Regression analysis convolutional neural network behavior analysis method, apparatus and device, and medium

Abstract

The method is mainly applied to the technical field of image processing. The invention discloses a regression analysis convolutional neural network behavior analysis method, device and equipment and a medium, and the method comprises the steps: firstly, inputting an input image into a regression analysis convolutional neural network, and obtaining an expected output; then, a population matrix is created in a random initialization mode, and each individual is a binary filter; and updating the population matrix through preset iteration and screening conditions to determine an optimal filter. And then, processing the input image by using the optimal filter to obtain a target image, and inputting the target image into the regression analysis convolutional neural network again to obtain target output. And finally, by comparing the difference between the target image and the original image and the difference between the target output and the expected output, analyzing the prediction behavior of the convolutional neural network. And the relationship between the input value and the output value of the convolutional neural network is identified, regressed and analyzed, so that a user can understand the prediction basis of the neural network model.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a regression analysis convolutional neural network behavior analysis method, device, equipment and medium. Background Art

[0002] In the related technical field, due to the black box nature of deep convolutional neural networks, when a specific image is given as the input of the deep convolutional neural network, it is usually difficult to understand why the deep convolutional neural network outputs a specific prediction value, which in turn limits the application of deep convolutional neural networks in some high-risk tasks. Moreover, the interpretability of deep convolutional neural networks is usually only applicable to deep convolutional neural networks in classification tasks, but not to deep convolutional neural networks in regression analysis tasks. Summary of the invention

[0003] The present invention provides a regression analysis convolutional neural network behavior analysis method, device, equipment and medium, which helps users understand the prediction basis of the neural network model by identifying the relationship between the input value and the output value of the regression analysis convolutional neural network.

[0004] The present invention provides a regression analysis convolutional neural network behavior analysis method, the method comprising: Obtaining an input image and a regression analysis convolutional neural network to be analyzed, and inputting the input image into the regression analysis convolutional neural network for processing to obtain a desired output value; Creating a population matrix in a random initialization manner, wherein each individual element in the population matrix is ​​a binary filter for the regression analysis convolutional neural network behavior analysis; The population matrix is ​​updated based on a preset number of iterations, and the optimal element individual is determined from the updated population matrix according to a preset screening condition; Performing a convolution operation on the binary filter corresponding to the optimal element individual and the input image to obtain a target image, and inputting the target image into the regression analysis convolutional neural network for processing to obtain a target output value; Based on the first difference and the second difference, the relationship between the input value and the output value of the regression analysis convolutional neural network is determined to analyze the prediction behavior of the regression analysis convolutional neural network, wherein the first difference is the difference between the target image and the input image, and the second difference is the difference between the target output value and the expected output value.

[0005] Further, the updating of the population matrix based on a preset number of iterations and determining the optimal element individual from the updated population matrix according to a preset screening condition includes: Update the number and characteristics of individual elements in the current population matrix; Determine the fitness of each individual element in the updated current population matrix; Based on the fitness of each of the individual elements, a plurality of individual elements are selected from the current population matrix; Generate a new population matrix based on the multiple screened element individuals; The new population matrix is ​​used as the current population matrix, and the step of updating the number of individual elements and the characteristics of individual elements in the current population matrix is ​​returned; When the number of times the new population matrix is ​​generated is equal to the preset number of iterations, the new population matrix is ​​used as the updated population matrix, and the element individual with the best fitness in the updated population matrix is ​​used as the best element individual.

[0006] Further, the determination of the fitness of each individual element in the updated current population matrix includes: Performing convolution operations on the input image and each binary filter in the population matrix respectively to obtain multiple images after convolution operations; Inputting each image after the convolution operation into the regression analysis convolutional neural network for processing to obtain an actual output value corresponding to each image after the convolution operation; The expected output value is compared with each of the actual output values ​​to determine the absolute error value corresponding to each of the binary filters, wherein the absolute error value is used as the fitness of the individual element.

