A radar jamming identification hyperparameter smoothing optimization method without playback condition
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- XIDIAN UNIV
- Filing Date
- 2026-04-24
- Publication Date
- 2026-08-07
AI Technical Summary
该类方法中保存历史样本、构建联合验证集或增加新的可训练模块，不仅降低了历史样本中数据的安全性，还因存储和计算开销的增加而降低了部署效率，与真实电子对抗环境中的保密要求和快速部署要求不相适应
本发明提供的无回放条件下雷达干扰识别超参数平滑优化方法，通过获取当前增量轮对应的当前任务数据集及当前类别数量、历史增量轮对应的历史全局系数及历史类别数量；对当前任务数据集进行随机下采样，得到当前评估子集；在预设范围内，通过一维有界搜索算法确定当前评估子集对应的局部候选系数；基于历史类别数量与当前类别数量对历史全局系数及局部候选系数进行加权，得到当前增量轮对应的雷达干扰识别超参数。一方面，通过本发明提供的方法确定雷达干扰识别超参数的过程中，无需进行历史样本回放或追加门控网络训练，不仅可以提高历史样本中数据的安全性，还可以减少存储和计算开销，从而提高了部署效率；另一方面，确定雷达干扰识别超参数的过程中参考了历史全局系数，对历史全局系数中包含的旧类别知识进行了显式保留，可以避免新任务更新对旧任务决策边界造成冲击；再一方面，基于历史类别数量与当前类别数量对历史全局系数及局部候选系数进行加权，得到当前增量轮对应的雷达干扰识别超参数，该过程中递推加权融合的方式可以削弱单轮异常波动对雷达干扰识别超参数的影响，提高了雷达干扰识别超参数随场景演变过程中的平稳性。
Smart Images
Figure CN122525502A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of radar interference identification, specifically relating to a hyperparameter smoothing optimization method for radar interference identification under non-replay conditions. Background Technology
[0002] In tasks such as radar interference identification, open set modulation identification, and fault diagnosis, to balance continuous learning capability with the ability to retain old knowledge, an increasing number of identification systems are adopting multi-subspace, multi-branch, or multi-expert parallel inference architectures. Such architectures typically store knowledge learned at different stages in different feature subspaces or classification branches, and fuse the outputs of each subspace during the inference stage.
[0003] In the aforementioned radar interference identification system, the output of the subspace to which the category belongs is typically assigned a dominant weight, while the output of the subspace not belonging to the category is assigned a radar interference identification hyperparameter. The magnitude of this radar interference identification hyperparameter directly determines the shape of the final decision boundary: a hyperparameter that is too small will lead to insufficient utilization of complementary information across subspaces, while a hyperparameter that is too large will amplify noise in irrelevant subspaces and introduce misjudgments. Furthermore, the optimal radar interference identification hyperparameter exhibits significant fluctuations under different interference-to-noise ratios, tending to a smaller value under high signal-to-noise conditions and significantly increasing under strong noise conditions. This means that fixed empirical parameters are difficult to apply across different scenarios.
[0004] In related technologies, end-to-end recalibration of multi-subspace outputs is performed by replaying historical samples, using joint validation sets, or training an additional gating network to improve the environmental adaptability of radar jamming identification hyperparameters. However, preserving historical samples, constructing joint validation sets, or adding new trainable modules in these methods not only reduces the security of historical data but also decreases deployment efficiency due to increased storage and computational overhead, making them incompatible with the confidentiality and rapid deployment requirements of real electronic warfare environments. Summary of the Invention
[0005] To address the aforementioned problems in the existing technology, this invention provides a hyperparameter smoothing optimization method for radar interference identification under non-replay conditions. The technical problem to be solved by this invention is achieved through the following technical solution: In a first aspect, the present invention provides a method for smoothing and optimizing hyperparameters for radar interference identification under non-replay conditions, comprising: Get the current task dataset and number of categories corresponding to the current increment round, and the historical global coefficients and number of categories corresponding to the historical increment rounds; The current task dataset is randomly downsampled to obtain the current evaluation subset; Within a preset range, the local candidate coefficients corresponding to the current evaluation subset are determined by a one-dimensional bounded search algorithm. The historical global coefficients and the local candidate coefficients are weighted based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round.
