Genetic algorithm-based classification method and system for divide-and-conquer feature selection

Abstract

The invention belongs to the technical field of data classification, and provides a genetic algorithm-based classification method and system for divide-and-conquer feature selection, and the method comprises the steps: obtaining to-be-classified data; performing feature grouping on the acquired data to be classified; and carrying out feature selection on the grouped features by adopting a genetic algorithm to obtain an optimal feature subset, namely, a classification result, so as to complete the classification of the genetic algorithm-based divide-and-conquer feature selection. According to the method, irrelevant features are eliminated, really relevant features are selected, the classification time of feature selection is shortened, and the feature selection process in a classification task is effectively accelerated.

Description

Technical Field

[0001] The present invention belongs to the technical field of data classification, and in particular relates to a classification method and system for feature selection based on a divide-and-conquer genetic algorithm. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In recent years, the speed and amount of data generation have increased rapidly, and various fields have posed challenges to high-dimensional data; many of the features of these high-dimensional data may be irrelevant to the classification task, and may even reduce the classification performance of the learning algorithm due to its large search space. Without any prior knowledge, it is an extremely difficult task to identify useful features from a large number of data features. Therefore, how to quickly and efficiently filter out useful features from a large number of data features has become a key issue hindering the development of machine learning and data mining.

[0004] Feature selection is a crucial preprocessing step. It aims to improve the accuracy of classification tasks, speed up the training of classification models, and reduce the risk of overfitting classification models by selecting the most relevant features from the original data set and reducing the complexity of the data. According to different feature evaluation criteria, feature selection methods can be divided into three categories: filtering, packaging, and embedding. The filtering method evaluates the importance of features based on the intrinsic properties of the data and selects features independently of the classification model; commonly used metrics include distance metrics, correlation metrics, consistency metrics, and information metrics. The packaging method uses the predictive performance of the classification algorithm as an evaluation indicator, and selects feature combinations that maximize model performance by continuously adjusting feature subsets during the training process; commonly used classification algorithms include K-nearest neighbors, support vector machines, neural networks, and Bayesian networks. The embedding method embeds the feature selection process into the model training process, completing feature selection and model construction at the same time; commonly used techniques include feature importance ranking and regularization of decision trees.

[0005] In order to improve the efficiency of feature selection, the divide and conquer strategy is introduced into the feature selection process. The divide and conquer method divides the high-dimensional feature space into several subspaces, selects features for each subspace separately, and finally constructs the global optimal feature subset through the fusion strategy; it greatly reduces the dimension of the search space, thereby reducing the computational complexity and accelerating the feature selection process. Feature selection can essentially be regarded as a complex combinatorial optimization problem. Assuming that the data set contains n original features, the goal of feature selection is to select n features from these 2. nThe exhaustive search is a process that selects an optimal subset from the potential feature combinations. When the number of features increases to a certain size, exhaustive search becomes almost infeasible because the computational complexity increases exponentially with the growth of the number of features.

[0006] According to the inventors' understanding, the prior art has the following technical problems:

[0007] Existing feature selection is mostly based on the operator's experience. The selected features are not representative or are representative but have limitations, which will lead to low classification accuracy, poor classification precision and slow classification speed. Summary of the invention

[0008] To solve the above problems, the present invention proposes a classification method and system for feature selection based on a divide-and-conquer genetic algorithm, which eliminates irrelevant features, selects truly relevant features, shortens the classification time of feature selection, and effectively accelerates the feature selection process in classification tasks.

