CN106599936A - Characteristic selection method based on binary ant colony algorithm and system thereof - Google Patents
Characteristic selection method based on binary ant colony algorithm and system thereof Download PDFInfo
- Publication number
- CN106599936A CN106599936A CN201611246351.6A CN201611246351A CN106599936A CN 106599936 A CN106599936 A CN 106599936A CN 201611246351 A CN201611246351 A CN 201611246351A CN 106599936 A CN106599936 A CN 106599936A
- Authority
- CN
- China
- Prior art keywords
- binary
- ant colony
- genetic
- algorithm
- individual
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The invention discloses a characteristic selection method based on a binary ant colony algorithm and a system thereof. The method comprises the following steps of acquiring a training sample set which needs to carry out characteristic selection; carrying out characteristic extraction on the training sample set and acquiring a sample characteristic set; using a binary genetic algorithm to classify the sample characteristic set, seeking maximum genetic fitness and acquiring an optimal solution, wherein the genetic fitness is a degree which makes a result of the binary genetic algorithm approach a target result; according to the optimal solution, setting visibility information of the binary ant colony algorithm and initializing an ant colony of the binary ant colony algorithm; and using the binary ant colony algorithm including the visibility information to carry out characteristic selection on the sample characteristic set. By using the method and the system of the invention, the binary genetic algorithm is used to provide appropriate visibility information for the binary ant colony algorithm so that a convergence speed and robustness of the binary ant colony algorithm can be increased and characteristic selection efficiency and performance are further increased too.
Description
Technical field
The present invention relates to mode identification technology, more particularly to a kind of feature selection based on binary ant colony algorithm
Method and system.
Background technology
Feature selection also known as feature subset selection, refer to and some most effective features are selected from primitive character to reduce number
According to the process of collection dimension, it is that the data of key in an important means, and pattern recognition for improve learning algorithm performance are located in advance
Reason step.
Feature selection is exactly the process of a search optimal feature subset in essence, i.e., discrete set is carried out excellent
The process of change.Ant group algorithm is particularly suitable for seeking optimal solution in multiple solutions of discrete space.Therefore can be by ant group algorithm
It is applied in feature selection.
Binary ant colony algorithm (BACO, binary ant colony optimization) is a kind of with the calculation of original ant colony
Swarm Intelligence Algorithm based on method.Continuous optimization problems are used to solve for often.But binary ant colony algorithm solution at the beginning
It is randomly generated, no utilizable visibility information causes the convergence rate of binary ant colony algorithm slower, feature choosing
The efficiency selected is restricted.
The content of the invention
It is an object of the invention to provide a kind of feature selection approach and system based on binary ant colony algorithm, is utilizing two
System ant group algorithm provides suitable visibility information before carrying out feature selection, improve the convergence rate of binary ant colony algorithm,
Improve the efficiency of feature selection.
For achieving the above object, the invention provides following scheme：
A kind of feature selection approach based on binary ant colony algorithm, including：
Acquisition needs to carry out the training sample set of feature selection；
Feature extraction is carried out to the training sample set, sample characteristics collection is obtained；
Classified and sought maximum genetic adaptation degree using binary strings genetic algorithm to the sample characteristics collection, obtained most
Excellent solution；The genetic adaptation degree is the degree for making the result of the binary strings genetic algorithm be close to objective result；
The visibility information of binary ant colony algorithm, the ant to the binary ant colony algorithm are arranged according to the optimal solution
Group is initialized；
Feature selection is carried out to the sample characteristics collection using the binary ant colony algorithm comprising the visibility information.
Optionally, the utilization binary strings genetic algorithm is classified to the sample characteristics collection and is sought maximum heredity and fitted
Response, obtains optimal solution, specifically includes：
Initialized for the parameter of binary strings genetic algorithm described in the sample training set pair, generated genetic groups；
The individual length of wherein described heredity is set to the feature quantity included by the sample characteristics collection；The heredity is individual for composition
The unit of genetic groups；
The genetic groups are decoded, fisrt feature subset is obtained；
The training sample set is classified using the fisrt feature subset, obtain the first classification results；
Calculate the first classification accuracy of first classification results；
The individual genetic adaptation degree of corresponding heredity is solved according to first classification accuracy；
Through multiple genetic manipulation, the maximum of the genetic adaptation degree is solved, maximum genetic adaptation degree is obtained；It is described most
The heredity that excellent solution is the maximum genetic adaptation degree is individual.
