Overfitting mitigation method, system and device based on adversarial learning and data enhancement, and storage medium

Abstract

The invention discloses an overfitting relieving method, system and device based on adversarial learning and data enhancement and a storage medium, and the method comprises the steps: S1, carrying out the pre-training of a living body model according to a first training set, and obtaining a first classification result; s2, performing attribute identification on the first training set by using an attribute discriminator, enabling a condition generation network to generate a first attribute fuzzy sample according to the attribute identification, the random noise vector and the first classification result, and adding the first attribute fuzzy sample into the first training set to obtain a second training set; s3, training the living body model according to the second training set to obtain a second classification result; and S4, enabling the conditional generation network to generate a second attribute fuzzy sample according to the attribute identifier, the random noise vector and the second classification result by using an attribute discriminator, adding the second attribute fuzzy sample into a second training set, and repeatedly training the living body model until the proportion of the attribute identifier in the second classification result accords with a preset range.

Description

[0001] Technology Neighborhood

[0002] The present invention relates to the field of living body recognition technology, and in particular to an overfitting mitigation method, system, device and storage medium based on adversarial learning and data enhancement. Background Art

[0003] In the adversarial training of liveness recognition models, overfitting is an issue that requires special attention. Overfitting refers to the phenomenon that the model performs well on the training set, but the performance on the test set (or in actual applications) drops significantly, which usually means that the model's generalization ability is weak.

[0004] In adversarial training of liveness detection models, overfitting may manifest as follows: as training progresses, the adversarial robustness of the model on the training set (i.e., the model's ability to identify adversarial samples) continues to increase, but the adversarial robustness on the test set begins to decrease. At the same time, the classification accuracy of both may also continue to increase, but the accuracy improvement on the test set may not be as significant as that on the training set.

[0005] The main reasons for overfitting are as follows:

[0006] Too high model complexity: When the complexity of the model is much higher than the complexity of the problem, the model may learn the noise and unnecessary features in the training data, resulting in overfitting.

[0007] Insufficient training data: If the training data is insufficient to support the complexity of the model, the model may overfit to the limited information in the training data and fail to generalize to new data.

[0008] Impure data: Noise, incorrect labels, or outliers contained in the training data may cause the model to learn incorrect features.

[0009] The disclosure of the above background technology content is only used to assist in understanding the inventive concept and technical solution of the present invention. It does not necessarily belong to the prior art of this patent application. In the absence of clear evidence that the above content has been disclosed on the filing date of this patent application, the above background technology should not be used to evaluate the novelty and creativity of the present application. Summary of the invention

[0010] To this end, the present invention introduces an adversarial network into the living body model training process, classifies body markers according to the attribute discriminator, and generates attribute fuzzy samples through a conditional generation network. Through iterative training, the living body model filters out the influence of identity and obtains a more robust recognition effect.

[0011] In a first aspect, the present invention provides an overfitting mitigation method based on adversarial learning and data enhancement, characterized in that it includes:

[0012] Step S1: pre-training the living body model according to the first training set to obtain a first classification result;

[0013] Step S2: using an attribute discriminator to identify the attributes of the first training set, and using a conditional generation network to generate a first attribute fuzzy sample according to the attribute identification, a random noise vector and the first classification result, and adding the first attribute fuzzy sample to the first training set to obtain a second training set;

[0014] Step S3: training the living model according to the second training set to obtain a second classification result;

[0015] Step S4: Use the attribute discriminator to enable the conditional generation network to generate a second attribute fuzzy sample according to the attribute identifier, the random noise vector and the second classification result, and add the second attribute fuzzy sample to the second training set, and repeatedly train the living model until the proportion of the attribute identifier in the second classification result meets the preset range.

[0016] Optionally, the overfitting mitigation method based on adversarial learning and data enhancement is characterized in that there are both living objects and prostheses under the same attribute identification in the first training set.