[0009] According to some embodiments, a first solution of the present invention provides a classification method for feature selection based on a divide-and-conquer genetic algorithm, using the following technical solution:

[0010] A classification method for feature selection based on a divide-and-conquer genetic algorithm, comprising:

[0011] Obtain data to be classified;

[0012] Perform feature grouping on the acquired data to be classified;

[0013] The genetic algorithm is used to select the grouped features to obtain the optimal feature subset, that is, the classification result, and the classification of feature selection based on the divide-and-conquer method of the genetic algorithm is completed;

[0014] Among them, in the process of selecting features after grouping by using genetic algorithm, the fitness of individuals is evaluated based on the fitness evaluation function, and individuals of the population are selected according to the obtained fitness values. The selected individuals are crossover operated, and the chromosome fragments are exchanged according to the preset probability to obtain new offspring individuals. The offspring individuals after crossover are mutated, and the chromosome genes are randomly changed based on probability to change the selection state of the features, generate a new population, and evaluate the fitness value of each individual in the new population; the current population is updated according to the obtained fitness value to obtain the global optimal individual; when the number of iterations reaches the maximum number of iterations, the evolution is stopped, and the obtained global optimal individual is the optimal feature subset, and the classification of feature selection is completed.

[0015] As a further technical limitation, in the process of performing the crossover operation on the offspring individuals after the crossover, for each individual, the chromosome fragments are exchanged between the selected parent individuals to generate new offspring individuals Experimental Example Cij ;Right now Among them, C ij represents the gene value of the generated offspring individual at position j, P 1j and P 2j is the gene value of two randomly selected parent individuals at position j; j is the position of the gene, selected from [1, D], D is the chromosome length (number of features); rand is a random number generated in the range [0, 1], CR is the crossover probability, which is used to control the probability of gene inheritance from the parent individual, j rand It is a randomly selected gene position that ensures that the offspring individuals have at least one gene that is different from the parent.

[0016] As a further technical limitation, in the process of performing mutation operation on the offspring individuals after crossover, for each individual, the values ​​of some genes in the chromosome are randomly changed to generate a new offspring individual experimental example M ij ;Right now Among them, M ij represents the gene value of the individual at position j after mutation, P j is the gene value of the target chromosome of the mutation operation at position j; j is the position of the gene, selected from [1, D], and D is the chromosome length; rand is a random number generated in the range of [0, 1], and MR is the mutation probability, which is used to control the possibility of gene mutation.

[0017] As a further technical limitation, in the process of updating the current population according to the obtained fitness value, for the generated new population, the individual optimal fitness value of each chromosome is updated, and the chromosome with the best fitness value is selected as the global optimal solution of the population;

[0018] When the updating process is based on the fitness value of each chromosome, for each chromosome in the population, the current fitness value is compared with its historical optimal fitness value. If the current fitness value is better than the historical optimal fitness value, the chromosome is used as the new individual optimal solution, otherwise the historical optimal solution remains unchanged; the fitness values ​​of the individual optimal solutions of all chromosomes in the population are compared, and the chromosome with the best fitness value is selected as the global optimal solution of the current population.

[0019] As a further technical limitation, before the acquired data to be classified is subjected to feature grouping, the acquired data to be classified needs to be normalized, and feature extraction and feature grouping are performed on the normalized data.

[0020] It should be noted that feature extraction of normalized data is a technology that should be known to those skilled in the art, and the present invention does not elaborate on it in detail.

[0021] Furthermore, the importance between the extracted data features and the data feature class labels is calculated, a three-dimensional cube is constructed based on the obtained importance, and the data features are grouped according to the side lengths of the obtained cube.

[0022] According to some embodiments, a second solution of the present invention provides a classification system for feature selection based on a divide-and-conquer genetic algorithm, using the following technical solution:

[0023] A classification system based on divide-and-conquer feature selection of genetic algorithm, including:

[0024] An acquisition module, configured to acquire data to be classified;

[0025] A grouping module, which is configured to perform feature grouping on the acquired data to be classified;

[0026] A classification module is configured to use a genetic algorithm to perform feature selection on the grouped features to obtain an optimal feature subset, that is, to obtain a classification result, thereby completing the classification of feature selection based on a divide-and-conquer method of the genetic algorithm;

[0027] Among them, in the process of selecting features after grouping by using genetic algorithm, the fitness of individuals is evaluated based on the fitness evaluation function, and individuals of the population are selected according to the obtained fitness values. The selected individuals are crossover operated, and the chromosome fragments are exchanged according to the preset probability to obtain new offspring individuals. The offspring individuals after crossover are mutated, and the chromosome genes are randomly changed based on probability to change the selection state of the features, generate a new population, and evaluate the fitness value of each individual in the new population; the current population is updated according to the obtained fitness value to obtain the global optimal individual; when the number of iterations reaches the maximum number of iterations, the evolution is stopped, and the obtained global optimal individual is the optimal feature subset, and the classification of feature selection is completed.

