic time; compared with the wegman method, the estimators are unbiased, making their computational load smaller.
Description
Technical Field
The invention relates to the field of machine learning, in particular to a classifier screening method and system based on dynamic programming and computer equipment.
Background
Classification is a supervised learning process in machine learning. The so-called classification problem, i.e. the classification of the examination results of a patient into sick and healthy by using a machine learning algorithm, is a medical classification problem (the data to be distinguished is divided into two categories). For another example, in an e-mail box, after receiving the mail, the e-mail box will classify our mail into advertisement mail, junk mail and normal mail, which is a multi-classification problem.
Receiver Operating Characteristic (ROC) analysis has found widespread use in both types of problems. An ROC curve is defined based on different decision threshold settings, which is a graph of false positive rate versus true positive rate (sensitivity). The area under the ROC curve (AUC) can be analytically or empirically calculated as an index to summarize the overall performance of the binary classifier. However, three types of problems are often encountered in practice, particularly in regional medicine. In order to evaluate the performance of the three classes of classifiers, researchers have proposed volume under three classes of ROC surfaces (VUS) as a value index, and others have proposed various methods to estimate the mean and variance of VUS. However, to our knowledge, existing methods are subject to heavy computational loads.
Chinese invention patent CN107562880A discloses "a classification result screening method and apparatus based on multi-stage classifier", published as 2018, month 01, 09: acquiring an initial request and a classification result of a multi-stage classifier; the classification result is the result of the multi-stage classifier on classification of the predicted text information; the classification result comprises a set of x classification paths, wherein x is a positive integer greater than or equal to 1; screening the classification result by using a preset rule corresponding to the initial request to obtain a target result; and the target result is output, a target result which is more accurate and meets the requirements and predicted text information are provided for a user, the interference of excessive results of the multi-stage classifier on useful information is avoided, the text classification efficiency and effect are improved, and in addition, the screening method can be arbitrarily combined, and the complex multi-directional screening requirement can be met. But the method uses multi-stage classifiers for screening any combination, and the computational load of the screening method is extremely large.
Disclosure of Invention
The invention provides a classifier screening method, a system and computer equipment based on dynamic programming, aiming at solving the technical defect of large computational load of the existing classifier screening method.
In order to realize the purpose, the technical scheme is as follows:
a classifier screening method based on dynamic programming comprises the following steps:
s1: sequencing the sample data, and converting the sample data into a vector matrix;
s2: obtaining VUS estimated values under the ROC curved surface of the three types of classifiers by using a dynamic programming algorithm according to the vector matrix;
s3: substituting the VUS estimated value and the vector matrix into the VUS variance expression to solve the VUS variance value;
s4: and comparing according to the VUS variance value to obtain a VUS maximum estimation value, and taking the classifier of the maximum estimation value as an optimal classifier.
In the scheme, the existing data is sequenced and converted into a vector matrix; obtaining VUS estimated values under the ROC curved surface of the three types of classifiers and corresponding variance values by using a dynamic programming algorithm according to the vector matrix; comparing VUS maximum estimation values obtained by a dynamic programming algorithm, and taking a classifier of the maximum estimation values as an optimal classifier, wherein the method is based on linear time and is faster than quadratic time; compared with the Wegener method, the estimation quantity is unbiased, so that the calculation load is smaller.
Preferably, in step S1, the vector matrix is solved as follows:
let D1，...，DN，N＝n1+n2+n3Is that
The D sequence is sequenced according to ascending order to obtain a sequence statistical information sequence;
D(1)＝…＝D(1)<…<D(j)＝…＝D(j)(＝D(i))<…<D(k)＝…＝D(K)
let a be when i 1i、bi、ciAre each X3k、X2j、X1iIn sequence with D(i)The same number of the technical vectors can be obtained
Forming a vector matrix:
n1、n2、n3sample sizes, X, of three classes, respectively3k、X2j、X1iSample sequences of three classes, respectively.
Preferably, in step S2, the VUS expression of the dynamic programming algorithm is:
wherein the content of the first and second substances,
is a VUS estimate.