Optionally, the visibility information that binary ant colony algorithm is set according to the optimal solution, to the binary system
The ant colony of ant group algorithm is initialized, and is specifically included：
Obtain the binary numeral of the optimal solution；
In the binary numeral 1 is replaced with into the first preset number, 0 in the binary numeral is replaced with into
Two preset numbers, obtain the visibility information of binary ant colony algorithm；
The parameter of the binary ant colony algorithm is initialized according to the visibility information, generate ant colony, it is described
Ant colony is the individual set for constituting of multiple Formica fuscas.
Optionally, it is described the sample characteristics collection to be carried out using the binary ant colony algorithm comprising the visibility information
Feature selection, specifically includes：
Candidate solution is searched in the ant colony, candidate solution set is obtained；
The candidate solution set is decoded, second feature subset is obtained；
The training sample set is classified using the second feature subset, obtain the second classification results；
Calculate the second classification accuracy of second classification results；
The individual Formica fusca fitness of corresponding Formica fusca is solved according to second classification accuracy；
Determine that optimum Formica fusca is individual through multiple computing；It is the maximum ant of the Formica fusca fitness that the optimum Formica fusca is individual
Ant is individual；
Decode the individual corresponding optimal feature subset of the optimum Formica fusca.
The invention also discloses a kind of feature selection system based on binary ant colony algorithm, including：
Sample acquisition module, needs to carry out the training sample set of feature selection for obtaining；
Feature extraction module, for carrying out feature extraction to the training sample set, obtains sample characteristics collection；
Hereditary optimizing module, for being classified and being sought maximum to the sample characteristics collection using binary strings genetic algorithm
Genetic adaptation degree, obtains optimal solution；The genetic adaptation degree is that the result for making the binary strings genetic algorithm is close to objective result
Degree；
Visibility generation module, for the visibility information of binary ant colony algorithm is arranged according to the optimal solution, to institute
The ant colony for stating binary ant colony algorithm is initialized；
Ant colony selecting module, for utilizing the binary ant colony algorithm comprising the visibility information to the sample characteristics
Collection carries out feature selection.
Optionally, the hereditary optimizing module, specifically includes：
Parameter initial cell, for carrying out initially for the parameter of binary strings genetic algorithm described in the sample training set pair
Change, generate genetic groups；The individual length of wherein described heredity is set to the feature quantity included by the sample characteristics collection；Institute
State the hereditary individual unit to constitute genetic groups；
Population decoding unit, for decoding to the genetic groups, obtains fisrt feature subset；
First taxon, for classifying to the training sample set using the fisrt feature subset, obtains
One classification results；
First accuracy rate computing unit, for calculating the first classification accuracy of first classification results；
Genetic adaptation degree computing unit, for solving the individual heredity of corresponding heredity according to first classification accuracy
Fitness；
Optimal solution computing unit, for passing through multiple genetic manipulation, solves the maximum of the genetic adaptation degree, obtains most
Big genetic adaptation degree；The heredity that the optimal solution is the maximum genetic adaptation degree is individual.
Optionally, the visibility generation module, specifically includes：
Binary number acquiring unit, for obtaining the binary numeral of the optimal solution；
Visibility computing unit, for 1 in the binary numeral is replaced with the first preset number, described two is entered
In numerical value processed 0 replaces with the second preset number, obtains the visibility information of binary ant colony algorithm；
Ant colony signal generating unit, for being carried out initially to the parameter of the binary ant colony algorithm according to the visibility information
Change, generate ant colony, the ant colony is the individual set for constituting of multiple Formica fuscas.
Optionally, the ant colony selecting module, specifically includes：
Candidate solution search unit, for searching for candidate solution in the ant colony, obtains candidate solution set；
Candidate solution decoding unit, for decoding to the candidate solution set, obtains second feature subset；
Second taxon, for classifying to the training sample set using the second feature subset, obtains
Two classification results；
Second accuracy rate computing unit, calculates the second classification accuracy of second classification results；
Formica fusca fitness computing unit, for solving the individual Formica fusca of corresponding Formica fusca according to second classification accuracy
Fitness；
For passing through multiple computing, optimum individual determining unit, determines that optimum Formica fusca is individual；The optimum Formica fusca individuality is
The maximum Formica fusca of the Formica fusca fitness is individual；
Optimal characteristics determining unit, for decoding the individual corresponding optimal feature subset of the optimum Formica fusca.
According to the specific embodiment that the present invention is provided, the invention discloses following technique effect：The present invention utilizes binary system
Genetic algorithm provides suitable visibility information for binary ant colony algorithm such that it is able to make the convergence speed of binary ant colony algorithm
Degree and robustness are improved, and further increase the efficiency and performance of feature selection.
Description of the drawings
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to institute in embodiment
The accompanying drawing that needs are used is briefly described, it should be apparent that, drawings in the following description are only some enforcements of the present invention
Example, for those of ordinary skill in the art, without having to pay creative labor, can be with according to these accompanying drawings
Obtain other accompanying drawings.