[0028] According to some embodiments, a third solution of the present invention provides a computer-readable storage medium, which adopts the following technical solution:

[0029] A computer-readable storage medium stores a program, which, when executed by a processor, implements the steps of the classification method for feature selection based on divide-and-conquer genetic algorithm as described in the first solution of the present invention.

[0030] According to some embodiments, a fourth solution of the present invention provides an electronic device, which adopts the following technical solution:

[0031] An electronic device comprises a memory, a processor and a program stored in the memory and running on the processor, wherein when the processor executes the program, the steps in the classification method of divide-and-conquer feature selection based on genetic algorithm as described in the first scheme of the present invention are implemented.

[0032] According to some embodiments, a fifth solution of the present invention provides a computer program product, which adopts the following technical solution:

[0033] A computer program product comprises software codes, wherein the program in the software codes executes the steps in the classification method of divide-and-conquer feature selection based on genetic algorithm as described in the first solution of the present invention.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] The present invention has high classification accuracy, strong classification robustness and fast classification speed; it has the advantages of simple implementation, fast mining speed, high classification accuracy and few adjustable parameters; 3. Selecting truly relevant features can further simplify the model and facilitate operation by practitioners; eliminating irrelevant features can further shorten the model training time and enhance its practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings in the specification that constitute a part of this embodiment are used to provide a further understanding of this embodiment. The schematic embodiments of this embodiment and their descriptions are used to explain this embodiment and do not constitute improper limitations on this embodiment.

[0037] Figure 1 This is a flow chart of a classification method for feature selection based on a divide-and-conquer genetic algorithm in Embodiment 1 of the present invention;

[0038] Figure 2 This is a flow chart of using a genetic algorithm to select features after grouping in Embodiment 1 of the present invention;

[0039] Figure 3 This is a fitness value change trend diagram of optimizing Prostate_GE data using the method of this embodiment and the original particle swarm algorithm in Embodiment 1 of the present invention;

[0040] Figure 4 This is a fitness value change trend diagram of optimizing Armstrong data using the method of this embodiment and the particle swarm algorithm in Embodiment 1 of the present invention;

[0041] Figure 5 This is a fitness value change trend diagram of optimizing Colon data using the method of this embodiment and the particle swarm algorithm in Embodiment 1 of the present invention;

[0042] Figure 6 This is a fitness value change trend diagram of optimizing DLBCL data using the method of this embodiment and the particle swarm algorithm in Embodiment 1 of the present invention;

[0043] Figure 7This is a fitness value change trend diagram of optimizing LSVT data using the method of this embodiment and the particle swarm algorithm in Embodiment 1 of the present invention;

[0044] Figure 8 This is a fitness value change trend diagram of optimizing CLLSUB data using the method of this embodiment and the particle swarm algorithm in Embodiment 1 of the present invention;

[0045] Fig. 9 This is a fitness value change trend diagram of optimizing CLLSUB data using the method of this embodiment and the particle swarm algorithm in Embodiment 1 of the present invention;

[0046] Fig.10 This is a fitness value change trend diagram of optimizing WarpPIE10P data using the method of this embodiment and the particle swarm algorithm in Embodiment 1 of the present invention;

[0047] Fig.11 This is a structural block diagram of a classification system for feature selection based on a divide-and-conquer genetic algorithm in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0048] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0049] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0050] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0051] In the present invention, terms such as "upper", "lower", "left", "right", "front", "back", "vertical", "horizontal", "side", "bottom", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. They are relational words determined only for the convenience of describing the structural relationships of the various parts or elements of the present invention, and do not specifically refer to any part or element in the present invention and should not be understood as limitations on the present invention.