[0006] In one embodiment of the present invention, determining the local candidate coefficients corresponding to the current evaluation subset using a one-dimensional bounded search algorithm includes: Multiple candidate coefficients are determined by particle swarm optimization algorithm, and the candidate coefficient that maximizes the value obtained by particle swarm optimization algorithm is taken as the local candidate coefficient.
[0007] In one embodiment of the present invention, the step of weighting the historical global coefficients and the local candidate coefficients based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round includes: When the current increment round number is 1, the local candidate coefficients are used as the radar interference identification hyperparameters; When the number of the current incremental rounds is greater than 1, the historical global coefficients and the local candidate coefficients are weighted based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters.
[0008] In one embodiment of the present invention, the method further includes: The radar interference identification hyperparameters for the next incremental round are determined based on the radar interference identification hyperparameters corresponding to the current incremental round and the local candidate coefficients corresponding to the evaluation subset of the next incremental round.
[0009] In one embodiment of the present invention, the method further includes: Obtain historical task datasets; Establish the correspondence between the sub-scores and categories of each subspace; each subspace includes the subspaces in the historical task dataset and the subspaces in the current task dataset; Based on the correspondence between the sub-scores and categories of each subspace, the sub-scores of the target subspace corresponding to each category are assigned unit weights. The sub-scores of other subspaces are weighted and attenuated using the radar interference identification hyperparameter to obtain the fusion scores of each category. The other subspaces are the subspaces other than the target subspace among the subspaces. The radar interference identification result is determined based on the fusion scores of the aforementioned categories.
[0010] In one embodiment of the present invention, establishing the correspondence between the sub-scores of each subspace and the categories includes: Obtain the categories in each subspace; The sub-scores for each subspace are determined based on the categories in each subspace and the representation of the target sample in each subspace; or, a classification head is output from the representation of the target sample in each subspace to obtain the sub-scores for each subspace.
[0011] In one embodiment of the present invention, determining the radar interference identification result based on the fusion scores of the various categories includes: The maximum value among the fusion scores of each category is taken as the radar interference identification result.
[0012] Secondly, the present invention provides a radar interference identification hyperparameter smoothing optimization device under non-replay conditions, comprising: The data acquisition module is used to acquire the current task dataset and the number of current categories corresponding to the current increment round, as well as the historical global coefficients and the number of historical categories corresponding to the historical increment rounds. The data sampling module is used to randomly downsample the current task dataset to obtain the current evaluation subset; The first determining module is used to determine the local candidate coefficients corresponding to the current evaluation subset within a preset range using a one-dimensional bounded search algorithm. The second determining module is used to weight the historical global coefficients and the local candidate coefficients based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round.
[0013] Thirdly, the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; The processor, when executing a program stored in memory, implements any of the methods provided in the first aspect.
[0014] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements any of the methods provided in the first aspect.
[0015] The beneficial effects of this invention are: The radar interference identification hyperparameter smoothing optimization method provided by this invention under no-replay conditions is as follows: The method obtains the current task dataset and current category count corresponding to the current increment round, and the historical global coefficients and historical category counts corresponding to the historical increment rounds; it then randomly downsamples the current task dataset to obtain the current evaluation subset; within a preset range, it determines the local candidate coefficients corresponding to the current evaluation subset using a one-dimensional bounded search algorithm; and finally, it weights the historical global coefficients and local candidate coefficients based on the historical category counts and the current category counts to obtain the radar interference identification hyperparameters corresponding to the current increment round. On the one hand, the method provided by this invention eliminates the need for historical sample replay or additional gating network training in determining radar interference identification hyperparameters. This not only improves the security of historical data but also reduces storage and computational overhead, thereby increasing deployment efficiency. On the other hand, the determination of radar interference identification hyperparameters references historical global coefficients and explicitly preserves the old category knowledge contained in the historical global coefficients, avoiding the impact of new task updates on the decision boundaries of old tasks. Furthermore, the historical global coefficients and local candidate coefficients are weighted based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round. The recursive weighted fusion method in this process weakens the impact of abnormal fluctuations in a single round on the radar interference identification hyperparameters, improving the stability of the radar interference identification hyperparameters as the scene evolves.
[0016] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0017] Figure 1 This is a schematic diagram illustrating the implementation process of a hyperparameter smoothing optimization method for radar interference identification under non-replay conditions provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a hyperparameter smoothing optimization device for radar interference identification under non-replay conditions provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the hardware entity of an electronic device to which embodiments of the present invention are applied. Detailed Implementation
[0018] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.