[0017] Optionally, the overfitting mitigation method based on adversarial learning and data enhancement is characterized in that, when generating the first attribute fuzzy sample or the second attribute fuzzy sample, the attribute identifier is removed for the sample misclassified by the living model, and the label is kept unchanged to remove the influence of the attribute identifier on the living model.

[0018] Optionally, the overfitting mitigation method based on adversarial learning and data enhancement is characterized in that step S2 comprises:

[0019] Step S21: using an attribute discriminator to identify the first image in the first training set, marking images with the same attribute as the same attribute identifier, and marking images with different attributes as different attribute identifiers;

[0020] Step S22: input the wrong samples in the first classification result and the random noise vector into the conditional generation network to generate N first attribute fuzzy samples; when the attribute discriminator identifies the first attribute fuzzy samples, the attribute identifier cannot be identified; wherein N≥1;

[0021] Step S23: adding the generated fuzzy samples of the first attribute to the first training set to obtain a second training set.

[0022] Optionally, the overfitting mitigation method based on adversarial learning and data enhancement is applicable to a live face model, and is characterized in that step S22 comprises:

[0023] Step S221: For the wrong samples in the first classification result, obtain the key areas of the face; wherein the key areas of the face include at least one of the glasses, the nose, and the mouth;

[0024] Step S222: using a conditional generative network to remove one of the plurality of key face regions, and respectively generate a plurality of samples;

[0025] Step S223: inputting a random noise vector, and using the conditional generation network to modify or replace the key face area, so that the attribute discriminator cannot recognize the attribute identifier, and obtaining multiple samples.

[0026] Optionally, the overfitting mitigation method based on adversarial learning and data enhancement is applicable to a palm live model, and is characterized in that step S22 comprises:

[0027] Step S224: obtaining a palm or vein pattern for an erroneous sample in the first classification result;

[0028] Step S225: for the living body sample, at least partially replace the palm or vein pattern with other living body samples through a conditional generation network to generate multiple samples;

[0029] Step S226: For the prosthetic sample, a random noise vector is input into a conditional generation network to at least partially replace the palm or vein pattern to generate multiple samples.

[0030] Optionally, the overfitting mitigation method based on adversarial learning and data enhancement is characterized by further comprising:

[0031] Step S227: For the prosthetic sample, the palm or vein pattern is at least partially replaced with the living sample through a conditional generation network to generate multiple samples.

[0032] In a second aspect, the present invention provides an overfitting mitigation system based on adversarial learning and data enhancement, which is used to implement any of the above-mentioned overfitting mitigation methods based on adversarial learning and data enhancement, and is characterized by comprising:

[0033] A first training module, used to pre-train the living body model according to the first training set to obtain a first classification result;

[0034] a first generating module, configured to use an attribute discriminator to perform attribute identification on the first training set, and enable a conditional generating network to generate a first attribute fuzzy sample according to the attribute identification, a random noise vector and the first classification result, and add the first attribute fuzzy sample to the first training set to obtain a second training set;

[0035] A second training module, used for training the living model according to the second training set to obtain a second classification result;

[0036] The second generation module is used to use the attribute discriminator to enable the conditional generation network to generate a second attribute fuzzy sample according to the attribute identifier, the random noise vector and the second classification result, and add the second attribute fuzzy sample to the second training set, and repeatedly train the living model until the proportion of the attribute identifier in the second classification result meets a preset range.

[0037] In a third aspect, the present invention provides an overfitting mitigation device based on adversarial learning and data enhancement, characterized in that it includes:

[0038] processor;

[0039] a memory storing executable instructions of the processor;

[0040] The processor is configured to execute the steps of any of the aforementioned overfitting mitigation methods based on adversarial learning and data enhancement by executing the executable instructions.