Fig. 1 is method flow diagram of the present invention based on the feature selection approach embodiment of binary ant colony algorithm；
Fig. 2 is that utilization genetic algorithm of the present invention based on the feature selection approach embodiment of binary ant colony algorithm is obtained most
The method flow diagram of excellent solution；
Fig. 3 is generation binary ant colony algorithm of the present invention based on the feature selection approach embodiment of binary ant colony algorithm
Visibility information method flow diagram；
Fig. 4 is utilization binary ant colony algorithm of the present invention based on the feature selection approach embodiment of binary ant colony algorithm
Carry out the method flow diagram of feature selection；
Fig. 5 is system construction drawing of the present invention based on the feature selection system embodiment of binary ant colony algorithm.
Specific embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation is described, it is clear that described embodiment is only a part of embodiment of the invention, rather than the embodiment of whole.It is based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under the premise of creative work is not made
Embodiment, belongs to the scope of protection of the invention.
It is understandable to enable the above objects, features and advantages of the present invention to become apparent from, it is below in conjunction with the accompanying drawings and concrete real
The present invention is further detailed explanation to apply mode.
Fig. 1 is method flow diagram of the present invention based on the feature selection approach embodiment of binary ant colony algorithm.
Referring to Fig. 1, the feature selection approach based on binary ant colony algorithm is somebody's turn to do, including：
Step 101：Acquisition needs to carry out the training sample set of feature selection；
Step 102：Feature extraction is carried out to the training sample set, sample characteristics collection is obtained；
Step 103：The sample characteristics collection is classified using binary strings genetic algorithm and sought maximum genetic adaptation
Degree, obtains optimal solution；The genetic adaptation degree is the degree for making the result of the binary strings genetic algorithm be close to objective result；
Step 104：The visibility information of binary ant colony algorithm is arranged according to the optimal solution, to the Binary Ant Colony
The ant colony of algorithm is initialized；
Step 105：Spy is carried out to the sample characteristics collection using the binary ant colony algorithm comprising the visibility information
Levy selection.
The present invention provides suitable visibility information for binary ant colony algorithm using binary strings genetic algorithm such that it is able to
Improve the convergence rate and robustness of binary ant colony algorithm, further increase the efficiency and performance of feature selection.
Fig. 2 is that utilization genetic algorithm of the present invention based on the feature selection approach embodiment of binary ant colony algorithm is obtained most
The method flow diagram of excellent solution.
Referring to Fig. 2, the utilization binary strings genetic algorithm is classified to the sample characteristics collection and is sought maximum heredity
Fitness, obtains optimal solution, specifically includes：
Step 201：Initialized for the parameter of binary strings genetic algorithm described in the sample training set pair, generate and lose
Propagate group；The individual length of wherein described heredity is set to the feature quantity included by the sample characteristics collection；The heredity
Body is the unit for constituting genetic groups；
In initialization procedure, initiation parameter also includes population scale, selection rate, intersection except hereditary individual length
Rate, aberration rate, maximum iteration time.Population scale is the individual quantity of heredity in genetic groups；
Step 202：The genetic groups are decoded, fisrt feature subset is obtained；
It is individual for hereditary, if the corresponding binary number of some feature is 0, then it represents that this feature does not have selected,
If the corresponding binary number of some feature is 1, then it represents that this feature is selected.Such as one heredity it is individual for 0,1,0,
0,1,0,0,0 }, represent that the individuality there are eight features, from left to right, the 2nd feature and the 5th feature are chosen, and other 6 special
Levy and be not chosen.
Step 203：The training sample set is classified using the fisrt feature subset, obtain the first classification knot
Really；
Step 204：Calculate the first classification accuracy of first classification results；
The computing formula of first classification accuracy Accuracy (i) is：
Wherein i represents hereditary individual；TPFor the quantity of correct positive example, if test sample is positive example and the test specimens
Originally it is positive example to be determined, then the test sample is considered correct positive example；TNFor the quantity of correct negative example, if test sample is
Negative example and the test sample to be determined be negative example, then the test sample is considered correctly to bear example；FPFor wrong positive example
Quantity, if test sample be negative example and the test sample to be determined be positive example, the test sample is considered mistake
Positive example；FNFor the quantity of false negative example, if test sample be positive example and the test sample to be determined be negative example, should
Test sample is considered false negative example.
Step 205：The individual genetic adaptation degree of corresponding heredity is solved according to first classification accuracy；
The computing formula of hereditary individual fitness F (i) is：
Wherein n represents total Characteristic Number, and n (i) represents the number of the feature that hereditary individuality i is selected, and λ is weighting parameters,
λ ∈ [0.9,0.98] under normal circumstances.