[0019] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating the implementation process of a hyperparameter smoothing optimization method for radar interference identification under no-replay conditions provided by an embodiment of the present invention. The hyperparameter smoothing optimization method for radar interference identification under no-replay conditions provided by this embodiment of the present invention may include the following steps: Step S101: Obtain the current task dataset and current category count corresponding to the current increment round, and the historical global coefficient and historical category count corresponding to the historical increment round.
[0020] The current task dataset contains the data corresponding to the current increment round, such as radar interference sample data of newly added categories. The current category number refers to the number of categories contained in the current increment round. The historical increment round refers to the previous increment round. The historical global coefficient refers to the radar interference identification hyperparameter used in the previous increment round. The historical category number refers to the total number of categories contained in all historical increment rounds.
[0021] In one possible implementation, if the current task dataset contains the number of current categories, then only the current task dataset corresponding to the current increment round, the historical global coefficients corresponding to the historical increment rounds, and the number of historical categories can be obtained to determine the radar interference identification hyperparameters based on the current task dataset, the historical global coefficients, and the number of historical categories.
[0022] Step S102: Randomly downsample the current task dataset to obtain the current evaluation subset.
[0023] The current evaluation subset refers to a lightweight evaluation subset constructed from the current task dataset. Random downsampling can be performed using methods such as random sampling, stratified sampling, or class-balanced sampling.
[0024] For example, the current task dataset can be randomly downsampled using methods such as random sampling, stratified sampling, or class-balanced sampling to obtain the current evaluation subset, which can then be used to determine the hyperparameters for subsequent radar interference identification.
[0025] Step S103: Within a preset range, determine the local candidate coefficients corresponding to the current evaluation subset using a one-dimensional bounded search algorithm.
[0026] The preset range can be the interval [0,1] or a preset physically feasible interval. One-dimensional bounded search algorithms can include grid search, golden section search, particle swarm optimization, Bayesian optimization, simulated annealing, genetic algorithms, or other derivativeless search algorithms.
[0027] In one possible implementation, multiple candidate coefficients can be determined using a particle swarm optimization algorithm, and the candidate coefficient that maximizes the value obtained by the particle swarm optimization algorithm can be taken as a local candidate coefficient.
[0028] Taking the current increment round as the nth increment round as an example, within the interval [0,1] or the preset physically feasible interval... Internally, the particle swarm search algorithm is used to evaluate the independent variables. Solve the problem to obtain local candidate coefficients. For specific implementation details, please refer to the following formulas (1) and (2): (1); (2); in, Represents local candidate coefficients. This represents the objective function corresponding to the particle swarm search algorithm. This represents a performance evaluation function, whose values can be accuracy, macro average F1, Kappa coefficient, or a weighted combination of accuracy, macro average F1, and Kappa coefficient. Indicates the current evaluation subset. Indicates the penalty weight. This represents a penalty term to suppress excessive participation by irrelevant subspaces. The corresponding values can be , Or other monotonically increasing functions.
[0029] Step S104: Weight the historical global coefficients and local candidate coefficients based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round.
[0030] In one possible implementation, when the current increment round number is 1, the local candidate coefficients are used as radar interference identification hyperparameters. When the current increment round number is greater than 1, the historical global coefficients and local candidate coefficients are weighted based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters. It is understood that when the current increment round number is 1, there are no historical increment rounds, so the local candidate coefficients corresponding to the current increment round can be directly used as radar interference identification hyperparameters. When the current increment round number is greater than 1, the historical global coefficients can be referenced to determine the radar interference identification hyperparameters.
[0031] For example, the radar interference identification hyperparameters corresponding to the current incremental wheel can be determined by the following formulas (3), (4) and (5): (3); (4); (5); in, Indicates the number of historical categories. Indicates the current number of categories. A collection representing historical categories. Represents the set of the current category. This represents the radar interference identification hyperparameter corresponding to the current increment wheel. Represents local candidate coefficients. This indicates the current increment round number. Represents the historical global coefficient.
[0032] In one possible implementation, the number of historical categories and the number of current categories in the above embodiment can be replaced by indicators that characterize the information scale of the historical increment and the current increment, such as the number of effective samples in the historical increment and the current increment, the prior weight of the categories, and the confidence-weighted number of categories.