[0041] In a fourth aspect, the present invention provides a computer-readable storage medium for storing a program, characterized in that when the program is executed, the steps of any of the aforementioned overfitting mitigation methods based on adversarial learning and data enhancement are implemented.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] The present invention introduces the adversarial network into the training of the living model, so that the training samples can be changed in a targeted manner according to the training situation. The deviations occurring during the training of the living model can be corrected in a timely manner, the stability of the training process can be improved, the success rate of the living model training can be improved, and the generalization ability of the training model can be improved.

[0044] The present invention utilizes an adversarial network to generate attribute fuzzy samples so that attribute identifiers are removed in a targeted manner, thereby enabling a living body model with good generalization ability to be trained with a small sample set. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without creative work. Other features, purposes and advantages of the present invention will become more obvious by reading the detailed description of the non-limiting embodiments with reference to the following drawings:

[0046] Figure 1 This is a flowchart of a method for alleviating overfitting based on adversarial learning and data enhancement in an embodiment of the present invention;

[0047] Figure 2 is a flow chart of steps for obtaining a second training set in an embodiment of the present invention;

[0048] Figure 3 This is a flow chart of steps for generating a fuzzy sample of a first attribute in an embodiment of the present invention;

[0049] Figure 4 This is another flowchart of steps for generating a fuzzy sample of a first attribute in an embodiment of the present invention;

[0050] Figure 5 This is another flowchart of steps for generating a fuzzy sample of a first attribute in an embodiment of the present invention;

[0051] Figure 6 It is a structural schematic diagram of an overfitting mitigation system based on adversarial learning and data enhancement in an embodiment of the present invention;

[0052] Figure 7 is a structural schematic diagram of an overfitting mitigation device based on adversarial learning and data enhancement in an embodiment of the present invention; and

[0053] Figure 8 Schematic diagram of the structure of a computer-readable storage medium in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several variations and improvements may be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0055] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention 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 interchanged where appropriate, so that the embodiments of the present invention described herein, for example, can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof 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.

[0056] An embodiment of the present invention provides an overfitting mitigation method based on adversarial learning and data enhancement, aiming to solve the problems existing in the prior art.

[0057] The following specific embodiments are used to describe in detail the technical solutions of the present invention and how the technical solutions of the present application solve the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0058] The present invention introduces an adversarial network into the living body model training process, classifies body markers according to the attribute discriminator, and generates attribute fuzzy samples through a conditional generation network. Through iterative training, the living body model filters out the influence of identity and obtains a more robust recognition effect.

[0059] Figure 1 FIG. 1 is a flowchart of a method for alleviating overfitting based on adversarial learning and data enhancement in an embodiment of the present invention. Figure 1 As shown, the steps of an overfitting mitigation method based on adversarial learning and data enhancement in an embodiment of the present invention include:

[0060] Step S1: Perform initial training on the living body model according to the first training set and obtain a first classification result.

[0061] In this step, there are both live bodies and prostheses under the same attribute identification in the first training set. Through pre-training, the live body model has certain learning and classification capabilities for the initial training data. The live body model is trained using the first training set (containing a certain amount of labeled data). During the training process, the live body model will learn the characteristics of the data and classify the data based on these characteristics. This step obtains the first classification result, that is, the preliminary classification of the data in the first training set by the live body model.

[0062] Step S2: Use an attribute discriminator to identify the attributes of the first training set, and use a conditional generation network to generate a first attribute fuzzy sample based on the attribute identification and the first classification result, and add the first attribute fuzzy sample to the first training set to obtain a second training set.

[0063] In this step, when generating the first attribute fuzzy sample, the attribute identifier is removed for the sample misclassified by the living model, and the label is kept unchanged to remove the influence of the attribute identifier on the living model. The attribute discriminator is used to identify the specific attributes of the data and mark the data with the corresponding attribute identifier. The conditional generation network is used to generate fuzzy samples with specific attributes based on the attribute identifier, the random noise vector and the first classification result of the living model. The attribute discriminator is used to identify the attributes of the data in the first training set. According to the attribute identifier, the random noise vector (used to increase the diversity of the data) and the first classification result, the conditional generation network generates a series of first attribute fuzzy samples. These first attribute fuzzy samples are added to the first training set to obtain a second training set. The second training set contains more data samples, and these samples have richer attributes and features, which helps the living model learn a more comprehensive data distribution.

