CN107818342A - Based on the more categorizing systems and method for limiting fuzzy rule under big data environment - Google Patents

Based on the more categorizing systems and method for limiting fuzzy rule under big data environment Download PDF

Info

Publication number
CN107818342A
CN107818342A CN201711024378.5A CN201711024378A CN107818342A CN 107818342 A CN107818342 A CN 107818342A CN 201711024378 A CN201711024378 A CN 201711024378A CN 107818342 A CN107818342 A CN 107818342A
Authority
CN
China
Prior art keywords
fuzzy
rule
sample
classification
primitive
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
Application number
CN201711024378.5A
Other languages
Chinese (zh)
Inventor
熊安萍
蒋亚雄
祝清意
段杭彪
丁世旺
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Post and Telecommunications
Original Assignee
Chongqing University of Post and Telecommunications
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN201711024378.5A priority Critical patent/CN107818342A/en
Publication of CN107818342A publication Critical patent/CN107818342A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The present invention relates to, based on the more categorizing systems and method for limiting fuzzy rule, belong to big data classification field under a kind of big data environment.The system includes fuzzy introduction, indistinct logic computer, primary knowledge base and defuzzifier；The point for inputting domain U determinations is mapped as the fuzzy set on U by fuzzy introduction one by one；Primary knowledge base is made up of some fuzzy rule " if " rules, if some fuzzy rules are divided into Ganlei, the fuzzy rule of each class is made up of data rule and primitive rule；Fuzzy set on domain U is mapped by indistinct logic computer on the basis of fuzzy logic principles, using fuzzy rule with the fuzzy set on output domain V；The point that fuzzy set on V is mapped as determining on V one by one by fuzzy arrester.The present invention is greatly improved classification effectiveness, while by pair positive and negative regular supplement of determination rule, decreases because the error that fuzzy operation is brought, also ensure that classification accuracy.