[0007] Further, based on the fitness of each of the individual elements, a plurality of individual elements are screened out from the current population matrix, including: Acquire a first candidate element individual and a second candidate element individual from the current population matrix; Determine whether the fitness of the first candidate element individual is greater than the fitness of the second candidate element individual, if so, use the second candidate element individual as the screened element individual, if not, use the first candidate element individual as the screened element individual; Return to the step of obtaining the first candidate element individual and the second candidate element individual from the current population matrix until the number of the screened element individuals is equal to a preset number threshold.

[0008] Furthermore, the updating of the number of individual elements and the characteristics of individual elements in the current population matrix includes: Dividing the current population matrix into a plurality of element groups, each of the element groups includes a first element individual and a second element individual; Based on the first preset probability value, determine a target element group from the multiple element groups, and generate multiple third element individuals according to the characteristics of the first element individuals and the characteristics of the second element individuals in each of the target element groups; Based on the second preset probability value, performing a mutation operation on each of the third element individuals to obtain a new element individual; The updated current population matrix is ​​generated based on the new element individuals.

[0009] Further, the determining of a target element group from the plurality of element groups based on the first preset probability value, and generating a plurality of third element individuals according to the characteristics of the first element individuals and the characteristics of the second element individuals in each of the target element groups, comprises: Based on a first preset probability value, determining a target element group from the plurality of element groups; The first individual element and the second individual element in the target element group are converted into a first vector and a second vector respectively, and the first vector and the second vector both include a plurality of parameters; At least one parameter in the first vector is exchanged with a target parameter in the second vector, and the first vector and the second vector after the parameter exchange are respectively converted into the third element individuals.

[0010] Further, based on the second preset probability value, performing a mutation operation on each of the third element individuals to obtain a new element individual includes: Convert the third element individual into a third vector, wherein the third vector includes a plurality of parameters; Based on the second preset probability value, the value of the target parameter among the multiple parameters is adjusted to obtain a third vector after parameter adjustment and convert it into the new element individual.

[0011] The present invention also provides a regression analysis convolutional neural network behavior analysis device, the device comprising: A first processing module is used to obtain an input image and a regression analysis convolutional neural network to be analyzed, and input the input image into the regression analysis convolutional neural network for processing to obtain a desired output value; A generation module, used for creating a population matrix in a random initialization manner, wherein each individual element in the population matrix is ​​a binary filter for the regression analysis convolutional neural network behavior analysis; An iteration module, used to update the population matrix based on a preset number of iterations, and determine the optimal element individual from the updated population matrix according to a preset screening condition; A second processing module is used to perform a convolution operation on the binary filter corresponding to the optimal element individual and the input image to obtain a target image, and input the target image into the regression analysis convolutional neural network for processing to obtain a target output value; A parsing module is used to determine the relationship between the input value and the output value of the regression analysis convolutional neural network based on a first difference and a second difference to parse the prediction behavior of the regression analysis convolutional neural network, wherein the first difference is the difference between the target image and the input image, and the second difference is the difference between the target output value and the expected output value.

[0012] The present invention also provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the regression analysis convolutional neural network behavior analysis method as described in any one of the above items is implemented.

[0013] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the regression analysis convolutional neural network behavior analysis method as described in any of the above items.

[0014] The present invention has at least the following beneficial effects: This technical solution analyzes the impact of the binary filter on the prediction output of the regression analysis convolutional neural network, wherein the first difference reveals the input image feature changes that cause the prediction change, and the second difference reveals how these input image feature changes affect the output of the regression analysis convolutional neural network. Analyzing the first difference and the second difference helps to reveal the working mechanism of the regression analysis convolutional neural network in the regression analysis task, thereby improving the interpretability of the model, and by converting the complex decision-making process of the regression analysis convolutional neural network into observable image feature changes, users can more intuitively understand how the model makes specific predictions, thereby helping users understand the prediction behavior of the regression analysis convolutional neural network. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings are used to provide a further understanding of the technical solution of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the technical solution of the present invention and do not constitute a limitation on the technical solution of the present invention.