[0064] Step S3: training the living body model according to the second training set to obtain a second classification result.

[0065] In this step, the in vivo model is trained using the second training set. During the training process, the in vivo model learns the features of the data in the second training set and classifies the data based on these features. This step obtains the second classification result, that is, the classification of the data in the second training set by the in vivo model.

[0066] Step S4: using the attribute discriminator to enable the conditional generation network to generate a second attribute fuzzy sample according to the attribute identifier and the second classification result, and adding the second attribute fuzzy sample to the second training set, and repeatedly training the living model until the proportion of the attribute identifier in the second classification result meets a preset range.

[0067] In this step, the attribute discriminator is used to attribute the data in the second training set. According to the attribute identification, the random noise vector and the second classification result, the conditional generation network generates a series of second attribute fuzzy samples. These second attribute fuzzy samples are added to the second training set to obtain an updated training set. The in vivo model is repeatedly trained, and the attribute discriminator is used to check whether the ratio of the attribute identification in the second classification result meets the preset range. If not, new attribute fuzzy samples are continuously generated and added to the training set until the ratio meets the requirements.

[0068] Through continuous iterative training, the classification ability of the living model is gradually improved, and it can better process data with different attributes. At the same time, due to the introduction of adversarial learning and data enhancement technology, this embodiment can alleviate the overfitting problem and improve the generalization ability of the model.

[0069] Figure 2 FIG. 1 is a flow chart of steps for obtaining a second training set in an embodiment of the present invention. Figure 2 As shown, in an embodiment of the present invention, a step of obtaining a second training set includes:

[0070] Step S21: using an attribute discriminator to identify the first image in the first training set, marking images with the same attribute as the same attribute identifier, and marking images with different attributes as different attribute identifiers.

[0071] In this step, accurate attribute identifiers are assigned to the images in the first training set. The first image in the first training set is identified using an attribute discriminator. Based on the identification results, images with the same attributes are marked with the same attribute identifier, while images with different attributes are marked with different attribute identifiers. Each image in the first training set is assigned a unique attribute identifier that reflects the specific attributes of the image.

[0072] Step S22: inputting the erroneous samples in the first classification result and the random noise vector into the conditional generation network to generate N first attribute fuzzy samples; when the attribute discriminator identifies the first attribute fuzzy samples, the attribute identifier cannot be identified.

[0073] In this step, erroneous samples are identified from the first classification results. These erroneous samples are input into the conditional generation network together with the random noise vector. The conditional generation network generates N first attribute fuzzy samples based on the input erroneous samples and random noise vector. These samples may be visually similar to the original erroneous samples, but more fuzzy in attributes, so that the attribute discriminator cannot accurately identify the attribute identity when trying to identify the attributes of these samples. The value of N should be greater than or equal to 1 to ensure that a sufficient number of fuzzy samples are generated to enrich the training set. This step generates a series of first attribute fuzzy samples that have uncertainty in attributes, which helps the living model learn more robust feature representations.

[0074] Step S23: adding the generated fuzzy samples of the first attribute to the first training set to obtain a second training set.

[0075] In this step, the generated first attribute fuzzy samples are added to the first training set to form a richer second training set. The second training set contains the original first training set data and the newly generated first attribute fuzzy samples. These newly added samples increase the diversity and complexity of the training set, which helps the living model learn a wider range of feature representations and more accurate classification capabilities during the training process.

[0076] Figure 3 FIG. 1 is a flow chart of steps for generating a fuzzy sample of a first attribute in an embodiment of the present invention. Figure 3 As shown, this embodiment is applicable to the training of a living face model. In the embodiment of the present invention, a step of generating a fuzzy sample of a first attribute includes:

[0077] Step S221: For the erroneous samples in the first classification result, obtain the key face area.