Description

Based on the more categorizing systems and method for limiting fuzzy rule under big data environment
Technical field
The invention belongs to big data classification field, it is related under big data environment based on the more categorizing systems for limiting fuzzy rule And method.
Background technology
In recent years, with the arrival in big data epoch, there is an urgent need to work out instrument more easily and effectively to receiving by people The magnanimity information collected is fast and accurately classified, so as to therefrom extract it is suiting the requirements, succinct, refining, be appreciated that Knowledge.Data are the treasures accumulated in enterprise operation, are present in operating environment and analysis environments.Operating environment supports Basic business running, analysis environments support the Analysis of Policy Making demand of enterprise on the basis of integrating, refine operating environment data, Therefore data play the role of irreplaceable for enterprise operation support and Analysis of Policy Making.The big wealth of big data and what is be worth greatly obtain Take, also with data volume it is big the reason for, become more difficult to handle and find.Classification is a kind of common mass data Knowledge Extraction Method, the research on this respect at present also made significant headway.Existing sorting algorithm has many kinds, more commonly used Have KNN (K-nearest neighbor), Native Bayes, Nerual Net, SVM (Support Vectormachine), LLSF (linear least square fit) and conventional Fuzzy Rule Classification etc. is become recently Deng.
Fuzzy Classifier can utilize the fuzzy rule of simple language expression, overcome based on symbolic rule or hard rule point The defects of class device, therefore cause the concern of related researcher.The problem of for existing during these algorithm process large-scale datas, The research of many related fields has been carried out both at home and abroad.Shi Shaoying and Wang little Mo proposes a kind of the adaptive of regular number determination Fuzzy Classifier.The grader is made up of multiple for distinguishing a kind of indistinct logic computer of classification, each indistinct logic computer bag Containing two fuzzy logics " if -- if " rule, the GREV number of grader determines by the classification number of pattern to be sorted, grader Learning training is carried out using error back propagation learning algorithm, can be avoided because of " the dimension that input space dimension increases and occurs Disaster ".The disaggregated model has stronger learning ability, Classification and Identification ability, and anti-interference, anti-data contamination ability.Gong Quiet and Huang Xinyang etc. proposes a kind of based on the file classification method for improving fuzzy grammar increasable algorithm.The algorithm uses difference module Text fragments are transformed into bottom frame by type, form fuzzy grammar, are finally converted into Fuzzy Combined integration more general Transition form.The algorithm performance is more steady, and the size influence of data is smaller, has the relatively low model re -training time.Fan Thunder and thunder hero et al. propose the quick intuition fuzzy core match tracing method based on weak greedy strategy.For existing intuition mould Paste nuclear matching tracing algorithm causes learning time long limitation using the optimal basic function of greedy algorithm search, based on weak greedy Greedy strategy, propose a kind of random intuitionistic fuzzy nuclear matching tracing algorithm.This method has in the case where keeping accuracy of identification suitable Effect shortens a match tracing time, and computational efficiency significantly improves, and gained model have it is openness it is good, generalization ability is high etc. Advantage.Xu Mingliang and Wang Shitong proposes a kind of algorithm of structure TSK (Takagi-Sugeno-Kang) grader.The algorithm knot Matched moulds pastes (C+P) mean cluster (FCPM) algorithm and SP-V- SVMs (SVM) sorting algorithm, not only improves system Generalization Capability, also make system that there is the adaptive yojan function of fuzzy rule so that system is more compact.
The adaptive fuzzy classifier that regular number determines can solve the problem that carries out mass data classification " dimension calamity with fuzzy rule It is difficult " the problem of, and be also one of best Fuzzy Rule Classification method of effect.And the adaptive fuzzy point that regular number determines Appliances have the ability of Fuzzy Classifier Coping with Reality world flexibility information, are very suitable for handling present internet mass letter Breath.The adaptive fuzzy classifier that referred to hereinafter as regular number determines is determination rule-based algorithm.
The weak point for determining rule-based algorithm is that efficiency is less desirable, and the time of training is long, for big data mould The foundation of type needs to spend very big time overhead；Secondly, determine that application of the rule-based algorithm to fuzzy rule needs further to be extended.