[0016] Figure 1 It is a flow chart of the steps of the behavior analysis method of the convolutional neural network for regression analysis; Figure 2 is a flow chart of step S103 in the regression analysis convolutional neural network behavior analysis method; Figure 3is a flowchart of step S201 in the regression analysis convolutional neural network behavior analysis method; Figure 4 is a flowchart of step S202 in the regression analysis convolutional neural network behavior analysis method; Figure 5 is a flow chart of step S203 in the regression analysis convolutional neural network behavior analysis method; Figure 6 It is a schematic diagram of the structure of a regression analysis convolutional neural network behavior analysis device; Figure 7 It is a structural diagram of an electronic device. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] It should be noted that in the relevant technical field, deep convolutional neural networks have been widely used in image-based regression analysis tasks. However, due to the black box nature of deep convolutional neural networks, when a specific image is given as the input of the deep convolutional neural network, it is usually difficult to understand why the deep convolutional neural network outputs a specific prediction value, which in turn limits the application of deep convolutional neural networks in some high-risk tasks.

[0019] There are some techniques for interpreting deep convolutional neural networks, including techniques based on category activation maps and perturbation-based techniques. However, these techniques are mainly proposed for deep convolutional neural networks in classification tasks, while relatively few techniques are proposed for interpreting deep convolutional neural networks in regression analysis tasks. There is an essential difference between deep convolutional neural networks for classification tasks and regression analysis tasks. The output of deep convolutional neural networks for classification tasks is discrete category labels, while the output of deep convolutional neural networks for regression analysis tasks is continuous numerical values.

[0020] In order to help users understand the prediction behavior of deep convolutional neural networks for regression analysis tasks, the technical solution of this application provides a regression analysis convolutional neural network behavior analysis method, device, equipment and medium. The technical solution uses a genetic algorithm to evolve the interpretable filter of the regression analysis convolutional neural network (DCNN for short), so that when the result of the convolution operation of the evolved filter and the image is used as the input of the convolutional neural network, the convolutional neural network can output a specific value, thereby identifying the pattern in the image that makes the convolutional neural network output a specific prediction value, helping users understand why the convolutional neural network makes a specific prediction when given a specific input. A technical solution of a regression analysis convolutional neural network behavior analysis method, device, equipment and medium includes the following embodiments.

[0021] Please refer to Figure 1 , Figure 1 It is a flow chart of the steps of the regression analysis convolutional neural network behavior analysis method.

[0022] This embodiment provides a regression analysis convolutional neural network behavior analysis method including: S101, obtaining an input image and a regression analysis convolutional neural network to be analyzed, and inputting the input image into the regression analysis convolutional neural network for processing to obtain a desired output value.

[0023] S102. Create a population matrix in a random initialization manner, where each individual element in the population matrix is ​​a binary filter for regression analysis convolutional neural network behavior analysis.

[0024] S103, updating the population matrix based on a preset number of iterations, and determining the optimal element individual from the updated population matrix according to a preset screening condition.

[0025] S104, performing a convolution operation on the binary filter corresponding to the optimal individual element and the input image to obtain a target image, and inputting the target image into a regression analysis convolutional neural network for processing to obtain a target output value.

[0026] S105. Determine the relationship between the input value and the output value of the regression analysis convolutional neural network based on the first difference and the second difference to analyze the prediction behavior of the regression analysis convolutional neural network, wherein the first difference is the difference between the target image and the input image, and the second difference is the difference between the target output value and the expected output value.

[0027] In some embodiments, the binary filter for regression analysis convolutional neural network behavior analysis is a filter comprising The binary filter corresponding to the matrix of values.