Preferably, in the three classes of classifiers, the dynamic programmingThe items represent the number of events that satisfy the relationship in respective brackets, based on the dynamic programming item S1、S2、.......、S9The expression of (a) is:
n1、n2、n3sample sizes, X, of three classes, respectively3、X2、X1Are respectively a sample of three classesThe sequence is shown in the specification.
Preferably, the dynamic planning S1、S2、.......、S9The satisfying conditions are as follows:
S1，ε(X3>X2>X1)；
S2，ε(X3>X′3>X2>X1) Or epsilon (X'3>X3>X2>X1)；
S3，ε(X3>X2>X′2>X1) Or ε (X)3>X′2>X2>X1)；
S4，ε(X3>X2>X′1>X1) Or ε (X)3>X2>X1>X′1)；
S5，ε(X3>X′3>X2>X′2>X1) Or epsilon (X'3>X3>X2>X′2>X1)
ε(X3>X′3>X′2>X2>X1) Or epsilon (X'3>X3>X′2>X2>X1)；
S6，ε(X3>X2>X′3>X′2>X1) Or epsilon (X'3>X′2>X3>X2>X1)；
S7，ε(X3>X′3>X2>X1>X′1) Or epsilon (X'3>X3>X2>X1>X′1)
ε(X3>X′3>X2>X′1>X1) Or epsilon (X'3＞X′3＞X2＞X′1＞X1)；
S8，ε(X3>X2>X′2>X1>X′1) Or ε (X)3>X′2>X2>X1>X′1)
ε(X3>X2>X′2>X′1>X1) Or ε (X)3>X′2>X2>X′1>X1)；
S9，ε(X3>X2>X1>X′2>X′1) Or ε (X)3>X′2>X′1>X2>X1)。
Preferably, in step S3, according to the dynamic programming algorithm, the obtained variance expression of the VUS is as follows:
wherein
A classifier screening system based on dynamic programming applies a classifier screening method based on dynamic programming, and comprises the following steps: the VUS variance detection device comprises a screening module, a VUS calculation module, a VUS variance calculation module and a classification module; the data of the screening module is transmitted to a VUS calculation module, the data of the VUS calculation module is transmitted to a VUS variance calculation module, and the data of the VUS variance calculation module is transmitted to a classification module;
the screening module carries out sorting processing on the sample data and converts the sample data into a vector matrix;
the VUS calculation module obtains VUS estimated values of the three types of classifiers by using a dynamic programming algorithm according to the vector matrix;
the VUS variance calculation module substitutes the VUS estimated value and the vector matrix into the VUS variance expression to calculate a VUS variance value;
and the classification module compares the VUS maximum estimation value according to the VUS variance value, and takes the classifier of the maximum estimation value as an optimal classifier.
Preferably, the screening module sorts the sample data from small to large, and converts the sample data into a vector matrix.
Preferably, the VUS calculation module uses a dynamic programming algorithm to obtain a VUS estimation value under an ROC curved surface of the three types of classifiers according to the vector matrix.
A computer device is characterized by comprising a screening method and a screening system, wherein a computer program is stored in the device, and when the computer program is executed by a processor, the computer program realizes a classifier screening method based on dynamic programming.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a method for processing the existing data in a sequencing way, and converting the existing data into a vector matrix; obtaining VUS estimated values under three types of classifier ROC curved surfaces and corresponding variance values by using a dynamic programming algorithm according to the vector matrix; comparing the VUS maximum estimation value obtained under the dynamic programming algorithm, and taking the classifier of the maximum estimation value as an optimal classifier, wherein the method is based on linear time and is faster than secondary time; compared with the wegman method, the estimators are unbiased, making their computational load smaller.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 shows the vector matrix calculation S of the present invention1A diagram of;
FIG. 3 is a graph of the ROC surface profile of the present invention;
FIG. 4 is a block data flow diagram of the present invention.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
the invention is further illustrated below with reference to the figures and examples.