Step 206：Through multiple genetic manipulation, the maximum of the genetic adaptation degree is solved, maximum genetic adaptation is obtained
Degree；The heredity that the optimal solution is the maximum genetic adaptation degree is individual.
Fig. 3 is generation binary ant colony algorithm of the present invention based on the feature selection approach embodiment of binary ant colony algorithm
Visibility information method flow diagram.
Referring to Fig. 3, the visibility information that binary ant colony algorithm is arranged according to the optimal solution, to the binary system
The ant colony of ant group algorithm is initialized, and is specifically included：
Step 301：Obtain the binary numeral of the optimal solution；
Step 302：In the binary numeral 1 is replaced with into the first preset number, by 0 in the binary numeral
The second preset number is replaced with, the visibility information of binary ant colony algorithm is obtained；
Specially：If the value of a certain position is 0 in the binary system of the optimal solution that binary strings genetic algorithm is obtained, value is 0
Visibility value be set to first preset number, be worth and be set to the second preset number, the first present count for 1 visibility value
Word is more than the second preset number.Or, if the value of a certain position is 1 in the binary system of optimal solution, value is set for 1 visibility value
First preset number being set to, being worth and the second preset number is set to for 0 visibility value, the first preset number is pre- more than second
If digital.
For example, sample characteristics collection includes 8 features, and the binary numeral of the optimal solution that binary strings genetic algorithm is obtained is
01010001.Another ηnof(1)=0.8, ηnof(0)=0.2, wherein nof=1,2 ..., NumoF；NumoF represents Characteristic Number.Then
Visibility value is set to：
η1(0)=η2(0)=...=η8(0)={ 0.8,0.2,0.8,0.2,0.8,0.8,0.8,0.2 }
η1(1)=η2(1)=...=η8(1)={ 0.2,0.8,0.2,0.8,0.2,0.2,0.2,0.8 }.
Step 303：The parameter of the binary ant colony algorithm is initialized according to the visibility information, generate ant
Group, the ant colony are the individual set for constituting of multiple Formica fuscas.
The initialization procedure also includes initialization binary ant colony algorithm visibility weight parameter, pheromone concentration weight ginseng
Number, ant colony scale.Wherein ant colony scale is equal with the population scale of binary strings genetic algorithm.
Fig. 4 is utilization binary ant colony algorithm of the present invention based on the feature selection approach embodiment of binary ant colony algorithm
Carry out the method flow diagram of feature selection.
It is referring to Fig. 4, described the sample characteristics collection to be entered using the binary ant colony algorithm comprising the visibility information
Row feature selection, specifically includes：
Step 401：Candidate solution is searched in the ant colony, candidate solution set is obtained；
There are two kinds of candidate states in each feature, that is, select and do not select, therefore can be by candidate's shape of each feature
State is mapped with binary code.In mapping process, each binary digit one feature of correspondence, the binary digit for obtaining
Length be equal to number of features, and with 1 represent corresponding feature be selected for classification, with 0 represent corresponding feature not by
Select.Then binary ant colony algorithm mark candidate solution is 0 probabilityFor：
Wherein, k represents Formica fusca individuality；Represent that Formica fusca individuality k moves to the probability of state 0, i.e. Formica fusca from feature l
The probability that feature l in individual k is not selected；Here l is variable, and span is [1, L], and wherein L represents sample characteristics collection
Length.Parameter alpha is pheromone concentration weight parameter, and α >=0；β is visibility weight parameter, and β >=0；τl(0) represent road
Footpath (l, 0) on pheromone concentration；τl(1) represent path (l, 1) on pheromone concentration；ηl(0) represent path (l, 0) on
Visibility, ηl(1) represent path (l, 1) on visibility.
Accordingly, in binary ant colony algorithm, labelling candidate solution is 1 probabilityFor
The probit that each feature in candidate solution is marked as 0 or 1 is calculated using above-mentioned formula, then using roulette
Mode selects feature to be marked as 0 or 1 state.Whether Formica fusca individuality is often selected once, selected with regard to one feature of labelling, and one
Until all of feature be chosen and labelling on.