[0028] In step S101 of some embodiments, an automated script may be used to retrieve an input image and a corresponding DCNN model from a database, or an input image may be uploaded through a user interface and a pre-trained DCNN model may be selected for analysis.

[0029] In step S102 of some embodiments, an initial population matrix is ​​randomly generated, wherein the filter parameter of each individual is a random binary value. Optionally, a filter that performs well in a previous optimization process is used as part of the initial population.

[0030] In step S103 of some embodiments, the population is updated using standard operations (selection, crossover, and mutation) of the genetic algorithm. Optionally, fitness ratio selection is introduced to preferentially select individuals with high fitness for crossover and mutation. In addition, the probability of crossover and mutation is dynamically adjusted during the iteration process to explore and develop different search areas.

[0031] In step S105 of some embodiments, the specific method of parsing the prediction behavior of the convolutional neural network through regression analysis includes: calculating the statistical data of the difference (such as mean, standard deviation) to analyze the prediction behavior, using machine learning methods to model the difference to predict the behavior of DCNN under different inputs, or displaying the difference through visualization tools to help users intuitively understand the prediction behavior of DCNN.

[0032] Please refer to Figure 2 , Figure 2 The present invention is a flowchart of step S103 in the regression analysis convolutional neural network behavior analysis method.

[0033] In some embodiments, step S103 includes: S201. Update the number and characteristics of individual elements in the current population matrix.

[0034] S202: Determine the fitness of each individual element in the updated current population matrix.

[0035] S203. Based on the fitness of each element individual, a plurality of selected element individuals are selected from the current population matrix.

[0036] S204: Generate a new population matrix based on the multiple screened element individuals.

[0037] S205: Use the new population matrix as the current population matrix and return to execute step S201.

[0038] S206. When the number of times the new population matrix is ​​generated is equal to the preset number of iterations, the new population matrix is ​​used as the updated population matrix, and the element individual with the best fitness in the updated population matrix is ​​used as the best element individual.

[0039] In a specific embodiment, in this specific embodiment, t=1, 2, 3, ..., T is set in turn, where T is the set maximum number of iterations. First, an initial population matrix is ​​generated, which contains multiple individual elements. Each individual element represents a binary filter. Optionally, by increasing the size of the population, or adjusting the size and shape of each filter, a wider feature space can be explored.

[0040] Next, each binary filter is applied to the input image and the resulting image is processed using DCNN. Then, the error between the image processed by each filter and the expected output is calculated. This error will be used as the fitness evaluation of each individual.

[0041] Then, based on the fitness evaluation, the top 10% of individuals with the smallest error are selected. These individuals are considered to be the optimal binary filters. The selected individuals will generate a new population matrix through crossover (combining the characteristics of two excellent individuals to generate new individuals) and mutation (randomly changing the characteristics of some individuals) operations. This process simulates natural selection and genetic variation in the hope of producing better binary filters. The newly generated population matrix will replace the old population matrix and repeat the above process. This process will continue until the preset number of iterations is reached. When the number of times a new population matrix is ​​generated is equal to the preset number of iterations, the new population matrix is ​​used as the updated population matrix, and the element individual with the best fitness in the updated population matrix is ​​used as the optimal element individual.

[0042] Finally, after all iterations are completed, the individual with the highest fitness, that is, the binary filter with the smallest error, is selected.

[0043] It can be understood that this embodiment optimizes the prediction performance of the deep convolutional neural network by iteratively updating and screening individuals in the population matrix, and ultimately finds the filter with the best parameter configuration to improve the accuracy and interpretability of the model.

[0044] Please refer to Figure 3 , Figure 3 The present invention is a flowchart of step S201 in the regression analysis convolutional neural network behavior analysis method.

[0045] In some embodiments, step S201 includes: S301, dividing the current population matrix into a plurality of element groups, each element group including a first element individual and a second element individual.