Example 1
As shown in fig. 1, fig. 2 and fig. 3, a classifier screening method based on dynamic programming includes the following steps:
s1: sequencing the sample data, and converting the sample data into a vector matrix;
s2: obtaining VUS estimated values under the ROC curved surface of the three types of classifiers by using a dynamic programming algorithm according to the vector matrix;
s3: substituting the VUS estimated value and the vector matrix into the VUS variance expression to solve the VUS variance value;
s4: and comparing according to the VUS variance value to obtain a VUS maximum estimation value, and taking the classifier of the maximum estimation value as an optimal classifier.
In the scheme, the existing data is sequenced and converted into a vector matrix; obtaining VUS estimated values under the ROC curved surface of the three types of classifiers and corresponding variance values by using a dynamic programming algorithm according to the vector matrix; comparing VUS maximum estimation values obtained by a dynamic programming algorithm, and taking a classifier of the maximum estimation values as an optimal classifier, wherein the method is based on linear time and is faster than quadratic time; compared with the Wegener method, the estimation quantity is unbiased, so that the calculation load is smaller.
Preferably, in step S1, the vector matrix is solved as follows:
let D1，...，DN，N＝n1+n2+n3Is that
The D sequence is sequenced according to ascending order to obtain a sequence statistical information sequence;
D(1)＝…＝D(1)<…<D(j)＝…＝D(j)(＝D(i))<…<D(k)＝…＝D(K)
let a be when i 1i、bi、ciAre each X3k、X2j、X1iIn sequence with D(i)The same number of the technical vectors can be obtained
Forming a vector matrix:
n1、n2、n3sample sizes, X, of three classes, respectively3k、X2j、X1iSample sequences of three classes, respectively.
In the foregoing solution, specifically, in an optional embodiment, the text of the sample data may be a news text. The classification problem is visible everywhere in life, such as gender classification, age classification, height classification, face recognition and the like. In the face of three classification problems, such as blood pressure measurement, three results can be obtained, hypotension, normotensive blood pressure, and hypertension, which are equivalent to three types of samples, forming ROC curve. For both three-class and multi-class problems, a dynamic programming algorithm can be used to solve the VUS and its variance.
Preferably, in step S2, the VUS expression of the dynamic programming algorithm is:
wherein the content of the first and second substances,
is a VUS estimate.
Preferably, in the three classes of classifiers, the dynamic programming term represents the number of events satisfying the relationship in respective brackets based on the dynamic programming term S1、S2、.......、S9The expression of (a) is:
n1、n2、n3sample sizes, X, of three classes, respectively3、X2、X1Sample sequences of three classes, respectively.
Preferably, the dynamic planning S1、S2、.......、S9The satisfying conditions are as follows:
S1，ε(X3>X2>X1)；
S2，ε(X3>X′3>X2>X1) Or epsilon (X'3>X3>X2>X1)；
S3，ε(X3>X2>X′2>X1) Or ε (X)3>X′2>X2>X1)；
S4，ε(X3>X2>X′1>X1) Or ε (X)3>X2>X1>X′1)；
S5，ε(X3>X′3>X2>X′2>X1) Or epsilon (X'3>X3>X2>X′2>X1)
ε(X3>X′3>X′2>X2>X1) Or epsilon (X'3>X3>X′2>X2>X1)；
S6，ε(X3>X2>X′3>X′2>X1) Or epsilon (X'3>X′2>X3>X2>X1)；
S7，ε(X3>X′3>X2>X1>X′1) Or epsilon (X'3>X3>X2>X1>X′1)
ε(X3>X′3>X2>X′1>X1) Or ε ((X)3＞X′3＞X2＞X′1>X1))；
S8，ε(X3>X2>X′2>X1>X′1) Or ε (X)3>X′2>X2>X1>X′1)
ε(X3>X2>X′2>X′1>X1) Or ε (X)3>X′2>X2>X′1>X1)；
S9，ε(X3>X2>X1>X′2>X′1) Or ε (X)3>X′2>X′1>X2>X1)。
Preferably, in step S3, according to the dynamic programming algorithm, the obtained variance expression of the VUS is as follows:
wherein
Example 2
As shown in fig. 4, a classifier screening system based on dynamic programming applies a classifier screening method based on dynamic programming, which includes: the VUS variance detection device comprises a screening module, a VUS calculation module, a VUS variance calculation module and a classification module; the data of the screening module is transmitted to the VUS calculation module, the data of the VUS calculation module is transmitted to the VUS variance calculation module, and the data of the VUS variance calculation module is transmitted to the classification module;
the screening module carries out sorting processing on the sample data and converts the sample data into a vector matrix;
the VUS calculation module obtains VUS estimated values of the three types of classifiers by using a dynamic programming algorithm according to the vector matrix;
the VUS variance calculation module substitutes the VUS estimated value and the vector matrix into the VUS variance expression to calculate a VUS variance value;
and the classification module compares the VUS maximum estimation value according to the VUS variance value, and takes the classifier of the maximum estimation value as an optimal classifier.