[0052] In the present invention, terms such as "fixed connection", "connected", "connection", etc. should be understood in a broad sense, indicating that it can be fixedly connected, integrally connected or detachably connected; it can be directly connected or indirectly connected through an intermediate medium. For relevant scientific research or technical personnel in this field, the specific meanings of the above terms in the present invention can be determined according to specific circumstances, and they cannot be understood as limitations on the present invention.

[0053] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0054] Embodiment 1

[0055] Embodiment 1 of the present invention introduces a classification method for feature selection based on a divide-and-conquer genetic algorithm.

[0056] like Figure 1 A classification method of feature selection based on a divide-and-conquer genetic algorithm is shown, comprising:

[0057] Obtain data to be classified;

[0058] Perform feature grouping on the acquired data to be classified;

[0059] The genetic algorithm is used to select the grouped features to obtain the optimal feature subset, that is, the classification result, and the classification of feature selection based on the divide-and-conquer method of the genetic algorithm is completed;

[0060] Among them, in the process of selecting features after grouping by using genetic algorithm, the fitness of individuals is evaluated based on the fitness evaluation function, and individuals of the population are selected according to the obtained fitness values. The selected individuals are crossover operated, and the chromosome fragments are exchanged according to the preset probability to obtain new offspring individuals. The offspring individuals after crossover are mutated, and the chromosome genes are randomly changed based on probability to change the selection state of the features, generate a new population, and evaluate the fitness value of each individual in the new population; the current population is updated according to the obtained fitness value to obtain the global optimal individual; when the number of iterations reaches the maximum number of iterations, the evolution is stopped, and the obtained global optimal individual is the optimal feature subset, and the classification of feature selection is completed.

[0061] As one or more implementation modes, before the acquired data to be classified is subjected to feature grouping, the acquired data to be classified needs to be normalized, and feature extraction and feature grouping are performed on the normalized data.

[0062] It should be noted that feature extraction of normalized data is a technology that should be known to those skilled in the art, and the present invention does not elaborate on it in detail.

[0063] As one or more implementation modes, the importance between the extracted data features and the data feature class labels is calculated according to three evaluation indicators: Pearson correlation coefficient, SU symmetric uncertainty, and Relieff coefficient. A three-dimensional cube is constructed based on the obtained importance, and the data features are grouped according to the side length of the obtained cube.

[0064] In this embodiment, the process of using genetic algorithm to select features after grouping is as follows: Figure 2 As shown, specifically:

[0065] Input the feature set for feature selection; encode the feature set into chromosomes, each chromosome represents a feature subset, and each gene of the chromosome corresponds to a feature (1 means the feature is selected, 0 means it is not selected);

[0066] Initialize the population size, maximum number of iterations and crossover probability; randomly generate the initial population, where each individual (chromosome) represents a feature subset;

[0067] Evaluate the fitness value of each individual in the population according to the accuracy of the fitness evaluation function classifier;

[0068] According to the fitness value, the individuals in the population are selected, and individuals with higher fitness are retained to enter the next generation;

[0069] Perform a crossover operation on the selected individuals, exchanging chromosome segments with a certain probability to generate new offspring individuals;

[0070] Perform mutation operation on the offspring individuals after crossover, and randomly change the genes in the chromosomes according to a certain probability (i.e. change the selection state of the characteristics);

[0071] Generate a new population and evaluate the fitness value of each individual in the new population;

[0072] Update the current population according to the fitness value and save the current global optimal individual;

[0073] Determine whether the current number of iterations has reached the maximum number of iterations; if so, stop the evolution and decode the global optimal individual into the selected optimal feature subset; otherwise, return to re-select the individual to continue the next round of evolution.

[0074] In this embodiment, the population size is set to 100, the maximum number of function evaluations is set to 100, and each individual (chromosome) represents a feature subset.