Step 402：The candidate solution set is decoded, second feature subset is obtained；
Citing below is illustrated.If certain candidate solution has eight feature { f1,f2,f3,f4,f5,f6,f7,f8, which two
Ary codes are 01010001, and the process of the decoding is that 1 feature is chosen constitutive characteristic for the binary code intermediate value of candidate solution
Collection, is worth and is rejected for 0 feature.So, the binary code 01010001 represents that the 2nd, the 4th and the 8th feature is selected
Constitutive characteristic subset, i.e. feature f2、f4And f8Selected constitutive characteristic subset { f2,f4,f8}。
Step 403：The training sample set is classified using the second feature subset, obtain the second classification knot
Really；
Step 404：Calculate the second classification accuracy of second classification results；
The computational methods of second classification accuracy are identical with the computational methods of the first classification accuracy.
Step 405：The individual Formica fusca fitness of corresponding Formica fusca is solved according to second classification accuracy；
The computational methods of the Formica fusca fitness are identical with the computational methods of hereditary individual fitness.
Step 406：Determine that optimum Formica fusca is individual through multiple computing；It is the Formica fusca fitness that the optimum Formica fusca is individual
Maximum Formica fusca is individual；
In this process, current optimal solution is selected according to Formica fusca fitness first, judges current optimal solution whether better than upper
The optimal solution that a generation is obtained, if current optimal solution retains current optimal solution better than the optimal solution that previous generation is obtained, if worked as
Front optimal solution is not better than the optimal solution that previous generation is obtained, then retain the optimal solution of previous generation.
Secondly, update the pheromone concentration ε in ant colony searching routel(z).Pheromone concentration εlZ () more new formula is：
Wherein, t represents iterationses；εlZ in the t+1 time iteration of () (t+1) expression, feature l is marked as the letter of state z
The plain concentration of breath；Here z has two kinds of values, and respectively 0 or 1.ρ represents pheromone rate of volatilization, and ρ ∈ [0,1],Represent and increase
Plus pheromone, and Represent the Formica fusca fitness of optimal solution.
0 or 1 state is labeled as according to feature l in optimal solution, using above- mentioned information element concentration more new formula to being labeled as
The pheromone of the feature of state 0 is updated, and is otherwise updated the pheromone of the feature of the state that is labeled as 1.
Finally, judge whether iterationses meet default iterationses, the execution step 407 if meeting, if discontented
Foot, then return and re-execute step 401.
Step 407：Decode the individual corresponding optimal feature subset of the optimum Formica fusca.
Fig. 5 is system construction drawing of the present invention based on the feature selection system embodiment of binary ant colony algorithm.
Referring to Fig. 5, the feature selection system based on binary ant colony algorithm is somebody's turn to do, including：
Sample acquisition module 501, needs to carry out the training sample set of feature selection for obtaining；
Feature extraction module 502, for carrying out feature extraction to the training sample set, obtains sample characteristics collection；
Hereditary optimizing module 503, for the sample characteristics collection being classified and being sought using binary strings genetic algorithm
Maximum genetic adaptation degree, obtains optimal solution；The genetic adaptation degree is that the result for making the binary strings genetic algorithm is close to target
As a result degree；
Visibility generation module 504, it is for the visibility information of binary ant colony algorithm is arranged according to the optimal solution, right
The ant colony of the binary ant colony algorithm is initialized；
Ant colony selecting module 505, for utilizing the binary ant colony algorithm comprising the visibility information to the sample
Feature set carries out feature selection.
Optionally, the hereditary optimizing module 503, specifically includes：
Parameter initial cell, for carrying out initially for the parameter of binary strings genetic algorithm described in the sample training set pair
Change, generate genetic groups；The individual length of wherein described heredity is set to the feature quantity included by the sample characteristics collection；Institute
State the hereditary individual unit to constitute genetic groups；
Population decoding unit, for decoding to the genetic groups, obtains fisrt feature subset；
First taxon, for classifying to the training sample set using the fisrt feature subset, obtains
One classification results；
First accuracy rate computing unit, for calculating the first classification accuracy of first classification results；
Genetic adaptation degree computing unit, for solving the individual heredity of corresponding heredity according to first classification accuracy
Fitness；
Optimal solution computing unit, for passing through multiple genetic manipulation, solves the maximum of the genetic adaptation degree, obtains most
Big genetic adaptation degree；The heredity that the optimal solution is the maximum genetic adaptation degree is individual.
Optionally, the visibility generation module 504, specifically includes：
Binary number acquiring unit, for obtaining the binary numeral of the optimal solution；
Visibility computing unit, for 1 in the binary numeral is replaced with the first preset number, described two is entered
In numerical value processed 0 replaces with the second preset number, obtains the visibility information of binary ant colony algorithm；
Ant colony signal generating unit, for being carried out initially to the parameter of the binary ant colony algorithm according to the visibility information
Change, generate ant colony, the ant colony is the individual set for constituting of multiple Formica fuscas.