Preferably, the screening module sorts the sample data from small to large, and converts the sample data into a vector matrix. Preferably, the VUS calculation module uses a dynamic programming algorithm to obtain a VUS estimation value under an ROC curved surface of the three types of classifiers according to the vector matrix.
It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. It will be apparent to those skilled in the art that other variations and modifications can be made on the basis of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (10)
1. A classifier screening method based on dynamic programming is characterized by comprising the following steps:
s1: sequencing the sample data, and converting the sample data into a vector matrix;
s2: obtaining VUS estimated values under three types of classifier ROC curved surfaces by using a dynamic programming algorithm according to the vector matrix;
s3: substituting the VUS estimated value and the vector matrix into the VUS variance expression to solve the VUS variance value;
s4: and comparing according to the VUS variance value to obtain a VUS maximum estimation value, and taking the classifier of the maximum estimation value as an optimal classifier.
2. The dynamic programming-based classifier screening method according to claim 1, wherein in step S1, the vector matrix is solved as follows:
let D1，...，DN，N＝n1+n2+n3Is X11，...，
X21，...，X2n2，X31，...，
The D sequence is sequenced according to ascending order to obtain a sequence statistical information sequence;
D(1)＝…＝D(1)<…<D(j)＝…＝D(j)(＝D(i))<…<D(k)＝…＝D(K)
let a be when i 1i、bi、ciAre each X3k、X2j、X1iIn sequence with D(i)The same number of the technical vectors can be obtained
Forming a vector matrix:
n1、n2、n3sample sizes, X, of three classes, respectively3k、X2j、X1iSample sequences of three classes, respectively.
3. The dynamic programming-based classifier screening method of claim 2, wherein in step S2, the VUS expression of the dynamic programming algorithm is:
wherein the content of the first and second substances,
is a VUS estimate.
4. A dynamic programming based classification as claimed in claim 3The method for screening the classifier is characterized in that in the three classes of classifiers, the dynamic programming items represent the number of events which satisfy the relationship in brackets of each classifier, and the dynamic programming items S are based on1、S2、.......、S9The expression of (a) is:
n1、n2、n3sample sizes, X, of three classes, respectively3、X2、X1Sample sequences of three classes, respectively.
