A data processing method, device, equipment and readable storage medium

Abstract

The present invention provides a data processing method, device, equipment and readable storage medium, wherein the method comprises: obtaining a face data set containing multiple face feature data; obtaining the original population corresponding to the face data set, wherein the original population includes multiple feature individuals, and each of the multiple feature individuals contains multiple and equal number of face feature data; initializing the original population according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population; updating the initialized population according to the fitness value of each feature individual in the initialized population to obtain an updated population; determining the target feature individual according to the fitness value of each feature individual in the updated population, and determining the face feature data corresponding to the target feature individual as the target face feature data. The scheme of the present invention can improve the accuracy and efficiency of face feature data recognition.

Description

Technical Field

[0001] The present invention relates to the field of computer information processing technology, and in particular to a data processing method, device, equipment and readable storage medium. Background Art

[0002] Feature selection plays a key preprocessing role in machine learning and data mining. Its main function is to eliminate irrelevant and duplicate data, and it is widely used in machine learning and image analysis. With the rapid development of data acquisition technology, a large amount of high-dimensional data, such as text data, image data, etc., is continuously generated, which also brings a large amount of irrelevant and redundant data. These data have brought unprecedented challenges to data mining. The expansion of high-dimensional feature space makes the training and prediction of the model more complicated, thus affecting the classification performance. Therefore, feature selection technology is crucial in high-dimensional data processing, which can effectively simplify the feature space and improve model performance. In recent years, it has received more and more attention from scholars.

[0003] In order to make up for the shortcomings of existing algorithms and solve complex problems, in 2014, Mirialili et al. proposed the Grey Wolf Optimizer (GWO), inspired by the hierarchy and hunting mechanism of grey wolves. However, the original grey wolf optimizer sometimes easily falls into local optimality and has low convergence accuracy. Although the algorithm has its own advantages, it also has limitations. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a data processing method, device, equipment and readable storage medium to perform feature selection on incomplete facial data containing a large amount of noise, obtain the target facial feature data required by the face recognition model, and overcome the problems of low feature recognition accuracy, slow speed and poor robustness in the prior art.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a data processing method, including:

[0006] Acquire a face data set, wherein the face data set includes a plurality of face feature data;

[0007] According to the face data set, an original population corresponding to the face data set is obtained, wherein the original population includes a plurality of characteristic individuals, and each of the plurality of characteristic individuals includes a plurality of face characteristic data of the same quantity;

[0008] Initializing the original population according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population;

[0009] According to the fitness value of each characteristic individual in the initialization population, the initialization population is updated to obtain an updated population;

[0010] According to the fitness value of each characteristic individual in the updated population, a target characteristic individual is determined, and the facial feature data corresponding to the target characteristic individual is determined as the target facial feature data.

[0011] In one embodiment, the original population is initialized according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population, including:

[0012] Determining the initial position information corresponding to each characteristic individual in the original population according to the preset circular chaos mapping algorithm;

[0013] Transform the initial position information according to the preset quantum algorithm to obtain updated position information corresponding to the initial position information;

[0014] An initialization population corresponding to the original population is determined according to the initial position information and the updated position information.

[0015] In one embodiment, according to the preset circular chaos mapping algorithm, determining the initial position information corresponding to each characteristic individual in the original population includes:

[0016] Determining the chaotic mapping value of each characteristic individual in the original population according to the preset circular chaotic mapping algorithm;

[0017] According to the chaotic map value, the initial position information corresponding to each characteristic individual is determined.

[0018] In one embodiment, the initial position information is transformed according to the preset quantum algorithm to obtain updated position information corresponding to the initial position information, including:

[0019] Determining the quantum space position corresponding to each characteristic individual according to the initial position information of each characteristic individual in the original population;

[0020] According to the preset quantum bits and the quantum space position corresponding to each characteristic individual data, the updated position information corresponding to the initial position of each characteristic individual is determined.

[0021] In one embodiment, determining the initialization population corresponding to the original population according to the initial position information and the updated position information includes:

[0022] Determine the characteristic individual corresponding to the updated position information as an updated characteristic individual, wherein the updated characteristic individual corresponds one-to-one to each characteristic individual in the original population;

[0023] The initialized population is determined according to the fitness value of the updated characteristic individual and the fitness value of the characteristic individual in the original population corresponding to the currently updated characteristic individual.

[0024] In one embodiment, updating the initialization population according to the fitness value of each characteristic individual in the initialization population to obtain an updated population includes:

[0025] According to the fitness value of each characteristic individual in the initialization population, a plurality of characteristic individuals in the initialization population are subjected to stratification processing to obtain a plurality of population layers of different levels;

[0026] The level-based learning strategy performs iterative position update processing on the characteristic individuals in the plurality of population layers of different levels according to a preset number of iterations, and obtains the updated population.

[0027] In one embodiment, according to the fitness value of each feature individual in the updated population, a target feature individual is determined, and the facial feature data corresponding to the target feature individual is determined as the target facial feature data, including:

[0028] Determine the characteristic individual with the largest fitness value and the characteristic individual with the smallest fitness value in the updated population in each round of position iterative update processing;

[0029] Determine, according to the position information corresponding to the feature individual with the smallest fitness value and the preset number of iterations, a candidate feature individual corresponding to the feature individual with the largest fitness value;

[0030] Determine a target feature individual from the candidate feature individuals and the remaining feature individuals after removing the feature individual with the largest fitness value in the updated population;

[0031] The facial feature data corresponding to the target feature individual is determined as the target facial feature data.

[0032] An embodiment of the present invention further provides a data processing device, comprising:

[0033] An acquisition module is used to acquire a face data set, wherein the face data set includes a plurality of face feature data;

[0034] A processing module is used to obtain an original population corresponding to the face data set according to the face data set, wherein the original population includes multiple feature individuals, and each of the multiple feature individuals contains multiple and equal amounts of face feature data; initialize the original population according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population; update the initialized population according to the fitness value of each feature individual in the initialized population to obtain an updated population; determine a target feature individual according to the fitness value of each feature individual in the updated population, and determine the face feature data corresponding to the target feature individual as the target face feature data.