[0046] S302: Based on a first preset probability value, determine a target element group from a plurality of element groups, and generate a plurality of third element individuals according to characteristics of the first element individuals and characteristics of the second element individuals in each target element group.

[0047] S303: Based on the second preset probability value, perform a mutation operation on each third element individual to obtain a new element individual.

[0048] S304: Generate an updated current population matrix based on the new element individuals.

[0049] Specifically, the specific implementation of step S302 includes: Divide into order Yes, record it as ,in Indicates the rounding down operation. For each element group A crossover operation is performed with a first preset probability pc.

[0050] In some embodiments, the specific implementation of step S302 is: Based on a first preset probability value, a target element group is determined from multiple element groups; a first element individual and a second element individual in the target element group are respectively converted into a first vector and a second vector, and the first vector and the second vector both include multiple parameters; at least one parameter in the first vector is exchanged with a target parameter in the second vector, and the first vector and the second vector after the parameter exchange are respectively converted into a third element individual.

[0051] For example, the two element individuals in each element group (i.e., the first element individual and the second element individual) are reshaped into one Then, we randomly select a bit from the vector and swap the parts after the bit of the two elements to get the third element, and then reshape the two third elements back to filter.

[0052] In some embodiments, the specific implementation of step S302 is: The third element individual is converted into a third vector, wherein the third vector includes multiple parameters; based on the second preset probability value, the numerical value of the target parameter among the multiple parameters is adjusted to obtain the parameter-adjusted third vector and convert it into a new element individual.

[0053] For example, first randomly select L parameters (L≤ ), and then perform a mutation operation on each selected parameter with a second preset probability pm, that is, if the original value of a parameter is 0, it will become 1 with a probability of pm after the mutation operation; if the original value of a parameter is 1, it will become 0 with a probability of pm after the mutation operation.

[0054] It is understandable that this embodiment optimizes the binary filter by simulating the genetic evolution process. First, the population matrix is ​​divided into multiple element groups, each group containing two individuals. Then, a crossover operation is performed with a preset probability to generate a new third element individual. Next, these new individuals are mutated to increase the diversity of the population. By iteratively updating the population matrix, the individual with the highest fitness is finally screened out. This optimization method improves the performance of the filter and enhances the model's ability to recognize image features, thereby improving the prediction accuracy and interpretability of the deep convolutional neural network in regression analysis tasks. In addition, through automated evolutionary strategies, the solution also helps to reduce manual intervention and improve the efficiency of model training.

[0055] Please refer to Figure 4 , Figure 4 The present invention is a flowchart of step S202 in the regression analysis convolutional neural network behavior analysis method.

[0056] In some embodiments, step S202 includes: S401, performing convolution operations on an input image and each binary filter in a population matrix respectively, to obtain images after multiple convolution operations.

[0057] S402: Input each image after the convolution operation into the regression analysis convolutional neural network for processing to obtain an actual output value corresponding to each image after the convolution operation.

[0058] S403: Compare the expected output value with each actual output value respectively to determine the absolute error value corresponding to each binary filter, wherein the absolute error value is used as the fitness of the individual element.

[0059] In a specific embodiment, a population of N individuals is randomly initialized, denoted as ,in Represents the nth individual in the 0th generation population. Each individual is a binary filter with a value of 0 or 1. The fitness calculation process for each individual in the population is as follows: First, the individual is convolved with the input image, denoted as x∗f, and then the convolved image is used as the input of the deep convolutional neural network F(). Then, the output F(x∗f) is obtained through the deep convolutional neural network F(). Finally, the absolute error between the expected output value y and F(x∗f) is calculated: |y−F(x∗f)|. This absolute error is the fitness of individual f.

[0060] It can be understood that this embodiment convolves the input image with each binary filter, then inputs it into the convolutional neural network, obtains the actual output value, and compares it with the expected output value to calculate the absolute error fitness of each filter. This process realizes the quantitative evaluation of the filter performance, optimizes the filter to improve the network's recognition accuracy of image features, and thus improves the prediction accuracy of the model in the regression analysis task.