The content of the invention
In view of this, it is an object of the invention to provide more classification based on restriction fuzzy rule under a kind of big data environment System and method, not only classifying rules application fuzzy rule, also the fuzzy mode of application is handled in the operation of sorting algorithm, Classification effectiveness is greatly improved, while by pair positive and negative regular supplement of determination rule, is decreased because fuzzy operation band The error come.
To reach above-mentioned purpose, the present invention provides following technical scheme：
Based on the more categorizing systems for limiting fuzzy rule under big data environment, the system includes fuzzy introduction, fuzzy pushed away Reason machine, primary knowledge base and defuzzifier；
The point for inputting domain U determinations is mapped as the fuzzy set on U by fuzzy introduction one by one；Primary knowledge base is by some Fuzzy rule " if yes " rule composition, if some fuzzy rules are divided into Ganlei, the fuzzy rule of each class is advised by data Then formed with primitive rule；Indistinct logic computer, will be fuzzy on domain U using fuzzy rule on the basis of fuzzy logic principles Gather and be mapped with the fuzzy set on output domain V；Fuzzy set on V is mapped as true on V by fuzzy arrester one by one Fixed point.
Further, described " if yes " rule is：
Rk:If x1ForAnd ..., and xnForThen y is Gk
In formula,ForOn fuzzy set,ForOn fuzzy set, x=(x1,...,xn)T∈ U1×...×UnFor linguistic variable, y ∈ V are linguistic variable；K=1,2 ..., M, M be fuzzy rule base in rule sum；
The opposition of two class fuzzy rule problems is referred from, each corresponding V there are two classes, three kinds of rule in the system Then, one kind is primitive rule, and another kind of is data rule, and data rule is divided into support rule and opposes rule, abstract expression again For：
R(k,i):If x1For f1'sAnd ..., and xnFor fn'sThen y is w Gk
In formula, R(k,i)The description of a rule-like is represented, if i=0, represents the opposition rule that this rule belongs in data rule Then, the opposing extent that this classification of k is belonged to sample x is described；If i=1, then it represents that the branch that this rule belongs in data rule Rule is held, describes the degree of support for belonging to this classification of k to sample x；If i ∈ r, r are the primitive rule set limited, represent This rule belongs to primitive rule；fiMiddle i value is 1,2 ... ..., n, represents that each attribute of sample is corresponding to rule and is subordinate to journey Degree, fiSpan be 0 to 1, oppose rule in, fi0 ith attribute for making difficulties rule is taken to belong to i-th of sample Property opposing extent uncertain, fiOther values make difficulties rule ith attribute to the ith attribute of sample Different degrees of opposition property value；Equally, in primitive rule, all fiValue all takes 1, it is desirable to sample and rule symbol completely Close；W represents that sample to be sorted belongs to and be not belonging to the weight of some classification, and span is 0 to 1, value in primitive rule It is fixed as 1；Data rule is that primitive rule is limited in few quantity, in the case where obtaining good accuracy rate, is reduced The quantity of rule；Primitive rule is as few as possible, and primitive rule determines that sample belongs to some classification.
It is based on the more sorting techniques for limiting fuzzy rule, this method step under big data environment：
Use fuzzy rule, i.e., in p → q reasoning process, substitute using fuzzy rule IF-THEN in classical reasoning P, q propositions, with it is fuzzy mend, it is fuzzy simultaneously, it is fuzzy hand over substitution, ∨, ∧ operator, then fuzzy rule is：
IF<FP1>THEN<FP2>
Assuming that FP1It is one and is defined on U=U1×…×UnOn fuzzy relation, FP2It is one and is defined on V=V1×…× VnOn fuzzy relation, vector x and y are the linguistic variable on U and V respectively；
The fuzzy rule represents Dienes-Rescher implications according to the different fuzzy operators of selection；Fuzzy IF-THEN Regular IF<FP1>THEN<FP2>, a fuzzy relation Q being construed in U × VD,
Fuzzy benefit isObscure and be μA∪B(x)=max [μA(x),μB(x)], fuzzy reasoning is subordinate to Spending function is It is fuzzy benefit, represents that sample to be sorted belongs to one Individual classification complementary operation operation, it is fuzzy to mend the degree for representing that a sample is not belonging to some classification；μA∪B(x) it is to obscure simultaneously, represents When one sample belongs to two classifications simultaneously, allow sample to belong to the bigger classification of membership function, allow sample fog-level more Gently；Fuzzy reasoning is represented, is mended fuzzyWith fuzzy and μA∪B(x) on the basis of classical reasoning p → q etc. Valency formFuzzy Representation form, for carrying out fuzzy reasoning.
The beneficial effects of the present invention are：The present invention is greatly improved classification effectiveness, while by pair determination rule just The supplement of anti-rule, decrease because the error that fuzzy operation is brought, also ensure that classification accuracy.