Optionally, the ant colony selecting module 505, specifically includes：
Candidate solution search unit, for searching for candidate solution in the ant colony, obtains candidate solution set；
Candidate solution decoding unit, for decoding to the candidate solution set, obtains second feature subset；
Second taxon, for classifying to the training sample set using the second feature subset, obtains
Two classification results；
Second accuracy rate computing unit, calculates the second classification accuracy of second classification results；
Formica fusca fitness computing unit, for solving the individual Formica fusca of corresponding Formica fusca according to second classification accuracy
Fitness；
For passing through multiple computing, optimum individual determining unit, determines that optimum Formica fusca is individual；The optimum Formica fusca individuality is
The maximum Formica fusca of the Formica fusca fitness is individual.
Specific case used herein is set forth to the principle and embodiment of the present invention, and above example is said
It is bright to be only intended to help and understand the method for the present invention and its core concept；Simultaneously for one of ordinary skill in the art, foundation
The thought of the present invention, will change in specific embodiments and applications.In sum, this specification content is not
It is interpreted as limitation of the present invention.
Claims (8)
1. a kind of feature selection approach based on binary ant colony algorithm, it is characterised in that include：
Acquisition needs to carry out the training sample set of feature selection；
Feature extraction is carried out to the training sample set, sample characteristics collection is obtained；
Classified and sought maximum genetic adaptation degree using binary strings genetic algorithm to the sample characteristics collection, obtained optimum
Solution；The genetic adaptation degree is the degree for making the result of the binary strings genetic algorithm be close to objective result；
The visibility information of binary ant colony algorithm is arranged according to the optimal solution, the ant colony of the binary ant colony algorithm is entered
Row initialization；
Feature selection is carried out to the sample characteristics collection using the binary ant colony algorithm comprising the visibility information.
2. a kind of feature selection approach based on binary ant colony algorithm according to claim 1, it is characterised in that described
Classified and sought maximum genetic adaptation degree using binary strings genetic algorithm to the sample characteristics collection, obtained optimal solution, had
Body includes：
Initialized for the parameter of binary strings genetic algorithm described in the sample training set pair, generated genetic groups；Wherein
The individual length of the heredity is set to the feature quantity included by the sample characteristics collection；The heredity is individual hereditary to constitute
The unit of population；
The genetic groups are decoded, fisrt feature subset is obtained；
The training sample set is classified using the fisrt feature subset, obtain the first classification results；
Calculate the first classification accuracy of first classification results；
The individual genetic adaptation degree of corresponding heredity is solved according to first classification accuracy；
Through multiple genetic manipulation, the maximum of the genetic adaptation degree is solved, maximum genetic adaptation degree is obtained；The optimal solution
The heredity of as described maximum genetic adaptation degree is individual.
3. a kind of feature selection approach based on binary ant colony algorithm according to claim 1, it is characterised in that described
The visibility information of binary ant colony algorithm is arranged according to the optimal solution, the ant colony of the binary ant colony algorithm is carried out just
Beginningization, specifically includes：
Obtain the binary numeral of the optimal solution；
In the binary numeral 1 is replaced with into the first preset number, 0 in the binary numeral second is replaced with into pre-
If digital, the visibility information of binary ant colony algorithm is obtained；
The parameter of the binary ant colony algorithm is initialized according to the visibility information, generate ant colony, the ant colony
For the individual set for constituting of multiple Formica fuscas.
4. a kind of feature selection approach based on binary ant colony algorithm according to claim 3, it is characterised in that described
Feature selection is carried out to the sample characteristics collection using the binary ant colony algorithm comprising the visibility information, is specifically included：
Candidate solution is searched in the ant colony, candidate solution set is obtained；
The candidate solution set is decoded, second feature subset is obtained；
The training sample set is classified using the second feature subset, obtain the second classification results；
Calculate the second classification accuracy of second classification results；
The individual Formica fusca fitness of corresponding Formica fusca is solved according to second classification accuracy；
Determine that optimum Formica fusca is individual through multiple computing；The individual Formica fusca for Formica fusca fitness maximum of the optimum Formica fusca
Body；
Decode the individual corresponding optimal feature subset of the optimum Formica fusca.
5. a kind of feature selection system based on binary ant colony algorithm, it is characterised in that include：
Sample acquisition module, needs to carry out the training sample set of feature selection for obtaining；
Feature extraction module, for carrying out feature extraction to the training sample set, obtains sample characteristics collection；
Hereditary optimizing module, for being classified and being sought maximum heredity using binary strings genetic algorithm to the sample characteristics collection
Fitness, obtains optimal solution；The genetic adaptation degree is the journey for making the result of the binary strings genetic algorithm be close to objective result
Degree；
Visibility generation module, for the visibility information of binary ant colony algorithm is arranged according to the optimal solution, to described two
The ant colony of system ant group algorithm is initialized；
Ant colony selecting module, for utilizing the binary ant colony algorithm comprising the visibility information to enter the sample characteristics collection
Row feature selection.