[0035] An embodiment of the present invention further provides a computing device, comprising:

[0036] A memory for storing one or more programs;

[0037] One or more processors are used to execute the one or more programs to implement the method described above.

[0038] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a program, and when the program is executed by a processor, the method described above is implemented.

[0039] The above solution of the present invention includes at least the following beneficial effects:

[0040] The data processing method, device, equipment set and readable storage medium provided by the above-mentioned scheme of the present invention include: obtaining a face data set, the face data set including multiple face feature data; obtaining an original population corresponding to the face data set according to the face data set, the original population including multiple feature individuals, each of the multiple feature individuals including multiple and equal number of face feature data; initializing the original population according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population; updating the initialized population according to the fitness value of each feature individual in the initialized population to obtain an updated population; determining a target feature individual according to the fitness value of each feature individual in the updated population, and determining the face feature data corresponding to the target feature individual as the target face feature data, so as to identify and select the target face feature data in incomplete face data containing a large amount of noise, improve the accuracy and efficiency of recognition, and overcome the problems of low feature recognition accuracy, slow speed and poor robustness in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flow chart of a data processing method provided by an embodiment of the present invention;

[0042] Figure 2 It is a schematic diagram of the distribution state of chaotic values ​​corresponding to the original chaotic mapping algorithm and the preset chaotic mapping algorithm provided by an optional embodiment of the present invention;

[0043] Figure 3 It is a schematic diagram of optimization effects of different algorithms on different face recognition data sets provided by an optional embodiment of the present invention;

[0044] Figure 4 is a box plot of classification error rates after a face recognition data set is processed by different algorithms provided in an optional embodiment of the present invention;

[0045] Figure 5 is a feature subset box plot after a face recognition data set is processed by different algorithms provided in an optional embodiment of the present invention;

[0046] Figure 6 is a histogram of average running time of different algorithms provided by an optional embodiment of the present invention running 30 times on different face recognition data sets;

[0047] Figure 7 is a schematic block diagram of a data processing device provided by an embodiment of the present invention;

[0048] Figure 8 is a schematic block diagram of an electronic device provided by an embodiment of the present invention;

[0049] Fig. 9 is a schematic block diagram of a computing device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0051] In the following description, certain specific details are set forth for the purpose of illustrating the various disclosed embodiments to provide a thorough understanding of the various disclosed embodiments. However, those skilled in the relevant art will recognize that the embodiments may be practiced without one or more of these specific details. In other cases, well-known devices, structures, and techniques associated with the present application may not be shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

[0052] References throughout the specification to "one embodiment" or "an embodiment" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of "in one embodiment" or "in an embodiment" in various places throughout the specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any manner in one or more embodiments.

[0053] In the following description, in order to clearly show the structure and working mode of the present invention, many directional words will be used for description, but the words "front", "back", "left", "right", "outside", "inside", "outward", "inward", "up", "down", etc. should be understood as convenient terms and should not be understood as restrictive terms.

[0054] A data processing method, apparatus, device and readable storage medium proposed in an embodiment of the present invention are based on an improved binary grey wolf optimizer, and integrate quantum computing and multi-strategy with the binary grey wolf optimizer to identify and select target facial feature data in a facial data set in high-dimensional data classification to reduce the risk of making wrong choices.

[0055] Here, we first introduce the feature extraction problem:

[0056] set up is a samples and each sample has The FS feature extraction problem aims to extract Select from the features indivual( ) features in order to optimize the objective function When solving feature selection problems, it is often necessary to encode possible feature selection schemes in a common and intuitive way, such as using a binary string Encode the solution to the FS problem, where Represents a feature selection scheme, each binary string The length of is equal to the total number of features, and each bit in the string corresponds to a feature, where "1" indicates that the feature is selected, and "0" indicates that the feature is not selected. The binary string can uniquely describe a features (that is, the number of digits with the value "1" in the string is , which is described as shown in formula (1):

[0057] ; (1)

[0058] In the FS problem, represents the feature selection scheme, Indicates the selection features, otherwise not selected. For classification, Usually the error rate. The FS goal is to minimize , that is, find the best feature combination to minimize the classification error rate. The description is shown in formula (2):

[0059] ; (2)

[0060] Secondly, the original Gray Wolf Optimizer theory involved in this application is introduced:

[0061] Mirjalili et al. proposed the Gray Wolf Optimization (GWO) algorithm in 2014, which was inspired by the hierarchical behavior of gray wolf social groups and imitated , , , Four-level structure: The wolf is the leader, responsible for decision-making and strategy; Wolf Assist Wolves organize and hunt; Wolves are guards, protecting borders; There are many wolves, and they obey the commands of other wolves. To facilitate the application in optimization problems, GWO corresponds to the optimal solution, suboptimal solution and third optimal solution as , and Wolf, the remaining candidate solutions are Wolf. The hunting process simulates the process of top wolves searching, tracking and surrounding their prey, thereby collaboratively finding the optimal solution in the search space.

[0062] The hunting process is abstracted into three main steps in the gray wolf optimization algorithm; the first is the "encirclement" stage, which simulates the behavior of wolves gradually approaching their prey; followed by the "hunting" stage, in which the wolves work together to narrow the encirclement; and finally the "attack prey" stage, in which the wolves launch an attack and capture the prey.

[0063] (1) The first is the "encircling the prey" stage: the gray wolves in the wolf pack will disperse and use their own search capabilities to find and track the prey. Each wolf will update its own position based on the approximate location of the prey. Eventually, a siege is formed. The mathematical model of wolves surrounding prey can be expressed as formula (3) and formula (4):

[0064] ; (3)

[0065] ; (4)

[0066] in, represents the distance between the wolf and the prey, represents the position vector of the prey, represents the position vector of each wolf.