Brief description of the drawings
In order that the purpose of the present invention, technical scheme and beneficial effect are clearer, the present invention provides drawings described below and carried out Explanation：
Fig. 1 is fuzzy classification system；
Fig. 2, which is limited, determines the multi-categorizer of fuzzy rule；
Fig. 3 is neural network model.
Embodiment
Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are described in detail.
In order to solve the above-mentioned technical problem, it is main to divide based on the more categorizing systems for limiting fuzzy rule under big data environment For fuzzy reasoning and fuzzy system two parts.
1. fuzzy reasoning
Fuzzy reasoning is that fuzzy rule is used in classical reasoning.In p → q reasoning process, fuzzy rule IF- is used THEN substitutes p, q proposition, with it is fuzzy mend, it is fuzzy simultaneously, it is fuzzy hand over to substitute, ∨, ∧ operator, then can represent fuzzy rule：
IF<FP1>THEN<FP2> (1)
Assuming that FP1It is one and is defined on U=U1×…×UnOn fuzzy relation, FP2It is one and is defined on V=V1×…× VnOn fuzzy relation, x and y are the linguistic variable (vector) on U and V respectively.Fuzzy the pushing away of p → q reasoning equivalence formula (1) Reason is represented by different reasoning implications according to the different fuzzy operators of selection, it is conventional have Dienes-Rescher implications, Three kinds of Lukasiewicz implications, Zadeh implications.The present invention uses Dienes-Rescher implications.Specifically, fuzzy IF- THEN rules IF<FP1>THEN<FP2>, a fuzzy relation Q being construed in U × VD, it obscures benefit and is represented by formula (2), obscure and be represented by formula (3), fuzzy reasoning membership function such as formula (4).
μA∪B(x)=max [μA(x),μB(x)] (3)
Formula (2) and formula (3) they are the commonly used relation of sample resume fuzzy membership functions, and formula (2) represents fuzzy and mended, that is, One sample is not belonging to the degree of some classification, when formula (3) represents that a sample belongs to two classifications simultaneously, obscures and is exactly Allow sample to belong to the bigger classification of membership function, make the fog-level of sample lighter.Formula (4) be then classical reasoning p → q etc. Valency formFuzzy Representation form, for carrying out fuzzy reasoning.
2. fuzzy classification system
It is proposition of the present invention to limit fuzzy, uses for reference and determines fuzzy thought, by the part missed in its algorithm model rule By largely repeating accurately experience or learning to add to turn into a kind of improved algorithm of primitive rule in single grader.Tool The fuzzy logic system block diagram of the limited multi-categorizer for determining fuzzy rule is as shown in Figure 1.In Fig. 1, fuzzy introduction will be inputted and discussed The point that domain U is determined is mapped as the fuzzy set on U one by one；Primary knowledge base is made up of some fuzzy " if yes " rules, if If dry fuzzy rule can be divided into Ganlei；The fuzzy rule of each class is made up of data rule and primitive rule；Fuzzy reasoning Machine is on the basis of fuzzy logic principles, using these fuzzy rules by the fuzzy set on domain U with exporting the mould on domain V Paste set is mapped；The point that fuzzy set on V is mapped as determining on V one by one by fuzzy arrester.
Some fuzzy " if yes " in the basic fuzzy rule base of two parts and data fuzzy rule base in rudimentary knowledge Rule has following form：
Rk:If x1ForAnd ..., and xnForThen y is Gk (5)
In formula,ForOn fuzzy set,ForOn fuzzy set x=(x1,...,xn)T∈U1 ×...×UnFor linguistic variable, y ∈ V are also linguistic variable.If M is rule sum in fuzzy rule base, i.e. k=1 in (5) formula, 2,...,M.The opposition of two class fuzzy rule problems is referred from, each corresponding V has two classes, three kinds of rules in the present system, One kind is primitive rule, and one kind is data rule, and data rule can be divided into support rule and oppose two kinds of rule again, be abstracted table Up to such as formula (6) Suo Shi：
R(k,i):If x1For f1'sAnd ..., and xn is fn'sThen y is w Gk (6)
In formula (6), R(k,i)The description of a rule-like is represented, if i=0, represents that this rule belongs to anti-in data rule To rule, the opposing extent that this classification of k is belonged to sample x is described；If i=1, then it represents that this rule belongs in data rule Support rule, the degree of support that this classification of k is belonged to sample x is described；If i ∈ r, r are the primitive rule set limited, Represent that this rule belongs to primitive rule.fiMiddle i value is 1,2 ... ..., n, represents that each attribute of sample is corresponding to rule and is subordinate to Category degree, fiSpan be 0 to 1, oppose rule in, fi0 is taken to make difficulties regular ith attribute to the i-th of sample Uncertain, the f of the opposing extent of individual attributeiOther values make difficulties rule ith attribute to i-th of sample The different degrees of opposition property value of attribute；Equally, in primitive rule, all fiValue all takes 1, it is desirable to which sample is complete with rule Meet.(6) w in formula represents that sample to be sorted belongs to and be not belonging to the weight of some classification, and span is 0 to 1, substantially Value is fixed as 1 in rule.The two rule-like primitive rules that data rule is then similar in document are limited in few quantity, So in the case where obtaining good accuracy rate, the quantity of rule can be reduced.Primitive rule is as few as possible, is from experience Or the rule outside more this two rule-like come in study, primitive rule can determine that sample belongs to some classification.