[0061] Please refer to Figure 5 , Figure 5 The present invention is a flowchart of step S203 in the regression analysis convolutional neural network behavior analysis method.

[0062] In some embodiments, step S203 includes: S501. Obtain a first candidate element individual and a second candidate element individual from the current population matrix.

[0063] S502: Determine whether the fitness of the first candidate element individual is greater than the fitness of the second candidate element individual. If so, use the second candidate element individual as the selected element individual. If not, use the first candidate element individual as the selected element individual.

[0064] S503, returning to step S501, until the number of individual elements after screening is equal to a preset number threshold.

[0065] In a specific embodiment, first, the current population matrix of the t-1 generation population is Randomly select two individuals and , as the first candidate element individual and the second candidate element individual, then, the first candidate element individual and the second candidate element individual If the fitness of The fitness is less than or equal to The fitness of is selected into the tth generation population, otherwise The selection process is repeated N times to ensure that the number of individuals in the t-th generation population is N. Finally, the current population matrix of the t-1th generation population is Select N individuals from the t-th generation population to obtain the new population matrix .

[0066] It can be understood that this embodiment achieves the survival of the fittest in the population by comparing the fitness of two individuals to be selected and selecting the individual with lower fitness as the element individual after screening. This process is repeated until the preset screening number threshold is reached, thereby ensuring that the individuals with the highest fitness are retained in the population. This method improves the overall quality of the population and promotes the improvement of model performance, especially in optimization problems, which helps to quickly converge to a better solution.

[0067] It is understandable that any of the above embodiments can analyze the effect of the binary filter on the prediction output of the regression analysis convolutional neural network, wherein the first difference reveals the input image feature changes that cause the prediction change, and the second difference reveals how these input image feature changes affect the output of the regression analysis convolutional neural network. Analyzing the first difference and the second difference helps to reveal the working mechanism of the regression analysis convolutional neural network in the regression analysis task, thereby improving the interpretability of the model, and by converting the complex decision-making process of the regression analysis convolutional neural network into observable image feature changes, users can more intuitively understand how the model makes specific predictions, thereby helping users understand the prediction behavior of the regression analysis convolutional neural network.

[0068] Please refer to Figure 6 , Figure 6 It is a structural schematic diagram of a regression analysis convolutional neural network behavior analysis device.

[0069] This embodiment provides a regression analysis convolutional neural network behavior analysis device comprising: The first processing module 601 is used to obtain an input image and a regression analysis convolutional neural network to be analyzed, and input the input image into the regression analysis convolutional neural network for processing to obtain a desired output value.

[0070] The generation module 602 is used to create a population matrix in a random initialization manner, and each element individual in the population matrix is ​​a binary filter for regression analysis convolutional neural network behavior analysis.

[0071] The iteration module 603 is used to update the population matrix based on a preset number of iterations, and determine the optimal element individual from the updated population matrix according to a preset screening condition.

[0072] The second processing module 604 is used to perform a convolution operation on the binary filter corresponding to the optimal element individual and the input image to obtain a target image, and input the target image into the regression analysis convolutional neural network for processing to obtain a target output value.

[0073] The analysis module 605 is used to determine the relationship between the input value and the output value of the regression analysis convolutional neural network based on the first difference and the second difference, so as to analyze the prediction behavior of the regression analysis convolutional neural network, wherein the first difference is the difference between the target image and the input image, and the second difference is the difference between the target output value and the expected output value.

[0074] It will be appreciated by those skilled in the art that all or some of the steps and devices in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. As is well known to those skilled in the art, communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0075] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0076] An embodiment of the present application also provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements any of the above regression analysis convolutional neural network behavior analysis methods.