[0067] and is the correlation coefficient vector, which is calculated by the following formula (5):

[0068] ; (5)

[0069] ; (6)

[0070] and is a random vector between [0, 1].

[0071] ; (7)

[0072] It decreases linearly as the number of iterations goes from 2 to 0.

[0073] (2) The second is the "hunting" stage: After determining the location of the prey, , , Wolf Leadership The wolves surrounded the prey from all sides. Through coordination and strategic adjustments, the wolves successfully captured the prey. , , Wolves know the potential location of their prey. The wolf updates its position accordingly. The mathematical model is summarized in the following formula (8): Position update is based on the leader wolf information.

[0074] ; (8)

[0075] , , Obtained by the following formula:

[0076] ; (9)

[0077] ; (10)

[0078] ; (11)

[0079] , , Obtained by the following formula:

[0080] ; (12)

[0081] ; (13)

[0082] ; (14)

[0083] in , , Represents the search agent and , , The distance between , and Respectively , , The wolf's position. They are the three best solutions so far. Correlation coefficient , , Give , , The wolf is given random weights so that it does not approach the prey too quickly and avoids premature killing. In the actual hunting process, it acts as an obstacle. Is the search agent in It can be seen that the update of its position is mainly affected by , , The influence of these three wolves.

[0084] (3) The third stage is the "attack prey" stage: when the prey stops moving, the wolf will attack it. This process is achieved by reducing It can be seen from formula (5) that , , The value range is between [-2, 2]. The value of decreases, , , The fluctuation of When , it emphasizes that the gray wolf approaches and attacks prey, and exploits prey; when When hunting, gray wolves move away from prey and look for new prey, emphasizing exploration.

[0085] The original population involved in the following embodiments of the present invention can be regarded as the gray wolf population in the gray wolf algorithm, each characteristic individual can be regarded as a gray wolf individual in the gray wolf population, and the multiple facial feature data contained in each characteristic individual can be regarded as the hunting target of the gray wolf individual.

[0086] like Figure 1As shown, an embodiment of the present invention provides a data processing method, including:

[0087] Step 11, obtaining a face data set, where the face data set includes a plurality of face feature data;

[0088] Step 12, obtaining an original population corresponding to the face data set according to the face data set, wherein the original population includes a plurality of characteristic individuals, and each of the plurality of characteristic individuals includes a plurality of face characteristic data of the same quantity;

[0089] Step 13, initializing the original population according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population;

[0090] Step 14, updating the initialized population according to the fitness value of each characteristic individual in the initialized population, and obtaining an updated population;

[0091] Step 15, determining the target feature individual according to the fitness value of each feature individual in the updated population, and determining the facial feature data corresponding to the target feature individual as the target facial feature data.

[0092] In this embodiment, the initial population of the face data set is initialized according to a preset circular chaos mapping algorithm and a preset quantum algorithm to enhance the randomness, distribution uniformity and diversity of the characteristic individuals in the initial population; during the initialization process, the complexity and ergodicity of the circular chaos mapping are used to enhance the randomness and distribution uniformity of the initial individuals, and the probability characteristics of quantum theory are combined to further increase the diversity of the characteristic individuals in the data set, thereby ensuring the accuracy of subsequent feature recognition and selection; here, the number of characteristic individuals in the initialized population is the same as the number of characteristic individuals in the original population;

[0093] Furthermore, based on the fitness value of each characteristic individual in the initialization population, the initialization population is updated to adjust the exploration and development capabilities of the initialization population during the optimization process; here, the fitness value represents the classification error when performing feature classification;

[0094] Furthermore, the target feature individual is determined according to the fitness value of each feature individual in the updated population; here, the target feature individual may include the preferred feature individual in the updated population and the newly generated candidate feature individual, the preferred feature individual corresponds to the feature individual with the smallest fitness value, and the candidate feature individual corresponds to the replacement feature individual generated after eliminating the feature individual with the largest fitness value, so as to optimize the population structure of the updated population and obtain feature individuals with better fitness, and determine the preferred feature individual and the candidate feature individual as the target feature individual, and at the same time, determine the facial feature data corresponding to the target feature individual as the target facial feature data required for face model recognition, so as to ensure the accuracy and efficiency of recognition.

[0095] In an optional embodiment of the present invention, the above step 12 may include:

[0096] Step 121, evenly segment the face data set, and use each group of face feature data among the multiple groups of face feature data obtained after segmentation as a feature individual, and use the population composed of the multiple feature individuals as the original population corresponding to the face data set.

[0097] Here, the original population can be regarded as the gray wolf population in the gray wolf algorithm, each characteristic individual can be regarded as a gray wolf individual in the gray wolf population, and the multiple facial feature data contained in each characteristic individual can be regarded as the hunting target of the gray wolf individual.

[0098] In an optional embodiment of the present invention, the above step 13 may include:

[0099] Step 131, determining the initial position information corresponding to each characteristic individual in the original population according to a preset circular chaos mapping algorithm;

[0100] In this embodiment, the circular chaos mapping is an iterative function based on points on a circle, which can generate a pseudo-random sequence with chaotic characteristics. According to the preset circular chaos mapping algorithm, the characteristic individuals in the original population are mapped and iteratively processed and their corresponding initial position information is determined, which can effectively improve the problems of low coverage rate and low differences between individuals in the solution space; here, the preset circular chaos mapping algorithm is improved based on the original circular chaos mapping algorithm to enhance the diversity of characteristic individuals in the original population and enhance the optimization ability of the algorithm;

[0101] Here, the original circular chaos mapping algorithm is specifically expressed as shown in formula (15):

[0102] ; (15)

[0103] Among them, a and b are setting parameters, indicating the mapping range of characteristic individuals. represents the modulo operation, represents the chaotic map value, which is also the point of the next iteration (the initial position sequence corresponding to the characteristic individual of the next iteration mapping), Represents the point of the current iteration (the initial position sequence corresponding to the feature individuals mapped in the current iteration);

[0104] Furthermore, the above step 131 may include:

[0105] Step 1311, determining the chaos mapping value of each characteristic individual in the original population according to a preset circular chaos mapping algorithm;

[0106] Step 1312, determining the initial position information corresponding to each characteristic individual according to the chaotic mapping value.