[0075] As one or more implementation modes, according to the encoding state of the chromosome in the feature selection space, the fitness value of each chromosome is calculated using the fitness evaluation function of the feature subset. This value is a criterion for evaluating the selected feature subset. The smaller the value, the better, that is, the smaller the classification error rate, the better the fitness value.

[0076] As one or more implementations, a selection operation is used to generate the next generation of population according to the fitness value of each individual in the population. Individuals with higher fitness (the larger the fitness value, the better) are preferentially selected to enter the next generation, and a roulette wheel selection method can be used to ensure that high-quality individuals are retained during the evolution process, thereby improving the overall population quality.

[0077] As one or more embodiments, for each individual, a crossover operation is used to exchange chromosome segments between selected parent individuals to generate new offspring individuals. ij ,Right now:

[0078]

[0079] Among them, C ij represents the gene value of the generated offspring individual at position j, P 1j and P 2j is the gene value of two randomly selected parent individuals at position j; j is the position of the gene, selected from [1, D], D is the chromosome length (number of features); rand is a random number generated in the range [0, 1], CR is the crossover probability, which is used to control the probability of gene inheritance from the parent individual, j rand It is a randomly selected gene position that ensures that the offspring individuals have at least one gene that is different from the parent.

[0080] As one or more implementations, for each individual, the mutation operation is used to randomly change the value of some genes in the chromosome to generate a new offspring individual experimental example M ij ;Right now:

[0081]

[0082] Among them, M ij represents the gene value of the individual at position j after mutation, P j is the gene value of the target chromosome of the mutation operation at position j; j is the position of the gene, selected from [1, D], and D is the chromosome length; rand is a random number generated in the range of [0, 1], and MR is the mutation probability, which is used to control the possibility of gene mutation.

[0083] As one or more implementations, a new population is generated through selection, crossover and mutation operations, and each individual (chromosome) represents a new feature subset. Subsequently, each individual in the new population is evaluated using a fitness evaluation function to calculate its fitness value.

[0084] As one or more implementation modes, for the generated new population, for the generated new population, the individual optimal fitness value of each chromosome is updated, and the chromosome with the best fitness value is selected as the global optimal solution of the population;

[0085] When the updating process is based on the fitness value of each chromosome, for each chromosome in the population, the current fitness value is compared with its historical optimal fitness value. If the current fitness value is better than the historical optimal fitness value, the chromosome is used as the new individual optimal solution, otherwise the historical optimal solution remains unchanged;

[0086] The fitness values ​​of the individual optimal solutions of all chromosomes in the population are compared, and the chromosome with the best fitness value is selected as the global optimal solution of the current population.

[0087] This embodiment may also determine an evaluation feature subset, specifically:

[0088] Feature evaluation is an important step in evaluating the quality of the selected feature subset in the feature selection process; since the K-nearest neighbor algorithm has efficient and easy-to-implement learning characteristics, this embodiment uses the K-nearest neighbor algorithm as a learning algorithm to evaluate the classification accuracy of the selected feature subset, where the K value is set to 5.

[0089] This embodiment establishes a fitness evaluation function for evaluating the selected feature subset, specifically:

[0090] Considering that the feature selection process is a multi-objective optimization task, that is, high classification accuracy and a small number of selected features. Therefore, this embodiment uses the classification error rate of the learning algorithm and the length of the feature subset as indicators for evaluating the quality of the selected feature subset, and connects them with a weight factor. Therefore, the fitness evaluation function of the feature subset can be expressed as:

[0091]

[0092] Among them, γ R (D) represents the classification error rate of the feature subset R relative to the target data set D; |S| represents the length of the selected feature subset; |N| represents the total number of features in the data set; α is a weight factor, which is a parameter used to adjust the ratio of the classification error rate and the feature subset length. In this embodiment, α is set to 0.9.

[0093] Since most data sets in reality are unbalanced, balanced accuracy is used to calculate the fitness evaluation function in this embodiment.

[0094] In order to avoid feature selection bias in the feature selection process, 5-fold cross validation is used to evaluate the classification accuracy on the training set. The classification error rate is calculated as follows:

[0095]

[0096] Among them, c represents the number of categories in the data set, TPR i Represents the accuracy of correct identification in category i.