[0077] refer to Figure 7 , Figure 7 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes: The processor 701 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application; The memory 702 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 702 can store operating devices and other applications. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 702, and the processor 701 calls and executes the regression analysis convolutional neural network behavior analysis method of the embodiment of this application; Input / output interface 703, used to implement information input and output; Communication interface 704, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.); A bus 705 that transmits information between various components of the device (e.g., the processor 701, the memory 702, the input / output interface 703, and the communication interface 704); The processor 701 , the memory 702 , the input / output interface 703 and the communication interface 704 are connected to each other in communication within the device via a bus 705 .

[0078] It can be understood that the contents of the above method embodiments are all applicable to the electronic device embodiment, the functions specifically implemented by the electronic device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0079] An embodiment of the present application also provides a computer-readable storage medium, which stores a program executable by a processor. When the program executable by the processor is executed by the processor, it is used to implement the regression analysis convolutional neural network behavior analysis method as described in any one of the above specific embodiments.

[0080] An embodiment of the present application also discloses a computer program product, including a computer program or computer instructions, wherein the computer program or computer instructions are stored in a computer-readable storage medium, a processor of a computer device reads the computer program or computer instructions from the computer-readable storage medium, and the processor executes the computer program or computer instructions, so that the computer device executes the regression analysis convolutional neural network behavior analysis method as described in any of the previous embodiments.

[0081] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0082] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein, for example. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0083] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. 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 an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0084] 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.

[0085] In addition, each functional unit in each embodiment of the present application 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. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0086] If the integrated unit 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 application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store program codes.

[0087] Although the description of the present application has been quite detailed and specifically describes several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, but should be regarded as providing a broad possible interpretation of these claims by reference to the attached claims, taking into account the prior art, so as to effectively cover the intended scope of the present application. In addition, the above description of the present application is based on the embodiments foreseeable by the inventor, and its purpose is to provide a useful description, and those non-substantial changes to the present application that have not yet been foreseen may still represent equivalent changes to the present application.

Claims

1. A regression analysis convolutional neural network behavior analysis method, characterized in that: The method comprises: Obtaining an input image and a regression analysis convolutional neural network to be analyzed, and inputting the input image into the regression analysis convolutional neural network for processing to obtain a desired output value; Creating a population matrix in a random initialization manner, wherein each individual element in the population matrix is ​​a binary filter for the regression analysis convolutional neural network behavior analysis; The population matrix is ​​updated based on a preset number of iterations, and the optimal element individual is determined from the updated population matrix according to a preset screening condition; Performing a convolution operation on the binary filter corresponding to the optimal element individual and the input image to obtain a target image, and inputting the target image into the regression analysis convolutional neural network for processing to obtain a target output value; Based on the first difference and the second difference, the relationship between the input value and the output value of the regression analysis convolutional neural network is determined to analyze the prediction behavior of the regression analysis convolutional neural network, wherein the first difference is the difference between the target image and the input image, and the second difference is the difference between the target output value and the expected output value.

2. The regression analysis convolutional neural network behavior analysis method according to claim 1, characterized in that: The updating of the population matrix based on a preset number of iterations and determining the optimal element individual from the updated population matrix according to a preset screening condition includes: Update the number and characteristics of individual elements in the current population matrix; Determine the fitness of each individual element in the updated current population matrix; Based on the fitness of each of the individual elements, a plurality of individual elements are selected from the current population matrix; Generate a new population matrix based on the multiple screened element individuals; The new population matrix is ​​used as the current population matrix, and the step of updating the number of individual elements and the characteristics of individual elements in the current population matrix is ​​returned; When the number of times the new population matrix is ​​generated is equal to the preset number of iterations, the new population matrix is ​​used as the updated population matrix, and the element individual with the best fitness in the updated population matrix is ​​used as the best element individual.