[0107] In this embodiment, the original circular chaos mapping algorithm is improved to obtain a preset circular chaos mapping algorithm as shown in formula (16): Figure 2 As shown in Figure 1, compared with the application of the original circular chaotic map, the application of formula (16) achieves a significant optimization of the distribution state of the chaotic map value, showing a more uniform and balanced distribution feature:

[0108] ; (16)

[0109] Before iterative mapping, a random initial position is randomly selected for each characteristic individual according to the number of characteristic individuals in the original population. , and set the number of iterative mappings at the same time; further, according to the preset number of iterations and the random initial position corresponding to each characteristic individual, the chaotic mapping of the random initial position is iteratively processed through formula (16), and the chaotic sequence corresponding to each characteristic individual is obtained , , …, , where M represents the number of iterative mappings;

[0110] Furthermore, the chaotic sequence corresponding to each feature individual is mapped to the multidimensional search space of the problem. The mapping here can use linear changes, that is, the mapping transformation can be performed through formula (17), and the initial position information corresponding to each feature data can be determined. Specifically, it can be expressed as:

[0111] ;

[0112] in, Indicates The characteristic individual The initial position sequence in dimensions (i.e., it can represent the j-th facial feature data in the current feature individual); , Represent the lower and upper limits of the dimension respectively, i=1, 2, 3, ..., N, N represents the total number of characteristic individuals; here, the initial position sequence can be Binary encoding is performed, wherein 1 indicates that the current j-th facial feature data is selected, and 0 indicates that the current j-th facial feature data is not selected; specifically, binary encoding can be performed using the following formula:

[0113] ;

[0114] It should be known that when each initial position information in the initial position sequence is transformed to obtain the corresponding updated position information, the facial feature data represented by the updated position information and the facial feature data represented by the initial position information corresponding to the updated position information are the same facial feature data, then whether the facial feature data represented by the updated position information is selected should be consistent with whether the facial feature data represented by the corresponding initial position information is selected.

[0115] Step 132, transforming the initial position information according to a preset quantum algorithm to obtain updated position information corresponding to the initial position information;

[0116] In this embodiment, the preset quantum algorithm is quantum computing. It should be known that one of the core characteristics of quantum computing is that quantum bits exist in a superposition state. In this step, by operating a group of quantum bits, the diversity of the original population can be effectively expanded and optimized, thereby ensuring the accuracy of the subsequent determination of the target facial feature data.

[0117] In an optional embodiment of the present invention, the above step 132 may include:

[0118] Step 1321, determining the quantum space position corresponding to each characteristic individual according to the initial position information of each characteristic individual in the original population;

[0119] Step 1322, based on the preset quantum bits and the quantum space position corresponding to each characteristic individual, determine the updated position information corresponding to the initial position of each characteristic individual.

[0120] It should be known that in quantum space, a quantum bit is the smallest unit of information, and its form is diverse, including single quantum bits, double quantum bits, and multiple quantum bits. In the embodiments of this application, a single quantum bit is used for processing. A quantum bit has two basic states, representing the logical state and , general quantum states are usually represented by Greek letters, such as ; The quantum bit in quantum computing can be a superposition of two states, as described in formula (18):

[0121] ; (18)

[0122] in represents the quantum state of superposition, and are the amplitudes belonging to one of the probabilities, and and is a complex number, and its specific expression is as follows:

[0123] ; (19)

[0124] When a quantum bit in superposition is observed, its value changes to The probability of collapse to 0, or The probability of collapse to 1, where and is the probability amplitude. Based on these amplitudes, the quantum bit can be expressed as a matrix containing sine and cosine elements; the specific expression is as follows:

[0125] ; (20)

[0126] in, = , = ;

[0127] In this embodiment, quantum properties are integrated into the initialization process of the original population, and the probability amplitude (α and β) of the quantum bit is used to construct the position of different characteristic individuals; the state matrix of the quantum bit directly corresponds to the quantum space position of the characteristic individual, so the quantum space position of the characteristic individual and the state matrix have a corresponding relationship shown in formula (21):

[0128] ;(twenty one)

[0129] in, represents the cosine position of the characteristic individual, represents the sinusoidal position of the characteristic individual, is the rotation angle, ,and ;

[0130] Furthermore, the mapping relationship between the initial position information of each characteristic individual and the quantum space can be established through linear transformation. , whose range is , where the corresponding response solution is , the value range is , it can be converted by formula (22) and formula (23):

[0131] ;(twenty two)

[0132] ;(twenty three)

[0133] Here, by establishing a mapping relationship between the initial position information of each characteristic individual and the quantum space, a more diverse original population can be obtained. Formulas (22) and (23) are the updated position information corresponding to the initial position of each characteristic individual.

[0134] The conversion process of the above formula (22) and formula (23) is equivalent to the corresponding quantum state collapsing into a certain state observation, and then restoring the superposition state. The current state is assumed to be the position of sine or cosine. The inverse transformation is shown in formulas (24) and (25):

[0135] ;(twenty four)

[0136] ; (25)

[0137] In this embodiment, a dynamic quantum rotating gate is used to perform quantum state logic conversion. The specific expression of the dynamic quantum rotating gate is shown in formula (26):

[0138] ; (26)

[0139] when When , a dynamic quantum rotating gate is used to perform quantum state logic conversion, and the update process is as shown in formula (27):

[0140] . (27)

[0141] Step 133: Determine the initialization population corresponding to the original population according to the initial position information and the updated position information.