[0097] In order to illustrate the superiority of this embodiment in processing feature selection of classification tasks, the fitness value convergence curves obtained by comparing the method in this embodiment with the particle swarm algorithm when processing eight classification data sets: Prostate_GE, Armstrong, Colon, DLBCL, LSVT, CLLSUB, Carcinom and WarpPIE10P are as follows:

[0098] Figure 3 This is a fitness value change trend diagram of the method of this embodiment and the original particle swarm algorithm in the Prostate_GE classification problem in the example; Figure 4 This is a fitness value change trend diagram of the Armstrong classification problem using the method of this embodiment and the particle swarm algorithm in the example; Figure 5 This is a fitness value change trend diagram of the method of this embodiment and the particle swarm algorithm in the Colon classification problem in the example; Figure 6 This is a fitness value change trend diagram of the DLBCL classification problem using the method of this embodiment and the particle swarm algorithm in the example; Figure 7 This is a fitness value change trend diagram of the method of this embodiment and the particle swarm algorithm in the LSVT classification problem in the example; Figure 8 This is a fitness value change trend diagram of the method of this embodiment and the particle swarm algorithm in the CLLSUB classification problem in the example; Fig. 9 This is a fitness value change trend diagram of the method of this embodiment and the particle swarm algorithm in the Carcinom classification problem in the example; Fig.10 The figure is a fitness value change trend chart of the WarpPIE10P classification problem using the method of this embodiment and the particle swarm algorithm in the example; in the example, 8 classic classification problems are given as examples, each classification problem is executed independently 30 times, the number of particles in the population is set to 100, the maximum number of iterations is set to 100, and the performance indicator value obtained each time is recorded.

[0099] Table 1 shows the statistical results of the classification accuracy of the optimal feature subset obtained by the genetic algorithm and the method of this embodiment in the classification task. CR (%) in the table represents the classification accuracy of the selected feature subset in the classifier. By comparison, it can be seen that the feature selection method proposed in this embodiment can better select feature subsets with higher classification accuracy, and the ability to remove useless features in the data set is significantly enhanced. The convergence speed of this embodiment has been significantly improved, which also reflects the high efficiency of the feature selection method proposed in this embodiment.

[0100] Table 1 Comparison of classification accuracy for eight classification problems

[0101]

[0102]

[0103] Evolutionary algorithms have been widely proven to have strong global search capabilities and can efficiently find global optimal or near-optimal solutions in complex search spaces. Genetic algorithms, as a type of evolutionary algorithm, are now widely used in solving practical problems because of their high efficiency, easy implementation and applicability to a variety of complex problems. In feature selection tasks, genetic algorithms have a natural feature representation method, which represents the state of feature selection through binary coding, can directly and efficiently handle feature subset selection problems, and are the preferred algorithms for feature selection. Compared with other evolutionary algorithms, genetic algorithms use fitness functions to guide selection, crossover and mutation operations to update populations by simulating natural selection and genetics principles. In the search process, each generation gradually approaches the global optimal solution by retaining individuals with higher fitness and introducing diversity in crossover and mutation. However, traditional genetic algorithms may face the problems of insufficient population diversity and premature convergence in feature selection problems. Therefore, this embodiment proposes a divide-and-conquer strategy to decompose complex feature selection problems into several smaller sub-problems to solve one by one, and effectively solve the problems of insufficient population diversity and premature convergence by dividing the feature space into multiple subspaces and optimizing them separately.

[0104] Embodiment 2

[0105] The second embodiment of the present invention introduces a classification system for feature selection based on a divide-and-conquer genetic algorithm.