[0078] In this step, the key areas of the face include at least one of the glasses, nose, and mouth. Filter out the wrong samples from the first classification results. Use face detection and feature point positioning algorithms (such as Dlib, MTCNN, etc.) to locate the key areas of the face. These algorithms can automatically detect and mark the key parts of the face such as eyes, nose, mouth, etc. These key areas are extracted and prepared for processing in subsequent steps.

[0079] Step S222: using a conditional generative network to remove one of the plurality of key face regions, and respectively generate a plurality of samples.

[0080] In this step, samples without attribute identifiers are generated by removing one or more of the key areas of the face, so that the living model can reduce its reliance on attribute identifiers during training, thereby improving its generalization ability. For each misclassified sample, a conditional generative network model (CGN) is used to remove one or more extracted key areas. When removing key areas, it is necessary to ensure that the generated sample still visually maintains the basic structure and features of the face, but the attribute identifiers have been effectively removed or blurred. For each misclassified sample, the above operation is performed to generate multiple samples without attribute identifiers.

[0081] Step S223: inputting a random noise vector, and using the conditional generation network to modify or replace the key face area, so that the attribute discriminator cannot recognize the attribute identifier, and obtaining multiple samples.

[0082] In this step, a random noise vector is input to further generate samples without attribute identification by modifying or replacing the key areas of the face. Unlike step S222, here the key areas are not simply removed, but modified or replaced to eliminate the influence of the attribute identification. For each misclassified sample, a conditional generative network model is used to modify or replace its key areas. This can be achieved by generating a new key area image and fusing it into the original face image. When modifying or replacing the key area, it is necessary to ensure that the generated sample still visually maintains the coherence and authenticity of the face, and the attribute identification has been completely removed or replaced with meaningless information (random noise vector). For each misclassified sample, the above operation is performed to generate more samples without attribute identification. After the samples are generated, they are verified using an attribute discriminator to ensure that the attribute identification has been effectively removed, that is, the attribute discriminator cannot recognize the attribute identification in the sample. This can be achieved by inputting the sample into the attribute discriminator and checking whether its output meets a certain threshold condition.

[0083] This embodiment helps to improve the generalization ability of the model and reduce the occurrence of overfitting.

[0084] Figure 4 FIG. 1 is another flow chart of steps for generating a fuzzy sample of the first attribute in an embodiment of the present invention. Figure 4 As shown, this embodiment is applicable to the training of a palm live model. Another step of generating a first attribute fuzzy sample in the embodiment of the present invention includes:

[0085] Step S224: For the erroneous samples in the first classification result, a palm or vein pattern is obtained.

[0086] In this step, palm or vein patterns are extracted from the misclassified samples. These patterns are key features in palm liveness recognition. By replacing or modifying in subsequent steps, samples for adversarial training can be generated. First, the wrong samples are screened out from the first classification results. These samples may be live objects that are mistakenly identified as prostheses, or prostheses that are mistakenly identified as live objects. Use image processing techniques (such as image segmentation, feature extraction, etc.) to extract palm or vein patterns from the misclassified samples. This can be achieved using deep learning models or traditional image processing algorithms.

[0087] Step S225: For the living body sample, the palm or vein pattern is at least partially replaced with other living body samples through a conditional generation network to generate multiple samples.

[0088] In this step, multiple new adversarial samples are generated by replacing the palm or vein patterns in the live samples. These samples will be used to train the palm live model to improve its recognition and generalization capabilities of live features. For each misclassified live sample, a conditional generative network model (such as GAN) is used to select part of the palm or vein pattern from other live samples for replacement. When replacing, it is necessary to ensure that the generated sample still visually maintains the basic structure and features of the palm while introducing the diversity of other live samples. Repeat this process to generate multiple adversarial samples for each misclassified live sample.