5. The dynamic programming-based classifier screening method according to claim 4, wherein the dynamic programming S is1、S2、.......、S9The satisfying conditions are as follows:
S1，ε(X3>X2>X1)；
S2，ε(X3>X′3>X2>X1) Or epsilon (X'3>X3>X2>X1)；
S3，ε(X3>X2>X′2>X1) Or ε (X)3>X′2>X2>X1)；
S4，ε(X3>X2>X′1>X1) Or ε (X)3>X2>X1>X′1)；
S5，ε(X3>X′3>X2>X′2>X1) Or epsilon (X'3>X3>X2>X′2>X1)ε(X3>X′3>X′2>X2>X1) Or epsilon (X'3>X3>X′2>X2>X1)；
S6，ε(X3>X2>X′3>X′2>X1) Or epsilon (X'3>X′2>X3>X2>X1)；
S7，ε(X3>X′3>X2>X1>X′1) Or epsilon (X'3>X3>X2>X1>X′1)ε(X3>X′3>X2>X′1>X1) Or epsilon (X'3>X3>X2>X′1>X1)；
S8，ε(X3>X2>X′2>X1>X′1) Or ε (X)3>X′2>X2>X1>X′1)ε(X3>X2>X′2>X′1>X1) Or ε (X)3>X′2>X2>X′1>X1)；
S9，ε(X3>X2>X1>X′2>X′1) Or ε (X)3>X′2>X′1>X2>X1)。
6. The dynamic programming-based classifier screening method according to claim 5, wherein in step S3, according to the dynamic programming algorithm, the obtained VUS variance expression is:
wherein
7. A classifier screening system based on dynamic programming, which applies the classifier screening method based on dynamic programming according to claim 1, and is characterized by comprising: the VUS variance detection device comprises a screening module, a VUS calculation module, a VUS variance calculation module and a classification module; the data of the screening module is transmitted to the VUS calculation module, the data of the VUS calculation module is transmitted to the VUS variance calculation module, and the data of the VUS variance calculation module is transmitted to the classification module;
the screening module carries out sorting processing on the sample data and converts the sample data into a vector matrix;
the VUS calculation module obtains VUS estimated values of the three types of classifiers by using a dynamic programming algorithm according to the vector matrix;
the VUS variance calculation module substitutes the VUS estimated value and the vector matrix into the VUS variance expression to calculate a VUS variance value;
and the classification module compares the VUS maximum estimation value according to the VUS variance value, and takes the classifier of the maximum estimation value as an optimal classifier.
8. The dynamic programming-based classifier filtering system of claim 7, wherein said filtering module sorts the sample data from small to large and converts it into a vector matrix.
9. The system of claim 8, wherein the VUS calculation module uses a dynamic programming algorithm to obtain VUS estimates for three types of classifiers ROC based on the vector matrix.
10. A computer device comprising a screening method and a screening system, wherein the device stores a computer program, and the computer program is executed by a processor to implement the dynamic programming-based classifier screening method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110797631.0A CN114330468A (en) | 2021-07-14 | 2021-07-14 | Classifier screening method and system based on dynamic programming and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110797631.0A CN114330468A (en) | 2021-07-14 | 2021-07-14 | Classifier screening method and system based on dynamic programming and computer equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114330468A true CN114330468A (en) | 2022-04-12 |
Family
ID=81044303
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110797631.0A Pending CN114330468A (en) | 2021-07-14 | 2021-07-14 | Classifier screening method and system based on dynamic programming and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114330468A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107171315A (en) * | 2017-05-18 | 2017-09-15 | 广东电网有限责任公司电力科学研究院 | A kind of transient stability evaluation in power system method based on RPTSVM |
| US20190114511A1 (en) * | 2017-10-16 | 2019-04-18 | Illumina, Inc. | Deep Learning-Based Techniques for Training Deep Convolutional Neural Networks |
| CN110866463A (en) * | 2019-10-29 | 2020-03-06 | 贵州大学 | Dynamic classifier selection method based on graph |
| CN111817803A (en) * | 2020-06-17 | 2020-10-23 | 广东工业大学 | Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium |
| CN111931829A (en) * | 2020-07-24 | 2020-11-13 | 广东工业大学 | Classifier screening method, classifier screening system, storage medium and computer equipment |
| CN112716498A (en) * | 2020-12-29 | 2021-04-30 | 北京理工大学 | Electrocardiosignal feature extraction method based on dynamic time warping and symbolic dynamics |
| CN112909939A (en) * | 2021-03-10 | 2021-06-04 | 山东科技大学 | Power system transient stability assessment method based on load flow characteristics before and after fault |