3. embodiment
Limit the multi-categorizer of fuzzy rule
For with n feature x=(x1,...,xn)T∈U1×...×Un, the other classification problem of M species, the present invention sets Meter is realized as shown in Figure 2 based on the multi-classifier system for limiting fuzzy rule.
Wherein, sample space is sample set to be sorted, chooses a sample with n attribute every time, enters Classify in grader group, finally discussed according to the result of classification, if passing through base rule storehouse classification either data The result of rule base, discussion obtain final classification results.Wherein each C in grader groupi(i=1,2 ..., m) be all with Grader of the Fuzzy Rule-Based Modeling as core is limited in Fig. 1.In grader, fuzzy introduction uses formula (8) ambiguity function Fuzzy value is produced, membership function uses Gaussian function, and the reasoning of the fuzzy rule of formula (7) description uses product inference, obscured Arrester uses center average blur arrester.
In formula, h, j span is the value in { (h, j) | (n, 0), (f, 1) } set.When h, j value are (f, 1) When,Value beRepresent classification k of the sample under regular k subjection degree；When h, j value are (n, 0),Value beRepresent classification k of the sample under regular k opposing extent.Constraints It so can be obtained by the final formula (8) assessed sample and belong to corresponding classification：
Formula (8) represents degrees of membership of the sample x to class categories k, whereinWithIt is formula (7) result of calculation respectively, Represent that classification k corresponds to support and opposition degree of the grader fuzzy rule to sample x, that is, support sample to belong to this classification Degree and opposing extent.N represents that sample participates in the number of the attribute of classification, does normalized.Sample is fuzzy by limiting After the multi-categorizer of rule, there can be M degree of membership to M class, the wherein maximum final classification as sample be chosen, such as formula (9)：
In assorting process, the present invention is using the method for BP backpropagations come to hiThe support of { h | f, n } fuzzy rule and Oppose that coefficient is adjusted, by gradient descent method come the situation of change of adjustment parameter.Concrete model uses two layers of hidden layer Three-layer neural network realize, the coefficient that first layer regulation is supported, the coefficient that second layer regulation is opposed, comprehensive discussion whether Feedback regulation is needed, as shown in neural network model Fig. 3.
4. arthmetic statement
For in theory, the attribute of sample can with random length, regular attribute can also random length, but in order to grind The convenience studied carefully, it is assumed that the description attribute of all samples is equally long, is equal to the length of rules properties.It is same in real life The data of aspect, also substantially with same attribute description.Algorithm is broadly divided into two parts, and one is model training, one Sample classification.
It is all one two that training set TraningSet (X, T) and test set TestSet (X, T) are inputted in category of model algorithm Tuple, wherein X represent input sample collection, are a vector set, T is a result vector collection, is representing X corresponding sample just Really classification.Output is a four-tuple for limiting fuzzy classification model LFCM (H1, H2, P, C), and wherein H1, H2 is two arrays, Array length is the number of intrerneuron, represents the coefficient of training pattern, and P is the decimal within one 1, the standard of presentation class True rate, C are an integers, represent parameter modification iterations.
Input sample x (x in algorithm1,x2,...,xn)T, wherein x1,x2…,xnDescribe, can characterize for sample x feature One sample；Output result vector G (g1,g2,...,gm)T, wherein g1,g2…,gmFor final classification result, belong to corresponding class Value be 1, other is all 0.
Finally illustrate, preferred embodiment above is merely illustrative of the technical solution of the present invention and unrestricted, although logical Cross above preferred embodiment the present invention is described in detail, it is to be understood by those skilled in the art that can be Various changes are made to it in form and in details, without departing from claims of the present invention limited range.