[0089] Step S226: For the prosthetic sample, a random noise vector is input into a conditional generation network to at least partially replace the palm or vein pattern to generate multiple samples.

[0090] In this step, multiple new adversarial samples are generated by replacing the palm or vein patterns in the prosthetic samples. These samples will be used to improve the recognition and generalization capabilities of the palm live model for prosthetic features. For each misclassified prosthetic sample, a random noise vector is used as input and the conditional generation network model is used to generate replacement content for replacement. This process is repeated to generate multiple adversarial samples for each misclassified prosthetic sample.

[0091] Figure 5 FIG. 1 is another flow chart of steps for generating a fuzzy sample of the first attribute in an embodiment of the present invention. Figure 5 As shown, compared with the above-mentioned embodiment, another step of generating a first attribute fuzzy sample in the embodiment of the present invention further includes:

[0092] Step S227: For the prosthetic sample, the palm or vein pattern is at least partially replaced with the living sample through a conditional generation network to generate multiple samples.

[0093] In this step, parts of the palm or vein pattern of the live sample are used to replace the corresponding parts in the prosthetic sample to generate multiple new adversarial samples. These samples will be used to train the palm live model to better identify those prosthetic samples that may be disguised as live. For each misclassified prosthetic sample, a conditional generation network model is used to select part of the palm or vein pattern from the live sample library for replacement. During the replacement process, it is necessary to ensure that the generated sample has a certain degree of visual authenticity, that is, the replaced part maintains a certain degree of coordination with the other parts of the prosthetic sample in terms of texture, color, etc. At the same time, the replaced part should contain enough information to simulate the live features so that the model can learn the differences between these features. Repeat this process to generate multiple adversarial samples for each misclassified prosthetic sample.

[0094] The adversarial samples generated by steps S224 to S227 can greatly enrich the training set of the palm live model, thereby improving the model's ability to identify live and prosthetic samples. In particular, the introduction of step S227 enables the model to better learn the subtle differences between live and prosthetic samples, reducing misjudgments caused by camouflage.

[0095] Figure 6 FIG. 1 is a schematic diagram of the structure of an overfitting mitigation system based on adversarial learning and data enhancement in an embodiment of the present invention. Figure 6 As shown, an overfitting mitigation system based on adversarial learning and data enhancement in an embodiment of the present invention includes:

[0096] A first training module, used to pre-train the living body model according to the first training set to obtain a first classification result;

[0097] a first generating module, configured to use an attribute discriminator to perform attribute identification on the first training set, and enable a conditional generating network to generate a first attribute fuzzy sample according to the attribute identification, a random noise vector and the first classification result, and add the first attribute fuzzy sample to the first training set to obtain a second training set;

[0098] A second training module, used for training the living model according to the second training set to obtain a second classification result;

[0099] The second generation module is used to use the attribute discriminator to enable the conditional generation network to generate a second attribute fuzzy sample according to the attribute identifier, the random noise vector and the second classification result, and add the second attribute fuzzy sample to the second training set, and repeatedly train the living model until the proportion of the attribute identifier in the second classification result meets a preset range.

[0100] This embodiment introduces an adversarial network into the living body model training process, classifies body markers according to the attribute discriminator, and generates attribute fuzzy samples through a conditional generation network. Through iterative training, the living body model filters out the influence of identity and obtains a more robust recognition effect.

[0101] An embodiment of the present invention also provides an overfitting mitigation device based on adversarial learning and data enhancement, comprising a processor, a memory, in which executable instructions of the processor are stored, wherein the processor is configured to execute the steps of an overfitting mitigation method based on adversarial learning and data enhancement by executing the executable instructions.

[0102] As mentioned above, this embodiment introduces the adversarial network into the living body model training process, classifies the body markers according to the attribute discriminator, and generates attribute fuzzy samples through the conditional generation network. Through iterative training, the living body model filters out the influence of identity and obtains a more robust recognition effect.