[0033] In one possible implementation, boundary projection can be performed after updating the radar interference identification hyperparameters corresponding to the current increment wheel. This restricts the radar interference identification hyperparameters corresponding to the current increment wheel to a preset feasible range, thereby limiting the variation of the radar interference identification hyperparameters corresponding to the current increment wheel with respect to historical global coefficients and improving the stability of the radar interference identification hyperparameters as the scene evolves.
[0034] The radar interference identification hyperparameter smoothing optimization method provided by this invention under no-replay conditions is as follows: The method obtains the current task dataset and current category count corresponding to the current increment round, and the historical global coefficients and historical category counts corresponding to the historical increment rounds; it then randomly downsamples the current task dataset to obtain the current evaluation subset; within a preset range, it determines the local candidate coefficients corresponding to the current evaluation subset using a one-dimensional bounded search algorithm; and finally, it weights the historical global coefficients and local candidate coefficients based on the historical category counts and the current category counts to obtain the radar interference identification hyperparameters corresponding to the current increment round. On the one hand, the method provided by this invention eliminates the need for historical sample replay or additional gating network training in determining radar interference identification hyperparameters. This not only improves the security of historical data but also reduces storage and computational overhead, thereby increasing deployment efficiency. On the other hand, the determination of radar interference identification hyperparameters references historical global coefficients and explicitly preserves the old category knowledge contained in the historical global coefficients, avoiding the impact of new task updates on the decision boundaries of old tasks. Furthermore, the historical global coefficients and local candidate coefficients are weighted based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round. The recursive weighted fusion method in this process weakens the impact of abnormal fluctuations in a single round on the radar interference identification hyperparameters, improving the stability of the radar interference identification hyperparameters as the scene evolves.
[0035] In one possible implementation, the radar interference identification hyperparameters for the next increment round can be determined based on the radar interference identification hyperparameters corresponding to the current increment round and the local candidate coefficients corresponding to the evaluation subset of the next increment round.
[0036] For example, when entering the next incremental round, the local candidate coefficients corresponding to the evaluation subset of the next incremental round can be determined by the method in the embodiment of the present invention. Based on the radar interference identification hyperparameters corresponding to the current incremental round and the local candidate coefficients corresponding to the evaluation subset of the next incremental round, the radar interference identification hyperparameters corresponding to the next incremental round are determined by the weighted fusion method in the embodiment of the present invention.
[0037] In the above embodiments, the radar interference identification hyperparameters for the next incremental round are determined based on the radar interference identification hyperparameters corresponding to the current incremental round. This not only explicitly preserves the old category knowledge contained in the radar interference identification hyperparameters corresponding to the current incremental round in the next incremental round, but also eliminates the need for historical sample replay or additional gating network training. This not only improves the security of data in historical samples, but also reduces storage and computational overhead, thereby improving deployment efficiency.
[0038] In one possible implementation, a historical task dataset can be obtained; a correspondence between the sub-scores of each subspace and the categories can be established; based on the correspondence between the sub-scores of each subspace and the categories, the sub-scores of the target subspace corresponding to each category can be assigned unit weights, and the sub-scores of other subspaces can be weighted by the radar interference identification hyperparameter to obtain the fusion score of each category; the radar interference identification result can be determined based on the fusion score of each category.
[0039] Each subspace includes the subspace in the historical task dataset and the subspace in the current task dataset. Other subspaces are the subspaces in each subspace except for the target subspace. The historical task dataset contains the data corresponding to each historical increment round, such as radar interference sample data of each category in each historical increment round.
[0040] In one possible implementation, the categories in each subspace can be obtained; the sub-scores of each subspace can be determined based on the categories in each subspace and the representation of the target sample in each subspace; or, the sub-scores of each subspace can be obtained by outputting a classification head on the representation of the target sample in each subspace.
[0041] For example, the sub-scores of each subspace can be determined by the following formula (6) or formula (7): (6); (7); in, Indicates the first The sub-scores of each subspace, Indicates the target sample at the th Representation on a subspace Indicate category In the Prototype or discriminant vectors in each subspace It can be a cosine similarity function, a negative Euclidean distance function, or other comparable scoring functions. This indicates the output of the classification header.
[0042] After obtaining the sub-scores for each subspace, a correspondence between the sub-scores and categories of each subspace can be established based on the obtained categories and sub-scores for each subspace.