[0142] The initial position information of each characteristic individual in the original population corresponds to two updated position information (cosine position , Sine position ), in this embodiment, the characteristic individuals are screened according to the updated position information and the initial position information, thereby obtaining an initialized population;

[0143] Specifically, the above step 133 may include:

[0144] Step 1331, determining the characteristic individual corresponding to the updated position information as the updated characteristic individual, and the updated characteristic individual corresponds to each characteristic individual in the original population one by one;

[0145] Step 1332, determining the initialization population according to the fitness value of the updated characteristic individual and the fitness value of the characteristic individual in the original population corresponding to the current updated characteristic data.

[0146] In this embodiment, the updated cosine position Corresponding to a virtual first updated feature individual, the sine position Corresponding to a virtual second updated feature individual, the feature individual corresponding to each initial position information is compared with the first updated feature individual and the second updated feature individual corresponding to it (each feature individual has a virtual second updated feature individual and a virtual second updated feature individual corresponding to it), the best feature individual is screened out, and the set of the best feature individuals after screening all the feature individuals in the original population is used as the final initialization population; here, when screening the feature individual and the first updated feature individual and the second updated feature individual corresponding to it, a preset packaging algorithm can be used. Preferably, the preset packaging algorithm can be a packaging algorithm based on the KNN classifier to ensure the quality of the facial features selected by the feature individual; here, the number of neighbors can be set to 5 (K-5); in order to reduce the risk of overfitting, a sample data set consisting of the facial feature data contained in the feature individual and the first updated feature individual and the second updated feature individual corresponding to it can be cross-validated 10 times, that is, the sample data set is divided into 10 times and distributed between the training data and the test data, with a ratio of 9:1, so as to eliminate any deviation in the test data. In the packaging algorithm based on the KNN classifier, the fitness value is represented by the classification error, and the specific expression of the fitness function is shown in formula (31):

[0147] ; (31)

[0148] in, represents the fitness value, is the classification error rate of the kth run, which is specifically expressed as follows:

[0149] ; (32)

[0150] Furthermore, the feature individual with the smallest fitness value among the feature individual, the first updated feature individual corresponding to the current feature individual, and the second updated feature individual is screened out and used as the optimal feature individual to form an initialized population.

[0151] Initializing the original population is equivalent to expanding the number of each characteristic individual in the original population by 2 times, and selecting an optimal characteristic individual after the expansion to ensure the accuracy of subsequent face recognition. Therefore, the initialized population may include the original characteristic individuals in the original population, and may also include virtual first updated characteristic individuals and second updated characteristic individuals. It should be known that the number of characteristic individuals in the initialized population should be consistent with the number of characteristic individuals in the original population.

[0152] In an optional embodiment of the present invention, the above step 14 may include:

[0153] Step 141, according to the fitness value of each characteristic individual in the initialization population, a plurality of characteristic individuals in the initialization population are stratified to obtain a plurality of population layers of different levels;

[0154] Step 142, based on the level learning strategy, iteratively update the positions of the characteristic individuals in the population layers of multiple different levels according to a preset number of iterations, and obtain an updated population.

[0155] In this embodiment, the characteristic individuals in the initialization population are first arranged in ascending order of fitness values, and the characteristic individuals in the arranged initialization population are layered. Specifically, the number of layers can be calculated by formula (29) ( ):

[0156] ; (29)

[0157] Here, NL represents the number of layers, which decreases exponentially with the increase of the number of iterations, and its lower and upper limits are 2 and 30 respectively. The calculation result is As shown in formula (30):

[0158] ; (30)

[0159] in, Indicates different levels, N represents the total number of characteristic individuals in the initialization population, r represents the current number of iterations, and R represents the maximum number of iterations

[0160] Furthermore, the sorted population is divided into The characteristic individuals with the lowest fitness value are assigned to the highest level. Here, the higher the level, the smaller the corresponding index of the level, so is the highest level, and is the lowest level. The number of individuals in each level (i.e., the level size) is the same, and It means; it is obvious . Note that the entire population may not be evenly distributed.

[0161] Furthermore, higher-level feature individuals are randomly selected as , in order to retain the most instructive information in the optimal feature individuals and prevent them from being weakened, The feature individuals in the hierarchy update their positions according to formula (8), because The individual information in the hierarchy is the richest, and their information should be preserved and protected from being weakened; here, the level-based learning strategy is used to The positions of the center forwards at other levels outside the level are iteratively updated. Specifically, the mathematical model for position update is initially established as follows:

[0162] ; (33)

[0163] ; (34)

[0164] ; (35)

[0165] As the algorithm continues to iterate, Reduce, This means that the algorithm's development ability gradually increases, while its exploration ability gradually decreases. In order to better understand the method of this application, it is assumed that the initial feature population is divided into three levels (i.e. ),so It is the highest level. Is the lowest level. Each feature individual in the hierarchy updates its position according to formula (8), or Each feature individual in the hierarchy updates its position according to the mathematical model of position update mentioned above; according to the above process, Each characteristic individual in the hierarchy needs to be Randomly select a characteristic individual in the level ,but The updated position of each feature individual in the hierarchy is the position vector of the random feature individual corresponding to formula (34); Each characteristic individual in the hierarchy needs to be Level or Randomly select a characteristic individual in the level ,but The updated position of each feature individual in the hierarchy is the position vector of the random feature individual corresponding to formula (35);

[0166] In this embodiment, a level-based learning strategy performs iterative position update processing on characteristic individuals in multiple population layers of different levels according to a preset number of iterations, which can improve the efficiency and effectiveness of the corresponding algorithm of the present application.

[0167] In an optional embodiment of the present invention, the above step 15 may include:

[0168] Step 151, determining the characteristic individual with the largest fitness value and the characteristic individual with the smallest fitness value in the updated population after each round of position iterative update processing;

[0169] Step 152, determining a candidate feature individual corresponding to the feature individual with the largest fitness value according to the position information corresponding to the feature individual with the smallest fitness value and a preset number of iterations;

[0170] Step 153, determining the target feature individual from the candidate feature individuals and the remaining feature individuals after removing the feature individual with the largest fitness value in the updated population;

[0171] Step 154, determining the facial feature data corresponding to the target feature individual as the target facial feature data.