[0103] It will be appreciated by those skilled in the art that various aspects of the present invention may be implemented as systems, methods or program products. Therefore, various aspects of the present invention may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, which may be collectively referred to herein as "circuits", "modules" or "platforms".

[0104] Figure 7 Schematic diagram of the structure of an overfitting mitigation device based on adversarial learning and data enhancement in an embodiment of the present invention. Figure 7 The electronic device 600 according to this embodiment of the present invention is described. Figure 7 The electronic device 600 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0105] like Figure 7 As shown, the electronic device 600 is in the form of a general computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including the storage unit 620 and the processing unit 610), a display unit 640, etc.

[0106] The storage unit stores program codes, which can be executed by the processing unit 610, so that the processing unit 610 performs the steps of various exemplary embodiments of the present invention described in the above-mentioned overfitting mitigation method based on adversarial learning and data enhancement. For example, the processing unit 610 can perform the following steps: Figure 1 Follow the steps shown in .

[0107] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache memory unit 6202 , and may further include a read-only memory unit (ROM) 6203 .

[0108] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a grid environment.

[0109] Bus 630 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0110] The electronic device 600 may also communicate with one or more external devices 700 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device 600, and / or any device that enables the electronic device 600 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed through an input / output (I / O) interface 650. Furthermore, the electronic device 600 may also communicate with one or more grids (e.g., a local area network (LAN), a wide area network (WAN), and / or a public grid, such as the Internet) through a grid adapter 660. The grid adapter 660 may communicate with other modules of the electronic device 600 through the bus 630. It should be understood that although Figure 7 Not shown, other hardware and / or software modules may be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms.

[0111] In an embodiment of the present invention, a computer-readable storage medium is also provided for storing a program, and when the program is executed, the steps of an overfitting mitigation method based on adversarial learning and data enhancement are implemented. In some possible implementations, various aspects of the present invention can also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps of various exemplary embodiments of the present invention described in the above-mentioned overfitting mitigation method based on adversarial learning and data enhancement section of this specification.

[0112] As shown above, this embodiment introduces an adversarial network into the living body model training process, classifies body markers according to the attribute discriminator, and generates attribute fuzzy samples through a conditional generation network. Through iterative training, the living body model filters out the influence of identity and obtains a more robust recognition effect.

[0113] Figure 8 Schematic diagram of the structure of a computer-readable storage medium in an embodiment of the present invention. Figure 8As shown, a program product 800 for implementing the above method according to an embodiment of the present invention is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0114] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0115] Computer readable storage media may include data signals propagated in baseband or as part of a carrier wave, wherein readable program codes are carried. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program codes contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0116] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of grid, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0117] This embodiment introduces an adversarial network into the living body model training process, classifies body markers according to the attribute discriminator, and generates attribute fuzzy samples through a conditional generation network. Through iterative training, the living body model filters out the influence of identity and obtains a more robust recognition effect.

[0118] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the embodiments can be referred to each other. The above description of the disclosed embodiments enables professionals and technicians in this field to implement or use the present invention. The various modifications to these embodiments will be obvious to professionals and technicians in this field, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown in this article, but will comply with the widest range consistent with the principles and novel features disclosed herein.

[0119] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention.

Claims

1. An overfitting mitigation method based on adversarial learning and data enhancement, characterized in that: include: Step S1: pre-training the living body model according to the first training set to obtain a first classification result; Step S2: using an attribute discriminator to identify the attributes of the first training set, and using a conditional generation network to generate a first attribute fuzzy sample according to the attribute identification, a random noise vector and the first classification result, and adding the first attribute fuzzy sample to the first training set to obtain a second training set; Step S3: training the living model according to the second training set to obtain a second classification result; Step S4: Use the attribute discriminator to enable the conditional generation network to generate a second attribute fuzzy sample according to the attribute identifier, the random noise vector and the second classification result, and add the second attribute fuzzy sample to the second training set, and repeatedly train the living model until the proportion of the attribute identifier in the second classification result meets the preset range.