3. The regression analysis convolutional neural network behavior analysis method according to claim 2, characterized in that: The step of determining the fitness of each individual element in the updated current population matrix comprises: Performing convolution operations on the input image and each binary filter in the population matrix respectively to obtain multiple images after convolution operations; Inputting each image after the convolution operation into the regression analysis convolutional neural network for processing to obtain an actual output value corresponding to each image after the convolution operation; The expected output value is compared with each of the actual output values ​​to determine the absolute error value corresponding to each of the binary filters, wherein the absolute error value is used as the fitness of the individual element.

4. The regression analysis convolutional neural network behavior analysis method according to claim 2, characterized in that: The step of selecting a plurality of selected individual elements from the current population matrix based on the fitness of each individual element includes: Acquire a first candidate element individual and a second candidate element individual from the current population matrix; Determine whether the fitness of the first candidate element individual is greater than the fitness of the second candidate element individual, if so, use the second candidate element individual as the screened element individual, if not, use the first candidate element individual as the screened element individual; Return to the step of obtaining the first candidate element individual and the second candidate element individual from the current population matrix until the number of the screened element individuals is equal to a preset number threshold.

5. The regression analysis convolutional neural network behavior analysis method according to claim 2, characterized in that: The updating of the number of individual elements and the characteristics of individual elements in the current population matrix includes: Dividing the current population matrix into a plurality of element groups, each of the element groups includes a first element individual and a second element individual; Based on the first preset probability value, determine a target element group from the multiple element groups, and generate multiple third element individuals according to the characteristics of the first element individuals and the characteristics of the second element individuals in each of the target element groups; Based on the second preset probability value, performing a mutation operation on each of the third element individuals to obtain a new element individual; The updated current population matrix is ​​generated based on the new element individuals.

6. The regression analysis convolutional neural network behavior analysis method according to claim 5, characterized in that: The step of determining a target element group from the plurality of element groups based on the first preset probability value, and generating a plurality of third element individuals according to the characteristics of the first element individuals and the characteristics of the second element individuals in each of the target element groups, comprises: Based on a first preset probability value, determining a target element group from the plurality of element groups; The first individual element and the second individual element in the target element group are converted into a first vector and a second vector respectively, and the first vector and the second vector both include a plurality of parameters; At least one parameter in the first vector is exchanged with a target parameter in the second vector, and the first vector and the second vector after the parameter exchange are respectively converted into the third element individuals.

7. The regression analysis convolutional neural network behavior analysis method according to claim 5, characterized in that: The step of performing a mutation operation on each of the third element individuals based on the second preset probability value to obtain a new element individual includes: Convert the third element individual into a third vector, wherein the third vector includes a plurality of parameters; Based on the second preset probability value, the value of the target parameter among the multiple parameters is adjusted to obtain a third vector after parameter adjustment and convert it into the new element individual.

8. A regression analysis convolutional neural network behavior analysis device, characterized in that: The device comprises: A first processing module is used to obtain an input image and a regression analysis convolutional neural network to be analyzed, and input the input image into the regression analysis convolutional neural network for processing to obtain a desired output value; A generation module, used for creating a population matrix in a random initialization manner, wherein each individual element in the population matrix is ​​a binary filter for the regression analysis convolutional neural network behavior analysis; An iteration module, used to update the population matrix based on a preset number of iterations, and determine the optimal element individual from the updated population matrix according to a preset screening condition; A second processing module is used to perform a convolution operation on the binary filter corresponding to the optimal element individual and the input image to obtain a target image, and input the target image into the regression analysis convolutional neural network for processing to obtain a target output value; A parsing module is used to determine the relationship between the input value and the output value of the regression analysis convolutional neural network based on a first difference and a second difference to parse the prediction behavior of the regression analysis convolutional neural network, wherein the first difference is the difference between the target image and the input image, and the second difference is the difference between the target output value and the expected output value.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the regression analysis convolutional neural network behavior analysis method described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the regression analysis convolutional neural network behavior analysis method described in any one of claims 1 to 7 is implemented.

Images