[0043] For example, the fusion score for each category can be determined using the following formula (8): (8); in, Indicate category The fusion score, This represents the scaling factor. It can be obtained through learning, or it can be a preset value based on actual needs. Indicate category The source of the incremental round, i.e., the category The target subspace that was initially learned This represents the component scores of the target subspace. Indicates the number of other subspaces. Indicates the first The sub-scores of each subspace, This represents the sum of the scores for each item in the other subspaces. In one possible implementation, the maximum value among the fusion scores of each category can be used as the radar interference identification result.
[0044] Among them, the radar interference identification result can be a numerical value, and different values correspond to different categories. Therefore, when the radar interference identification result is a numerical value, the corresponding category can be directly determined based on the radar interference identification result.
[0045] For example, the maximum value of the fusion score for each category can be determined by directly comparing the sizes or by the following formula (9), so that the maximum value of the fusion score for each category can be used as the radar interference identification result.
[0046] (9); in, This indicates the radar interference identification result. A collection representing historical categories. Represents the set of the current category. Indicate category The fusion score.
[0047] Please see Figure 2 , Figure 2 This is a schematic diagram of a hyperparameter smoothing optimization device for radar interference identification under no-replay conditions provided by an embodiment of the present invention. The radar interference identification hyperparameter smoothing optimization device under no-replay conditions provided by the present invention includes a data acquisition module 201, a data sampling module 202, a first determination module 203, and a second determination module 204. Data acquisition module 201 is used to acquire the current task dataset and current category count corresponding to the current increment round, and the historical global coefficient and historical category count corresponding to the historical increment round; Data sampling module 202 is used to randomly downsample the current task dataset to obtain the current evaluation subset; The first determining module 203 is used to determine the local candidate coefficients corresponding to the current evaluation subset within a preset range using a one-dimensional bounded search algorithm; The second determining module 204 is used to weight the historical global coefficients and local candidate coefficients based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round.
[0048] Please see Figure 3 , Figure 3 This is a schematic diagram of the hardware entity of an electronic device to which embodiments of the present invention apply. The electronic device provided in this embodiment includes a processor 301, a communication interface 302, a memory 303, and a communication bus 304. The processor 301, communication interface 302, and memory 303 communicate with each other via the communication bus 304. The memory 303 stores computer programs. When the processor 301 executes the program stored in the memory 303, it implements the steps provided in the above method embodiments.
[0049] The communication bus 304 mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus 304 can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, it is represented by only one thick line in the figure, but this does not indicate that there is only one bus or one type of bus.
[0050] Communication interface 302 is used for communication between the above-mentioned electronic device and other devices.
[0051] The memory 303 may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.
[0052] The processor 301 mentioned above can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0053] The method provided in this invention can be applied to electronic devices. Specifically, the electronic device can be a desktop computer, a portable computer, a smart mobile terminal, a server, etc. No limitation is made herein; any electronic device that can implement this invention falls within the protection scope of this invention.
[0054] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps provided in the above-described method embodiments.
[0055] For the embodiments of the device / electronic device / storage medium, since they are basically similar to the method embodiments, the description is relatively simple. For specific details and beneficial effects, please refer to the description of the method embodiments.
[0056] It should be noted that the device, electronic device and storage medium in the embodiments of the present invention are respectively the radar interference identification hyperparameter smoothing optimization device, electronic device and storage medium applied above. Therefore, all embodiments of the radar interference identification hyperparameter smoothing optimization method described above are applicable to the device, electronic device and storage medium, and can achieve the same or similar beneficial effects.
[0057] The terminal device provided by the embodiments of the present invention can display proper nouns and / or fixed phrases for users to select, thereby reducing user input time and improving user experience.
[0058] This terminal device exists in various forms, including but not limited to: (1) Mobile communication devices: These devices are characterized by their mobile communication capabilities and are primarily designed to provide voice and data communication. These terminals include smartphones (e.g., iPhones), multimedia phones, feature phones, and low-end phones.
[0059] (2) Ultra-mobile personal computer devices: These devices fall under the category of personal computers, possessing computing and processing capabilities, and generally also have mobile internet access features. These terminals include PDAs, MIDs, and UMPCs, such as the iPad.
[0060] (3) Portable entertainment devices: These devices can display and play multimedia content. This category includes audio and video players (such as iPods), handheld game consoles, e-book readers, as well as smart toys and portable car navigation devices.
[0061] (4) Other electronic devices with data interaction functions.
[0062] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.
[0063] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. In addition, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.
[0064] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.