2. According to claim 1, the overfitting mitigation method based on adversarial learning and data enhancement is characterized in that: In the first training set, there are both living bodies and prosthetic bodies under the same attribute identification.

3. According to claim 1, the overfitting mitigation method based on adversarial learning and data enhancement is characterized in that: When generating the first attribute fuzzy sample or the second attribute fuzzy sample, the attribute identifier is removed for the sample misclassified by the living body model, and the label is kept unchanged to remove the influence of the attribute identifier on the living body model.

4. According to claim 1, the overfitting mitigation method based on adversarial learning and data enhancement is characterized in that: Step S2 includes: Step S21: using an attribute discriminator to identify the first image in the first training set, marking images with the same attribute as the same attribute identifier, and marking images with different attributes as different attribute identifiers; Step S22: input the wrong samples in the first classification result and the random noise vector into the conditional generation network to generate N first attribute fuzzy samples; when the attribute discriminator identifies the first attribute fuzzy samples, the attribute identifier cannot be identified; wherein N≥1; Step S23: adding the generated fuzzy samples of the first attribute to the first training set to obtain a second training set.

5. The overfitting mitigation method based on adversarial learning and data enhancement according to claim 4 is applicable to a living face model, characterized in that: Step S22 includes: Step S221: For the wrong samples in the first classification result, obtain the key areas of the face; wherein the key areas of the face include at least one of the glasses, the nose, and the mouth; Step S222: using a conditional generative network to remove one of the plurality of key face regions, and respectively generate a plurality of samples; Step S223: inputting a random noise vector, and using the conditional generation network to modify or replace the key face area, so that the attribute discriminator cannot recognize the attribute identifier, and obtaining multiple samples.

6. The overfitting mitigation method based on adversarial learning and data enhancement according to claim 4 is applicable to a palm live model, characterized in that: Step S22 includes: Step S224: obtaining a palm or vein pattern for an erroneous sample in the first classification result; Step S225: for the living body sample, at least partially replace the palm or vein pattern with other living body samples through a conditional generation network to generate multiple samples; Step S226: For the prosthetic sample, a random noise vector is input into a conditional generation network to at least partially replace the palm or vein pattern to generate multiple samples.

7. The overfitting mitigation method based on adversarial learning and data enhancement according to claim 6, characterized in that: Also includes: Step S227: For the prosthetic sample, the palm or vein pattern is at least partially replaced with the living sample through a conditional generation network to generate multiple samples.

8. An overfitting mitigation system based on adversarial learning and data enhancement, used to implement the overfitting mitigation method based on adversarial learning and data enhancement as claimed in any one of claims 1 to 7, characterized in that: include: A first training module, used to pre-train the living body model according to the first training set to obtain a first classification result; a first generating module, configured to use an attribute discriminator to perform attribute identification on the first training set, and enable a conditional generating network to generate a first attribute fuzzy sample according to the attribute identification, a random noise vector and the first classification result, and add the first attribute fuzzy sample to the first training set to obtain a second training set; A second training module, used for training the living body model according to the second training set to obtain a second classification result; The second generation module is used to use the attribute discriminator to enable the conditional generation network to generate a second attribute fuzzy sample according to the attribute identifier, the random noise vector and the second classification result, and add the second attribute fuzzy sample to the second training set, and repeatedly train the living model until the proportion of the attribute identifier in the second classification result meets a preset range.

9. An overfitting mitigation device based on adversarial learning and data enhancement, characterized in that: include: processor; a memory storing executable instructions of the processor; Wherein, the processor is configured to perform the steps of the overfitting mitigation method based on adversarial learning and data enhancement as described in any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium for storing a program, characterized in that: When the program is executed, the steps of the overfitting mitigation method based on adversarial learning and data enhancement described in any one of claims 1 to 7 are implemented.

Images