6. a kind of feature selection system based on binary ant colony algorithm according to claim 5, it is characterised in that described
Hereditary optimizing module, specifically includes：
Parameter initial cell, for being initialized for the parameter of binary strings genetic algorithm described in the sample training set pair,
Generate genetic groups；The individual length of wherein described heredity is set to the feature quantity included by the sample characteristics collection；It is described
The hereditary individual unit to constitute genetic groups；
Population decoding unit, for decoding to the genetic groups, obtains fisrt feature subset；
First taxon, for classifying to the training sample set using the fisrt feature subset, obtains first point
Class result；
First accuracy rate computing unit, for calculating the first classification accuracy of first classification results；
Genetic adaptation degree computing unit, for solving the individual genetic adaptation of corresponding heredity according to first classification accuracy
Degree；
Optimal solution computing unit, for passing through multiple genetic manipulation, solves the maximum of the genetic adaptation degree, obtains maximum something lost
Pass fitness；The heredity that the optimal solution is the maximum genetic adaptation degree is individual.
7. a kind of feature selection system based on binary ant colony algorithm according to claim 5, it is characterised in that described
Visibility generation module, specifically includes：
Binary number acquiring unit, for obtaining the binary numeral of the optimal solution；
Visibility computing unit, for 1 in the binary numeral is replaced with the first preset number, by the binary number
In value 0 replaces with the second preset number, obtains the visibility information of binary ant colony algorithm；
Ant colony signal generating unit, for being initialized to the parameter of the binary ant colony algorithm according to the visibility information,
Ant colony is generated, the ant colony is the individual set for constituting of multiple Formica fuscas.
8. a kind of feature selection system based on binary ant colony algorithm according to claim 7, it is characterised in that described
Ant colony selecting module, specifically includes：
Candidate solution search unit, for searching for candidate solution in the ant colony, obtains candidate solution set；
Candidate solution decoding unit, for decoding to the candidate solution set, obtains second feature subset；
Second taxon, for classifying to the training sample set using the second feature subset, obtains second point
Class result；
Second accuracy rate computing unit, calculates the second classification accuracy of second classification results；
Formica fusca fitness computing unit, adapts to for solving the individual Formica fusca of corresponding Formica fusca according to second classification accuracy
Degree；
For passing through multiple computing, optimum individual determining unit, determines that optimum Formica fusca is individual；It is described that the optimum Formica fusca is individual
The maximum Formica fusca of Formica fusca fitness is individual；
Optimal characteristics determining unit, for decoding the individual corresponding optimal feature subset of the optimum Formica fusca.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611246351.6A CN106599936A (en) | 2016-12-29 | 2016-12-29 | Characteristic selection method based on binary ant colony algorithm and system thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611246351.6A CN106599936A (en) | 2016-12-29 | 2016-12-29 | Characteristic selection method based on binary ant colony algorithm and system thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106599936A true CN106599936A (en) | 2017-04-26 |
Family
ID=58605146
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611246351.6A Pending CN106599936A (en) | 2016-12-29 | 2016-12-29 | Characteristic selection method based on binary ant colony algorithm and system thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106599936A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107437098A (en) * | 2017-07-31 | 2017-12-05 | 湖北工业大学 | A kind of hyperspectral image band selection method based on the improved binary ant colony algorithm of differential evolution |
| CN107590195A (en) * | 2017-08-14 | 2018-01-16 | 百度在线网络技术（北京）有限公司 | Textual classification model training method, file classification method and its device |
| CN107808164A (en) * | 2017-10-17 | 2018-03-16 | 湖北工业大学 | A kind of texture image feature selection approach based on fireworks algorithm |
| US20180330258A1 (en) * | 2017-05-09 | 2018-11-15 | Theodore D. Harris | Autonomous learning platform for novel feature discovery |
| CN108985010A (en) * | 2018-06-15 | 2018-12-11 | 河南师范大学 | Gene sorting method and device |
| CN110034920A (en) * | 2019-04-09 | 2019-07-19 | 中国人民解放军战略支援部队信息工程大学 | The mapping method and device of the restructural cryptologic array of coarseness |