[0106] like Fig.11 A classification system of feature selection based on divide-and-conquer genetic algorithm is shown, comprising:

[0107] An acquisition module, configured to acquire data to be classified;

[0108] A grouping module, which is configured to perform feature grouping on the acquired data to be classified;

[0109] A classification module is configured to use a genetic algorithm to perform feature selection on the grouped features to obtain an optimal feature subset, that is, to obtain a classification result, thereby completing the classification of feature selection based on a divide-and-conquer method of the genetic algorithm;

[0110] Among them, in the process of selecting features after grouping by using genetic algorithm, the fitness of individuals is evaluated based on the fitness evaluation function, and individuals of the population are selected according to the obtained fitness values. The selected individuals are crossover operated, and the chromosome fragments are exchanged according to the preset probability to obtain new offspring individuals. The offspring individuals after crossover are mutated, and the chromosome genes are randomly changed based on probability to change the selection state of the features, generate a new population, and evaluate the fitness value of each individual in the new population; the current population is updated according to the obtained fitness value to obtain the global optimal individual; when the number of iterations reaches the maximum number of iterations, the evolution is stopped, and the obtained global optimal individual is the optimal feature subset, and the classification of feature selection is completed.

[0111] The detailed steps are the same as the classification method of divide-and-conquer feature selection based on genetic algorithm provided in Example 1, and will not be repeated here.

[0112] Embodiment 3

[0113] Embodiment 3 of the present invention provides a computer-readable storage medium.

[0114] A computer-readable storage medium stores a program, which, when executed by a processor, implements the steps of the classification method for feature selection based on divide-and-conquer genetic algorithm as described in Embodiment 1 of the present invention.

[0115] The detailed steps are the same as the classification method of divide-and-conquer feature selection based on genetic algorithm provided in Example 1, and will not be repeated here.

[0116] Embodiment 4

[0117] A fourth embodiment of the present invention provides an electronic device.

[0118] An electronic device comprises a memory, a processor and a program stored in the memory and running on the processor, wherein when the processor executes the program, the steps in the classification method of divide-and-conquer feature selection based on genetic algorithm as described in the first embodiment of the present invention are implemented.

[0119] The detailed steps are the same as the classification method of divide-and-conquer feature selection based on genetic algorithm provided in Example 1, and will not be repeated here.

[0120] Embodiment 5

[0121] Embodiment 5 of the present invention provides a computer program product.

[0122] A computer program product includes software codes, wherein the program in the software codes executes the steps in the classification method for feature selection based on divide-and-conquer genetic algorithm as described in the first embodiment of the present invention.

[0123] The detailed steps are the same as the classification method of divide-and-conquer feature selection based on genetic algorithm provided in Example 1, and will not be repeated here.

[0124] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0125] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0126] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0127] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0128] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0129] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0130] The above description is only a preferred embodiment of the present embodiment and is not intended to limit the present embodiment. For those skilled in the art, the present embodiment may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present embodiment shall be included in the protection scope of the present embodiment.

Claims

1. A classification method for feature selection based on a divide-and-conquer genetic algorithm, characterized in that: include: Obtain data to be classified; Perform feature grouping on the acquired data to be classified; The genetic algorithm is used to select the grouped features to obtain the optimal feature subset, that is, the classification result, and the classification of feature selection based on the divide-and-conquer method of the genetic algorithm is completed; Among them, in the process of selecting features after grouping by using genetic algorithm, the fitness of individuals is evaluated based on the fitness evaluation function, and individuals of the population are selected according to the obtained fitness values. The selected individuals are crossover operated, and the chromosome fragments are exchanged according to the preset probability to obtain new offspring individuals. The offspring individuals after crossover are mutated, and the chromosome genes are randomly changed based on probability to change the selection state of the features, generate a new population, and evaluate the fitness value of each individual in the new population; the current population is updated according to the obtained fitness value to obtain the global optimal individual; when the number of iterations reaches the maximum number of iterations, the evolution is stopped, and the obtained global optimal individual is the optimal feature subset, and the classification of feature selection is completed.