[0172] In this embodiment, the characteristic individual with the highest fitness value in the updated population is first identified and eliminated, and candidate characteristic individuals corresponding to the eliminated characteristic individuals with the highest fitness value are generated to fill the vacancies after elimination, so as to continuously push the overall performance of the population to a higher level while maintaining the diversity of the population;

[0173] Here, according to the position information corresponding to the feature individual with the smallest fitness value and the preset number of iterations, the candidate feature individual corresponding to the feature individual with the largest fitness value is determined. Specifically, the specific position information corresponding to the candidate feature individual can be determined by formula (36):

[0174] ; (36)

[0175] in, Indicates the location information corresponding to the candidate feature individual, and They represent the characteristic individual with the smallest fitness value and the randomly selected characteristic individual, represents the dynamic radius defined by formula (37):

[0176] ; (37)

[0177] in, Indicates the current iteration, Indicates the maximum number of iterations. Dynamic radius and random number Multiplication constitutes the elastic radius. Therefore, in the case of overall contraction, the generation range of newly generated candidate feature individuals still has a certain probability of expansion, thereby avoiding the aggregation of newly generated candidate feature individuals in the later stage. Therefore, the elastic radius determines the generation range of newly generated candidate feature individuals in different iteration stages, thereby achieving large-scale exploration in the early stage and small-scale development in the later stage;

[0178] After determining the candidate feature individuals, the candidate feature individuals and other feature individuals after excluding the feature individual with the highest fitness value in the updated population are used to determine the target feature individuals, and the facial feature data corresponding to the target feature individuals are determined as the target facial feature data to facilitate face recognition and ensure the accuracy and efficiency of face recognition.

[0179] When the method provided by the above embodiment of the present invention is applied to face recognition, it is compared with five latest feature extraction algorithms, namely, the first feature extraction algorithm of BMNABC, the second feature extraction algorithm of BMPA_TVSinV, the third feature extraction algorithm of sin_cos_bIAVOA, the fourth feature extraction algorithm of BGWOPSO, and the fifth feature extraction algorithm of BGWO; the results are shown in the attached figure. Figure 3 To Attachment Figure 6 As shown, specifically:

[0180] Figure 3 The optimization effects of the method provided by the above embodiment of the present invention and the other five algorithms on different face recognition data sets are demonstrated. Global performance: In all data sets, the method provided by the above embodiment of the present invention generally shows a faster convergence speed, especially in the first few iterations, the error rate of the method provided by the above embodiment of the present invention drops significantly, showing its strong initial optimization ability. Final convergence result: On each data set, the final error rate value of the method provided by the above embodiment of the present invention is significantly lower than that of other algorithms, which shows that the method has an advantage in accuracy. Compared with other algorithms: Compared with other algorithms, the error rate reduction curve of the method provided by the above embodiment of the present invention on different data sets is steeper, indicating that it more effectively reduces the error rate in each iteration. In summary, the method provided by the above embodiment of the present invention shows a higher convergence speed and a lower final error rate on these experimental data sets, verifying its high efficiency.

[0181] Figure 4 A box plot of the classification error rate is drawn. It can be seen from the box plot that the method provided by the above embodiment of the present invention shows lower median error, smaller error distribution range and fewer outliers on different data sets compared with other algorithms, indicating that it has significant stability and consistency in reducing the error rate, verifying the superiority of the method provided by the above embodiment of the present invention in the optimization process.

[0182] Figure 5 A box plot of feature subsets is drawn. From the box plot analysis, it can be seen that the feature subset size of the method provided by the above embodiment of the present invention on different data sets is significantly smaller than that of other algorithms, and its distribution is more concentrated, and the interquartile range is smaller, indicating that the algorithm is more stable in feature selection and the results are more consistent. The small interquartile range means that the feature subset size of the method provided by the above embodiment of the present invention fluctuates less in the data set and has higher robustness. In addition, the method provided by the above embodiment of the present invention also has relatively few outliers, further demonstrating its superior performance. These results show that the method provided by the above embodiment of the present invention can effectively reduce the size of feature subsets, improve feature extraction efficiency, and show the best effect in feature selection tasks.

[0183] Figure 6 It is a bar chart of the average running time of each algorithm running 30 times on each data set. From the comparison of the average running time, it can be seen that the method provided by the above embodiment of the present invention shows a short and stable running time on different data sets, which is better than most other algorithms, especially compared with sin_cos_bIAVOA, which greatly reduces the calculation time. This shows that the method provided by the above embodiment of the present invention not only has a high feature selection effect and a low error rate, but also maintains excellent performance in operation efficiency, which verifies its practicality and overall advantages in optimization tasks.

[0184] Furthermore, as shown in Tables 1 to 5 below, the classification error rates of different algorithms on 5 different face recognition data sets (ORL first data set, orlraws10P second data set, warpAR10P third data set, warpPIE10P fourth data set, Yale fifth data set) are compared (ranked first in bold). The method provided by the above embodiment of the present invention performs well on 5 different face data sets and shows strong feature selection performance. For small-scale data sets and high-dimensional complex data sets, the dimension can be effectively reduced while maintaining high classification accuracy. At the same time, the stability of the method provided by the above embodiment of the present invention is significantly better than other algorithms, and the performance fluctuation is small, which ensures the consistency of output on different data sets. In contrast, although algorithms such as BGWO and sin_cos_bIAVOA perform well on some data sets, the overall ranking fluctuates greatly and is unstable, which further highlights the superiority of the method provided by the above embodiment of the present invention. The third to last row (Mean) in the table represents the Friedman average ranking, and the average ranking is the lowest (1.733). The second to last row represents Friedman's final ranking. It can be seen from the table that the method provided by the above embodiment of the present invention ranks first. The last row represents the result of the Wilcoxon rank sum test, "+" represents a significant difference, "=" represents similarity, and "-" represents no significant difference. It can be seen that this shows that the method provided by the above embodiment of the present invention shows significant advantages in most comparisons, and the effect is better than other algorithms. In summary, the method provided by the above embodiment of the present invention has a significant difference in classification error rate with the algorithms involved in the comparison. The method provided by the above embodiment of the present invention shows excellent ability in minimizing classification errors, feature selection and stability, and is an ideal choice for classification tasks.