Claims (3)

1. based on the more categorizing systems for limiting fuzzy rule under big data environment, it is characterised in that：The system includes fuzzy produce Device, indistinct logic computer, primary knowledge base and defuzzifier；
The point for inputting domain U determinations is mapped as the fuzzy set on U by fuzzy introduction one by one；Primary knowledge base is by some moulds Paste regular " if yes " rule composition, if some fuzzy rules are divided into Ganlei, the fuzzy rule of each class by data rule and Primitive rule forms；Indistinct logic computer is on the basis of fuzzy logic principles, using fuzzy rule by the fuzzy set on domain U It is mapped with the fuzzy set on output domain V；Fuzzy set on V is mapped as what is determined on V by fuzzy arrester one by one Point.
2. based on the more categorizing systems for limiting fuzzy rule under big data environment according to claim 1, it is characterised in that： " if yes " rule is：
Rk:If x1ForAnd ..., and xnForThen y is Gk
In formula,ForOn fuzzy set,ForOn fuzzy set, x=(x1,...,xn)T∈U1 ×...×UnFor linguistic variable, y ∈ V are linguistic variable；K=1,2 ..., M, M be fuzzy rule base in rule sum；
The opposition of two class fuzzy rule problems is referred from, each corresponding V has two classes, three kinds of rules in the system, and one Class is primitive rule, and another kind of is data rule, and data rule is divided into support rule and opposes rule again, and abstract expression is：
R(k,i):If x1For f1'sAnd ..., and xnFor fn'sThen y is w Gk
In formula, R(k,i)The description of a rule-like is represented, if i=0, represents the opposition rule that this rule belongs in data rule, Description belongs to the opposing extent of this classification of k to sample x；If i=1, then it represents that the support that this rule belongs in data rule Rule, the degree of support that this classification of k is belonged to sample x is described；If i ∈ r, r are the primitive rule set limited, this is represented Rule belongs to primitive rule；fiMiddle i value is 1,2 ... ..., n, represents that each attribute of sample is corresponding to rule and is subordinate to journey Degree, fiSpan be 0 to 1, oppose rule in, fi0 ith attribute for making difficulties rule is taken to belong to i-th of sample Property opposing extent uncertain, fiOther values make difficulties rule ith attribute to the ith attribute of sample Different degrees of opposition property value；Equally, in primitive rule, all fiValue all takes 1, it is desirable to sample and rule symbol completely Close；W represents that sample to be sorted belongs to and be not belonging to the weight of some classification, and span is 0 to 1, value in primitive rule It is fixed as 1；Data rule is that primitive rule is limited in few quantity, in the case where obtaining good accuracy rate, is reduced The quantity of rule；Primitive rule is as few as possible, and primitive rule determines that sample belongs to some classification.
3. based on the more sorting techniques for limiting fuzzy rule under big data environment, it is characterised in that：This method step is：
Fuzzy rule is used in classical reasoning, i.e., in p → q reasoning process, substitutes p, q using fuzzy rule IF-THEN Proposition, with it is fuzzy mend, it is fuzzy simultaneously, it is fuzzy hand over substitution-, ∨, ∧ operator, then fuzzy rule be：
IF<FP1>THEN<FP2>
Assuming that FP1It is one and is defined on U=U1×…×UnOn fuzzy relation, FP2It is one and is defined on V=V1×…×VnOn Fuzzy relation, vector x and y are the linguistic variable on U and V respectively；
The fuzzy rule represents Dienes-Rescher implications according to the different fuzzy operators of selection；Fuzzy IF-THEN rules IF<FP1>THEN<FP2>, a fuzzy relation Q being construed in U × VD,
Fuzzy benefit isObscure and be μA∪B(x)=max [μA(x),μB(x)], fuzzy reasoning degree of membership letter Number is It is fuzzy benefit, represents that sample to be sorted belongs to a class Other complementary operation operation, it is fuzzy to mend the degree for representing that a sample is not belonging to some classification；μA∪B(x) it is to obscure simultaneously, represents one When sample belongs to two classifications simultaneously, allow sample to belong to the bigger classification of membership function, make the fog-level of sample lighter；Fuzzy reasoning is represented, is mended fuzzyWith fuzzy and μA∪B(x) classical reasoning p → q equivalence on the basis of FormFuzzy Representation form, for carrying out fuzzy reasoning.
CN201711024378.5A 2017-10-27 2017-10-27 Based on the more categorizing systems and method for limiting fuzzy rule under big data environment Pending CN107818342A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711024378.5A CN107818342A (en) 2017-10-27 2017-10-27 Based on the more categorizing systems and method for limiting fuzzy rule under big data environment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711024378.5A CN107818342A (en) 2017-10-27 2017-10-27 Based on the more categorizing systems and method for limiting fuzzy rule under big data environment