[0065] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus (devices), or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects, all of which are collectively referred to herein as "modules" or "systems." Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The computer program may be stored / distributed in a suitable medium, provided with or as part of other hardware, or may take other distribution forms, such as via the Internet or other wired or wireless telecommunications systems.
[0066] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0067] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0068] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0069] In addition, this embodiment of the invention also provides a display device, which may include the display substrate provided in the above embodiments. The display device can be any product or component with display function, such as an LTPO display device, a Micro LED display device, a liquid crystal panel, electronic paper, an OLED panel, an AMOLED panel, a mobile phone, a tablet computer, a television, a monitor, a laptop computer, or a digital photo frame.
[0070] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.
Claims
1. A method for smoothing and optimizing hyperparameters for radar interference identification under no-replay conditions, characterized in that, include: Get the current task dataset and number of categories corresponding to the current increment round, and the historical global coefficients and number of categories corresponding to the historical increment rounds; The current task dataset is randomly downsampled to obtain the current evaluation subset; Within a preset range, the local candidate coefficients corresponding to the current evaluation subset are determined by a one-dimensional bounded search algorithm. The historical global coefficients and the local candidate coefficients are weighted based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round.
2. The hyperparameter smoothing optimization method for radar interference identification under no-replay conditions according to claim 1, characterized in that, The step of determining the local candidate coefficients corresponding to the current evaluation subset using a one-dimensional bounded search algorithm includes: Multiple candidate coefficients are determined by particle swarm optimization algorithm, and the candidate coefficient that maximizes the value obtained by particle swarm optimization algorithm is taken as the local candidate coefficient.
3. The hyperparameter smoothing optimization method for radar interference identification under no-replay conditions according to claim 1, characterized in that, The step of weighting the historical global coefficients and the local candidate coefficients based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round includes: When the current increment round number is 1, the local candidate coefficients are used as the radar interference identification hyperparameters; When the number of the current incremental rounds is greater than 1, the historical global coefficients and the local candidate coefficients are weighted based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters.
4. The hyperparameter smoothing optimization method for radar interference identification under no-replay conditions according to claim 1, characterized in that, The method further includes: The radar interference identification hyperparameters for the next incremental round are determined based on the radar interference identification hyperparameters corresponding to the current incremental round and the local candidate coefficients corresponding to the evaluation subset of the next incremental round.
5. The hyperparameter smoothing optimization method for radar interference identification under no-replay conditions according to claim 1, characterized in that, The method further includes: Obtain historical task datasets; Establish the correspondence between the sub-scores and categories of each subspace; each subspace includes the subspaces in the historical task dataset and the subspaces in the current task dataset; Based on the correspondence between the sub-scores and categories of each subspace, the sub-scores of the target subspace corresponding to each category are assigned unit weights. The sub-scores of other subspaces are weighted and attenuated using the radar interference identification hyperparameter to obtain the fusion scores of each category. The other subspaces are the subspaces other than the target subspace among the subspaces. The radar interference identification result is determined based on the fusion scores of the aforementioned categories.
6. The hyperparameter smoothing optimization method for radar interference identification under non-replay conditions according to claim 5, characterized in that, The establishment of the correspondence between the sub-scores of each subspace and the categories includes: Obtain the categories in each subspace; The sub-scores for each subspace are determined based on the categories in each subspace and the representation of the target sample in each subspace; or, a classification head is output from the representation of the target sample in each subspace to obtain the sub-scores for each subspace.
7. The hyperparameter smoothing optimization method for radar interference identification under non-replay conditions according to claim 5, characterized in that, The determination of radar interference identification results based on the fusion scores of the aforementioned categories includes: The maximum value among the fusion scores of each category is taken as the radar interference identification result.
8. A hyperparameter smoothing optimization device for radar interference identification under non-replay conditions, characterized in that, include: The data acquisition module is used to acquire the current task dataset and the number of current categories corresponding to the current increment round, as well as the historical global coefficients and the number of historical categories corresponding to the historical increment rounds. The data sampling module is used to randomly downsample the current task dataset to obtain the current evaluation subset; The first determining module is used to determine the local candidate coefficients corresponding to the current evaluation subset within a preset range using a one-dimensional bounded search algorithm. The second determining module is used to weight the historical global coefficients and the local candidate coefficients based on the number of historical categories and the number of current categories to obtain the radar interference identification hyperparameters corresponding to the current incremental round.
9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.