-
2021
- 2021-07-14 CN CN202110797631.0A patent/CN114330468A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107171315A (en) * | 2017-05-18 | 2017-09-15 | 广东电网有限责任公司电力科学研究院 | A kind of transient stability evaluation in power system method based on RPTSVM |
| US20190114511A1 (en) * | 2017-10-16 | 2019-04-18 | Illumina, Inc. | Deep Learning-Based Techniques for Training Deep Convolutional Neural Networks |
| CN110832596A (en) * | 2017-10-16 | 2020-02-21 | 因美纳有限公司 | Deep convolutional neural network training method based on deep learning |
| CN110870019A (en) * | 2017-10-16 | 2020-03-06 | 因美纳有限公司 | Semi-supervised learning for training deep convolutional neural network sets |
| CN110866463A (en) * | 2019-10-29 | 2020-03-06 | 贵州大学 | Dynamic classifier selection method based on graph |
| CN111817803A (en) * | 2020-06-17 | 2020-10-23 | 广东工业大学 | Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium |
| CN111931829A (en) * | 2020-07-24 | 2020-11-13 | 广东工业大学 | Classifier screening method, classifier screening system, storage medium and computer equipment |
| CN112716498A (en) * | 2020-12-29 | 2021-04-30 | 北京理工大学 | Electrocardiosignal feature extraction method based on dynamic time warping and symbolic dynamics |
| CN112909939A (en) * | 2021-03-10 | 2021-06-04 | 山东科技大学 | Power system transient stability assessment method based on load flow characteristics before and after fault |
Non-Patent Citations (4)
| Title |
|---|
| SHUN LIU.ET AL: "Fast and Unbiased Estimation of Volume Under Ordered Three-Class ROC Surface (VUS) Based on Dynamic Programming", 《IEEE》, vol. 7, pages 63972 - 63982, XP011726935, DOI: 10.1109/ACCESS.2019.2915642 * |
| SHUN LIU.ET AL: "Fast and Unbiased Estimation of Volume Under Ordered Three-Class ROC Surface (VUS) With Continuous or Discrete Measurements", 《IEEE》 * |
| 刘会河等: "基于SVM的降维方法在三类ROC分析中的应用", 《计算机与现代化》, no. 07 * |
| 孙旭: "接收机工作特性曲线研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 11 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190228527A1 (en) | System and method of classification of biological particles | |
| CN101814147A (en) | Method for realizing classification of scene images | |
| CN110363228B (en) | Noise label correction method | |
| JP2019207561A (en) | Information processing device, information processing method and program | |
| JP5905375B2 (en) | Misclassification detection apparatus, method, and program | |
| JP2023517155A (en) | Methods and devices for determining chromosomal aneuploidies and building classification models | |
| CN114330468A (en) | Classifier screening method and system based on dynamic programming and computer equipment | |
| CN112256881B (en) | User information classification method and device | |
| US20160026851A1 (en) | System and method for improved detection of objects of interest in image data by management of false positives | |
| CN113378682A (en) | Millimeter wave radar fall detection method and system based on improved clustering algorithm | |
| CN112347246A (en) | Self-adaptive document clustering method and system based on spectral decomposition | |
| Xu et al. | A fast algorithm for unbiased estimation of variance of AUC based on dynamic programming | |
| Charytanowicz et al. | Discrimination of wheat grain varieties using x-ray images | |
| JP2004348523A (en) | System for filtering document, and program | |
| JP2020067948A (en) | Analysis apparatus, analysis method, and analysis program | |
| CN111931829B (en) | Classifier screening method, system, storage medium and computer equipment | |
| US20140010433A1 (en) | System and method for improved detection of objects of interest in image data by management of false positives | |
| Soans et al. | Automatic assessment of the degree of TB-infection using images of ZN-stained sputum smear: New results | |
| Prasad et al. | Feature subset selection using ICA for classifying emphysema in HRCT images | |
| CN106933805B (en) | Method for identifying biological event trigger words in big data set | |
| Rohrlach et al. | BREADR: An R package for the Bayesian estimation of genetic relatedness from low-coverage genotype data | |
| JP2021124886A (en) | Information processing device, information processing method and information processing program | |
| Ozaki et al. | A method of two-stage clustering based on cluster validity measures | |
| Mohanraj et al. | Classification of rice grains based on quality using probabilistic neural network | |
| CN113742543B (en) | Data screening method and device, electronic equipment and storage medium |
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 |