Publications (1)

Publication Number Publication Date
CN107818342A true CN107818342A (en) 2018-03-20

Family

ID=61604176

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711024378.5A Pending CN107818342A (en) 2017-10-27 2017-10-27 Based on the more categorizing systems and method for limiting fuzzy rule under big data environment

Country Status (1)

Country Link
CN (1) CN107818342A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108564136A (en) * 2018-05-02 2018-09-21 北京航空航天大学 A kind of airspace operation Situation Assessment sorting technique based on fuzzy reasoning
CN110349188A (en) * 2019-07-18 2019-10-18 深圳大学 Multi-object tracking method, device and storage medium based on TSK fuzzy model

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108564136A (en) * 2018-05-02 2018-09-21 北京航空航天大学 A kind of airspace operation Situation Assessment sorting technique based on fuzzy reasoning
CN110349188A (en) * 2019-07-18 2019-10-18 深圳大学 Multi-object tracking method, device and storage medium based on TSK fuzzy model
CN110349188B (en) * 2019-07-18 2023-10-27 深圳大学 Multi-target tracking method, device and storage medium based on TSK fuzzy model

Similar Documents

Publication Publication Date Title
CN106779087B (en) A kind of general-purpose machinery learning data analysis platform
CN106845717B (en) Energy efficiency evaluation method based on multi-model fusion strategy
CN101604322B (en) Decision level text automatic classified fusion method
CN108319987A (en) A kind of filtering based on support vector machines-packaged type combined flow feature selection approach
CN108363810A (en) Text classification method and device
Mansoori et al. A weighting function for improving fuzzy classification systems performance
CN109871809A (en) A kind of machine learning process intelligence assemble method based on semantic net
Zhu Topic recommendation system using personalized fuzzy logic interest set
Wang et al. Design of the Sports Training Decision Support System Based on the Improved Association Rule, the Apriori Algorithm.
Chu et al. Co-training based on semi-supervised ensemble classification approach for multi-label data stream
CN112529638A (en) Service demand dynamic prediction method and system based on user classification and deep learning
CN107818342A (en) Based on the more categorizing systems and method for limiting fuzzy rule under big data environment
CN108154380A (en) The method for carrying out the online real-time recommendation of commodity to user based on extensive score data
CN108268461A (en) A kind of document sorting apparatus based on hybrid classifer
Xie et al. Multi-classification method for determining coastal water quality based on SVM with grid search and KNN
Yang et al. Microblog sentiment analysis algorithm research and implementation based on classification
Jamal et al. Sentimental analysis based on hybrid approach of latent dirichlet allocation and machine learning for large-scale of imbalanced twitter data
CN115563284B (en) Deep multi-instance weak supervision text classification method based on semantics
CN110196911B (en) Automatic classification management system for civil data
Peng et al. Hybrid feature selection model based on machine learning and knowledge graph
CN107644074A (en) A kind of method of the readable analysis of the Chinese teaching material based on convolutional neural networks
Pang et al. Application of the algorithm based on the PSO and improved SVDD for the personal credit rating
CN103207893A (en) Classification method of two types of texts on basis of vector group mapping
Wang et al. Content-based weibo user interest recognition
Park et al. Advanced technologies in blockchain, machine learning, and Big Data

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20180320