[0185] Table 1, comparison of error rates of different algorithms on the first ORL dataset (ranked first in bold)

[0186]

[0187] Table 2, comparison of error rates of different algorithms on the orlraws10P second dataset (ranked first in bold)

[0188]

[0189] Table 3, error rate comparison of different algorithms on the warpAR10P third dataset (ranked first in bold)

[0190]

[0191] Table 4, error rate comparison of different algorithms on the fourth data set of warpPIE10P (ranked first in bold)

[0192]

[0193] Table 5, comparison of error rates of different algorithms on the Yale fifth dataset (ranked first in bold)

[0194]

[0195] As shown in Tables 6 to 10 below, the feature subset size comparison of different algorithms on 5 face recognition data sets is shown (the first ranking is indicated in bold). The table shows the best feature size, intermediate feature size, worst feature size, average feature size and standard deviation of feature size of these methods. In order to highlight the best performing algorithm, Table 2 gives the best results under each evaluation index so that readers can quickly identify which algorithms perform best in terms of the effectiveness and stability of feature selection. It can be seen from the table that the method provided by the above embodiment of the present invention has achieved significant advantages in multiple evaluation dimensions, and all five evaluation indicators are low. Combined with the error evaluation results, it is verified that the method provided by the above embodiment of the present invention not only has a higher dimensionality reduction ability, but also shows stronger stability and effectiveness in feature selection. The third to last row (Mean) in the table represents the Friedman average ranking, and the average ranking is the lowest (1.587). The second to last row represents the Friedman final ranking. It can be seen from the table that the present invention ranks first. The last row represents the results of the Wilcoxon rank sum test. It can be seen that this shows that the method provided by the above embodiment of the present invention has shown significant advantages in all comparisons and is better than other algorithms. In summary, the method provided by the above embodiment of the present invention has significant differences in feature subset size compared with the algorithm involved.

[0196] Table 6, comparison of feature subset sizes of different algorithms on the ORL first dataset (ranked first in bold)

[0197]

[0198] Table 7, comparison of feature subset sizes of different algorithms on the orlraws10P second dataset (ranked first in bold)

[0199]

[0200] Table 8. Comparison of feature subset sizes of different algorithms on the warpAR10P third dataset (ranked first in bold)

[0201]

[0202] Table 9, comparison of feature subset sizes of different algorithms on the fourth warpPIE10P dataset (ranked first in bold)

[0203]

[0204] Table 10, comparison of feature subset sizes of different algorithms on the Yale fifth dataset (ranked first in bold)

[0205]

[0206] The original population of the acquired face data set is initialized by a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population; according to the fitness value of each feature individual in the initialized population, the initialized population is updated to obtain an updated population; according to the fitness value of each feature individual in the updated population, the target feature individual is determined, and the face feature data corresponding to the target feature individual is determined as the target face feature data, so as to improve the accuracy and efficiency of face feature data recognition.

[0207] like Figure 7 As shown, an embodiment of the present invention further provides a data processing device 70, comprising:

[0208] An acquisition module 71 is used to acquire a face data set, where the face data set includes a plurality of face feature data;

[0209] The processing module 72 is used to obtain an original population corresponding to the face data set according to the face data set, wherein the original population includes multiple feature individuals, and each of the multiple feature individuals contains multiple and equal amounts of face feature data; initialize the original population according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population, wherein the initialized population contains multiple feature individuals; update the initialized population according to the fitness value of each feature individual in the initialized population to obtain an updated population; determine the target feature individual according to the fitness value of each feature individual in the updated population, and determine the face feature data corresponding to the target feature individual as the target face feature data.

[0210] It should be noted that the device is a device corresponding to the above-mentioned data processing method, and all implementation methods in the above-mentioned method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0211] like Figure 8As shown, an embodiment of the present invention further provides an electronic device 50, including: a memory 51 for storing one or more computer programs; one or more processors 52 for executing one or more computer programs, and when the computer program is run by the processor, the data processing method described above is executed. All implementations in the above method embodiment are applicable to this embodiment, and the same technical effect can be achieved. The electronic device 50 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices. The components shown in the present invention, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required in the present invention.

[0212] like Fig. 9 As shown, the electronic device 50 is a computing device, or a computer system, which may include a CPU 501 (computing unit), which can perform various appropriate actions and processes according to a computer program stored in a ROM 502 (read-only memory) or a computer program loaded from a storage unit 508 into a random access RAM 503 (memory). In the RAM 503, various programs and data required for the operation of the device 500 may also be stored. The CPU 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An I / O interface 505 (input / output interface) is also connected to the bus 504.

[0213] A number of components in the electronic device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a disk, an optical disk, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0214] CPU 501 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of CPU 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. CPU 501 performs the various methods and processes described above. For example, in some embodiments, the data processing method 10 may be implemented as a computer software program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by CPU 501, one or more steps of the data processing method 10 described above may be performed. Alternatively, in other embodiments, CPU 501 may be configured to execute the data processing method 10 in any other appropriate manner (e.g., by means of firmware).

[0215] The embodiment of the present invention further provides a computer-readable storage medium storing instructions, which, when executed on a computer, enables the computer to execute the data processing method described above. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0216] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0217] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0218] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0219] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0220] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0221] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, disk or optical disk, etc., which can store program codes.