2. A classification method for feature selection based on a divide-and-conquer genetic algorithm as claimed in claim 1, characterized in that: In the process of performing the crossover operation on the offspring individuals after the crossover, for each individual, the chromosome fragments are exchanged between the selected parent individuals to generate new offspring individuals Experimental Example C ij ;Right now Among them, C ij represents the gene value of the generated offspring individual at position j, P 1j and P 2j is the gene value of two randomly selected parent individuals at position j; j is the position of the gene, selected from [1, D], D is the chromosome length (number of features); rand is a random number generated in the range [0, 1], CR is the crossover probability, which is used to control the probability of gene inheritance from the parent individual, j rand It is a randomly selected gene position that ensures that the offspring individuals have at least one gene that is different from the parent.

3. A classification method for feature selection based on divide and conquer genetic algorithm as claimed in claim 1, characterized in that: In the process of performing mutation operation on the offspring individuals after crossover, for each individual, the values ​​of some genes in the chromosome are randomly changed to generate a new offspring individual experimental example M ij ;Right now Among them, M ij represents the gene value of the individual at position j after mutation, P j is the gene value of the target chromosome of the mutation operation at position j; j is the position of the gene, selected from [1, D], and D is the chromosome length; rand is a random number generated in the range of [0, 1], and MR is the mutation probability, which is used to control the possibility of gene mutation.

4. A classification method for feature selection based on divide and conquer genetic algorithm as claimed in claim 1, characterized in that: In the process of updating the current population according to the obtained fitness value, for the generated new population, the individual optimal fitness value of each chromosome is updated, and the chromosome with the best fitness value is selected as the global optimal solution of the population; When the updating process is based on the fitness value of each chromosome, for each chromosome in the population, the current fitness value is compared with its historical optimal fitness value. If the current fitness value is better than the historical optimal fitness value, the chromosome is used as the new individual optimal solution, otherwise the historical optimal solution remains unchanged; the fitness values ​​of the individual optimal solutions of all chromosomes in the population are compared, and the chromosome with the best fitness value is selected as the global optimal solution of the current population.

5. A classification method for feature selection based on divide and conquer genetic algorithm as claimed in claim 1, characterized in that: Before the acquired data to be classified is grouped by features, the acquired data to be classified needs to be preprocessed. Before the acquired data to be classified is grouped by features, the acquired data to be classified needs to be normalized, and feature extraction and feature grouping are performed on the normalized data.

6. A classification method for feature selection based on divide and conquer genetic algorithm as claimed in claim 5, characterized in that: The importance between the extracted data features and the data feature class labels is calculated, a three-dimensional cube is constructed based on the obtained importance, and the data features are grouped according to the side length of the obtained cube.

7. A classification system based on divide-and-conquer feature selection of genetic algorithm, characterized in that: include: An acquisition module, configured to acquire data to be classified; A grouping module, which is configured to perform feature grouping on the acquired data to be classified; A classification module is configured to use a genetic algorithm to perform feature selection on the grouped features to obtain an optimal feature subset, that is, to obtain a classification result, and complete the classification of feature selection based on the divide-and-conquer method of the genetic algorithm; Among them, in the process of selecting features after grouping by using genetic algorithm, the fitness of individuals is evaluated based on the fitness evaluation function, and individuals of the population are selected according to the obtained fitness values. The selected individuals are crossover operated, and the chromosome fragments are exchanged according to the preset probability to obtain new offspring individuals. The offspring individuals after crossover are mutated, and the chromosome genes are randomly changed based on probability to change the selection state of the features, generate a new population, and evaluate the fitness value of each individual in the new population; the current population is updated according to the obtained fitness value to obtain the global optimal individual; when the number of iterations reaches the maximum number of iterations, the evolution is stopped, and the obtained global optimal individual is the optimal feature subset, and the classification of feature selection is completed.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the classification method of divide-and-conquer feature selection based on a genetic algorithm as described in any one of claims 1 to 6 are implemented.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the steps of the classification method of divide-and-conquer feature selection based on genetic algorithm as described in any one of claims 1 to 6 are implemented.

10. A computer program product comprising software code, characterized in that The program in the software code executes the steps of the classification method of divide-and-conquer feature selection based on genetic algorithm as described in any one of claims 1 to 6.

Images