| CN110442128A (en) * | 2019-07-20 | 2019-11-12 | 河北科技大学 | AGV paths planning method based on feature point extraction ant group algorithm |
| CN112612886A (en) * | 2020-12-18 | 2021-04-06 | 湖北工业大学 | Text feature selection method based on rice breeding-binary ant colony hybrid algorithm |
-
2016
- 2016-12-29 CN CN201611246351.6A patent/CN106599936A/en active Pending
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180330258A1 (en) * | 2017-05-09 | 2018-11-15 | Theodore D. Harris | Autonomous learning platform for novel feature discovery |
| US11586960B2 (en) * | 2017-05-09 | 2023-02-21 | Visa International Service Association | Autonomous learning platform for novel feature discovery |
| CN107437098B (en) * | 2017-07-31 | 2019-11-26 | 湖北工业大学 | A kind of hyperspectral image band selection method based on the improved binary ant colony algorithm of differential evolution |
| CN107437098A (en) * | 2017-07-31 | 2017-12-05 | 湖北工业大学 | A kind of hyperspectral image band selection method based on the improved binary ant colony algorithm of differential evolution |
| CN107590195A (en) * | 2017-08-14 | 2018-01-16 | 百度在线网络技术（北京）有限公司 | Textual classification model training method, file classification method and its device |
| CN107808164A (en) * | 2017-10-17 | 2018-03-16 | 湖北工业大学 | A kind of texture image feature selection approach based on fireworks algorithm |
| CN108985010B (en) * | 2018-06-15 | 2022-04-08 | 河南师范大学 | Gene classification method and apparatus |
| CN108985010A (en) * | 2018-06-15 | 2018-12-11 | 河南师范大学 | Gene sorting method and device |
| CN110034920A (en) * | 2019-04-09 | 2019-07-19 | 中国人民解放军战略支援部队信息工程大学 | The mapping method and device of the restructural cryptologic array of coarseness |
| CN110034920B (en) * | 2019-04-09 | 2023-03-31 | 中国人民解放军战略支援部队信息工程大学 | Mapping method and device for coarse-grained reconfigurable cipher logic array |
| CN110442128A (en) * | 2019-07-20 | 2019-11-12 | 河北科技大学 | AGV paths planning method based on feature point extraction ant group algorithm |
| CN110442128B (en) * | 2019-07-20 | 2022-08-16 | 河北科技大学 | AGV path planning method based on characteristic point extraction ant colony algorithm |
| CN112612886A (en) * | 2020-12-18 | 2021-04-06 | 湖北工业大学 | Text feature selection method based on rice breeding-binary ant colony hybrid algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106599936A (en) | Characteristic selection method based on binary ant colony algorithm and system thereof | |
| CN108897989B (en) | Biological event extraction method based on candidate event element attention mechanism | |
| CN108073711B (en) | Relation extraction method and system based on knowledge graph | |
| CN106407352B (en) | Traffic image search method based on deep learning | |
| CN109543180A (en) | A kind of text emotion analysis method based on attention mechanism | |
| CN109299262A (en) | A kind of text implication relation recognition methods for merging more granular informations | |
| CN104298873B (en) | A kind of attribute reduction method and state of mind appraisal procedure based on genetic algorithm and rough set | |
| CN112527938A (en) | Chinese POI matching method based on natural language understanding | |
| CN102411687B (en) | Deep learning detection method of unknown malicious codes | |
| CN109885824A (en) | A kind of Chinese name entity recognition method, device and the readable storage medium storing program for executing of level | |
| CN110110330A (en) | Text based keyword extracting method and computer equipment | |
| CN108009405A (en) | A kind of method based on machine learning techniques prediction Bacterial outer membrane proteins matter | |
| CN107437098B (en) | A kind of hyperspectral image band selection method based on the improved binary ant colony algorithm of differential evolution | |
| CN109933656A (en) | Public sentiment polarity prediction technique, device, computer equipment and storage medium | |
| CN111931505A (en) | Cross-language entity alignment method based on subgraph embedding | |
| CN107992890B (en) | A kind of multi-angle of view classifier and design method based on local feature | |
| CN111429977B (en) | Novel molecular similarity search algorithm based on attention of graph structure | |
| Yang et al. | Active matting | |
| CN112884802B (en) | Attack resistance method based on generation | |
| CN105893876A (en) | Chip hardware Trojan horse detection method and system | |
| CN109448794A (en) | A kind of epistasis site method for digging based on heredity taboo and Bayesian network | |
| CN103258147A (en) | Parallel evolution super-network DNA micro array gene data sorting system and method based on GPU | |
| CN114676353B (en) | Address matching method based on segmentation inference | |
| CN112364238A (en) | Deep learning-based user interest point recommendation method and system | |
| CN113268612A (en) | Heterogeneous information network knowledge graph completion method and device based on mean value fusion |
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 |
Application publication date: 20170426 |
|
| RJ01 | Rejection of invention patent application after publication |