[0222] In addition, it should be pointed out that in the apparatus and method of the present invention, it is obvious that each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. Moreover, the steps of performing the above series of processing can naturally be performed in chronological order according to the order of description, but it is not necessary to perform them in chronological order, and some steps can be performed in parallel or independently of each other. For those of ordinary skill in the art, it is understandable that all or any steps or components of the method and apparatus of the present invention can be implemented in hardware, firmware, software or a combination thereof in any computing device (including processors, storage media, etc.) or a network of computing devices, which can be achieved by those of ordinary skill in the art using their basic programming skills after reading the description of the present invention.

[0223] Therefore, the purpose of the present invention can also be achieved by running a program or a group of programs on any computing device. The computing device can be a well-known general-purpose device. Therefore, the purpose of the present invention can also be achieved by simply providing a program product containing a program code for implementing a method or device. That is to say, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any well-known storage medium or any storage medium developed in the future. It should also be pointed out that in the device and method of the present invention, it is obvious that each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. In addition, the steps of performing the above-mentioned series of processing can naturally be performed in chronological order according to the order of description, but it is not necessary to perform them in chronological order. Some steps can be performed in parallel or independently of each other.

[0224] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A data processing method, characterized in that: include: Acquire a face data set, wherein the face data set includes a plurality of face feature data; According to the face data set, an original population corresponding to the face data set is obtained, wherein the original population includes a plurality of characteristic individuals, and each of the plurality of characteristic individuals includes a plurality of face characteristic data of the same quantity; Based on the binary grey wolf optimizer and according to a preset circular chaos mapping algorithm and a preset quantum algorithm, the original population is initialized to obtain an initialized population corresponding to the original population; According to the fitness value of each characteristic individual in the initialization population, the initialization population is updated to obtain an updated population; Determine a target feature individual according to the fitness value of each feature individual in the updated population, and determine the facial feature data corresponding to the target feature individual as the target facial feature data; The initialization process of the original population is performed according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population, including: Determining the initial position information corresponding to each characteristic individual in the original population according to the preset circular chaos mapping algorithm; Transform the initial position information according to the preset quantum algorithm to obtain updated position information corresponding to the initial position information; An initialization population corresponding to the original population is determined according to the initial position information and the updated position information.

2. The data processing method according to claim 1, characterized in that: According to the preset circular chaos mapping algorithm, determining the initial position information corresponding to each characteristic individual in the original population includes: Determining the chaotic mapping value of each characteristic individual in the original population according to the preset circular chaotic mapping algorithm; According to the chaotic map value, the initial position information corresponding to each characteristic individual is determined.

3. The data processing method according to claim 1, characterized in that: The initial position information is transformed according to the preset quantum algorithm to obtain updated position information corresponding to the initial position information, including: Determining the quantum space position corresponding to each characteristic individual according to the initial position information of each characteristic individual in the original population; According to the preset quantum bits and the quantum space position corresponding to each characteristic individual data, the updated position information corresponding to the initial position of each characteristic individual is determined.

4. The data processing method according to claim 1, characterized in that: Determining, according to the initial position information and the updated position information, an initialization population corresponding to the original population, comprising: Determine the characteristic individual corresponding to the updated position information as an updated characteristic individual, wherein the updated characteristic individual corresponds one-to-one to each characteristic individual in the original population; The initialized population is determined according to the fitness value of the updated characteristic individual and the fitness value of the characteristic individual in the original population corresponding to the currently updated characteristic individual.

5. The data processing method according to claim 1, characterized in that: According to the fitness value of each characteristic individual in the initialization population, the initialization population is updated to obtain an updated population, including: According to the fitness value of each characteristic individual in the initialization population, a plurality of characteristic individuals in the initialization population are subjected to stratification processing to obtain a plurality of population layers of different levels; The level-based learning strategy performs iterative position update processing on the characteristic individuals in the plurality of population layers of different levels according to a preset number of iterations, and obtains the updated population.

6. The data processing method according to claim 5, characterized in that: Determining a target feature individual according to the fitness value of each feature individual in the updated population, and determining the facial feature data corresponding to the target feature individual as the target facial feature data, including: Determine the characteristic individual with the largest fitness value and the characteristic individual with the smallest fitness value in the updated population in each round of position iterative update processing; Determine, according to the position information corresponding to the feature individual with the smallest fitness value and the preset number of iterations, a candidate feature individual corresponding to the feature individual with the largest fitness value; Determine a target feature individual from the candidate feature individuals and the remaining feature individuals after removing the feature individual with the largest fitness value in the updated population; The facial feature data corresponding to the target feature individual is determined as the target facial feature data.

7. A data processing device, characterized in that: include: An acquisition module is used to acquire a face data set, wherein the face data set includes a plurality of face feature data; A processing module, configured to obtain, based on the face data set, an original population corresponding to the face data set, wherein the original population includes a plurality of characteristic individuals, and each of the plurality of characteristic individuals includes a plurality of face characteristic data of the same quantity; Based on the binary gray wolf optimizer and according to the preset circular chaos mapping algorithm and the preset quantum algorithm, the original population is initialized to obtain an initialized population corresponding to the original population; according to the fitness value of each characteristic individual in the initialized population, the initialized population is updated to obtain an updated population; according to the fitness value of each characteristic individual in the updated population, a target characteristic individual is determined, and the facial feature data corresponding to the target characteristic individual is determined as the target facial feature data; The initialization process of the original population is performed according to a preset circular chaos mapping algorithm and a preset quantum algorithm to obtain an initialized population corresponding to the original population, including: Determining the initial position information corresponding to each characteristic individual in the original population according to the preset circular chaos mapping algorithm; Transform the initial position information according to the preset quantum algorithm to obtain updated position information corresponding to the initial position information; An initialization population corresponding to the original population is determined according to the initial position information and the updated position information.

8. A computing device, characterized in that include: A memory for storing one or more programs; One or more processors, configured to execute the one or more programs to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images