Synthetic pseudo text enhanced small sample radar interference identification method and system

Abstract

The invention discloses a synthetic pseudo text enhanced small sample radar interference identification method and system, and the method comprises the steps: generating a group of k-classification n-sample support sets for radar interference types; generating a pseudo text corresponding to each interference type by using the generative large language model, and filtering to obtain a pseudo text data set; obtaining a sample image according to the support set, generating an image-text pair in a random sampling mode according to the sample image of each interference type and the pseudo-text data set, and forming a training set; according to the training set, utilizing a total loss function to finely adjust parameters of a contrast language-image pre-training CLIP model; an interference signal to be recognized is converted into an image, an encoder of the fine-tuned CLIP model is used for extracting image features and text features corresponding to the support set, the similarity between the image features and the text features of each category is calculated, and the text type with the maximum similarity is taken as the output interference type. The interference pattern recognition performance under the small sample condition can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of radar interference recognition, and in particular to a small sample radar interference recognition method and system enhanced by synthetic pseudo-text. Background Art

[0002] Electromagnetic interference has a serious impact on radar detection. Based on the interference pattern, radar selects appropriate waveforms and signal processing methods, which are common anti-interference methods. The performance of interference pattern recognition plays an important role in subsequent anti-interference, so interference recognition has received extensive attention.

[0003] In the actual electronic battlefield, radar jamming is unpredictable and inaccessible, which makes it difficult to collect a data set with enough samples. Usually, only a few sampling frames of different jamming types can be obtained, based on which the recognition model is trained. At this time, the problem of jamming pattern recognition is transformed into a small sample learning problem. The scarcity of training data seriously reduces the accuracy of recognition. Existing small sample classification methods are mainly concentrated in the field of computer vision, and there is little research in the field of small sample radar jamming recognition. Compared with images, radar signals are more diverse and complex, requiring the model to have better generalization ability. Summary of the invention

[0004] Purpose of the invention: To address the problem of interference pattern recognition under the scarcity of existing radar interference sample data, the present invention proposes a method that is not affected by the number of samples and can identify the interference type even under small sample conditions.

[0005] Technical solution: In order to achieve the above-mentioned invention object, the present invention adopts the following technical solution:

[0006] In a first aspect, a small sample radar interference recognition method enhanced by synthetic pseudo text comprises the following steps:

[0007] A support set of k-classified n samples is generated for radar interference types. A generative large language model is used to randomly give prompt templates and temperature parameters based on the support set and a preset prompt template list for interference type queries, generate pseudo texts corresponding to each interference type, and filter the pseudo texts to obtain a pseudo text dataset for each interference type.

[0008] According to the support set, the interference signal is converted into sample image data using the time-frequency transformation method; according to the sample image data and pseudo-text data set of each interference type, image-text pairs are generated by random sampling, and a total of k classified image-text pair data sets are generated to form a training set;

[0009] The training set is input into the image encoder and text encoder of the comparative language-image pre-trained CLIP model, image feature extraction and text feature extraction are performed respectively, and the feature similarity between the image and the text is calculated. The cross entropy loss is calculated based on the similarity, and the triple loss of the image feature and the triple loss of the text feature are calculated to form a total loss function. The model parameters are back-propagated according to the total loss function and the parameters are updated until the required number of training rounds is reached to achieve fine-tuning of the CLIP model.

[0010] The interference signal to be identified is converted into time-frequency domain image data using the time-frequency transformation method. The time-frequency domain image data and the interference type category name are respectively input into the image encoder and text encoder of the fine-tuned CLIP model to extract image features and category text features respectively, and the similarity between the image features and each category text feature is calculated. The text with the largest similarity is taken as the best matching description, and this text type is the output interference type.

[0011] Preferably, the generative large language model includes Llama-3-8b-instruct, Gemma-7b-it, Openchat-7b, Mistral-7b-instruct, and the temperature parameter is a random floating point number between 0 and 1.

[0012] Preferably, a generative large language model is used to randomly provide a prompt template and a temperature parameter according to a support set and a preset prompt template list for interference type query, and generate pseudo text corresponding to each interference type, including:

[0013] Determine, according to the support set, a list of prompt templates for interference type query, wherein the prompt template is a prompt statement for inquiring about the characteristics of a specified interference type;

[0014] According to the selected generative large language model, a large language model list is formed;

[0015] Execute the query. For each query process, randomly select the prompt template and the large language model from the prompt template list and the large language model list according to the temperature parameter to obtain the corresponding pseudo text.

[0016] Preferably, the pseudo text is filtered to obtain pseudo text data of each interference type, including:

[0017] The query response text is split with the specified delimiter to obtain short texts. For short texts, sentences whose length is not within the specified range, irrelevant texts, and existing texts of the same type are removed to obtain a high-quality pseudo-text dataset for each interference type.

[0018] Preferably, the feature similarity between the image and the text is expressed as:

[0019]

[0020] They represent the features of the image and text respectively, b represents the batch number, and d represents the feature dimension.

[0021] Preferably, the total loss function is expressed as:

[0022]

[0023] TripletMarginLoss represents triplet loss, and CrossEntropyLoss represents cross entropy loss.

[0024] Preferably, the time-frequency transformation method adopts short-time Fourier transform.

[0025] In the second aspect, a small sample radar interference recognition system enhanced by synthetic pseudo-text includes:

[0026] The pseudo text data generation module is used to generate a support set of k-classified n samples for radar interference types; using a generative large language model, according to the support set and a preset prompt template list for interference type query, randomly give prompt templates and temperature parameters, generate pseudo text corresponding to each interference type, and filter the pseudo text to obtain a pseudo text data set for each interference type;

[0027] The training set generation module is used to convert the interference signal into sample image data by using the time-frequency transformation method according to the support set; according to the sample image data and pseudo-text data set of each interference type, an image-text pair is generated by random sampling, and a total of k classified image-text pair data sets are generated to form a training set;

[0028] A model training module is used to input the training set into the image encoder and text encoder of the comparative language-image pre-training CLIP model, perform image feature extraction and text feature extraction respectively, calculate the feature similarity between the image and the text, calculate the cross entropy loss based on the similarity, and calculate the triplet loss of the image feature and the triplet loss of the text feature to form a total loss function, perform backpropagation on the model parameters according to the total loss function and update the parameters until the required number of training rounds is reached, so as to achieve fine-tuning of the CLIP model;

[0029] The interference recognition module is used to convert the interference signal to be identified into time-frequency domain image data by using the time-frequency transformation method, input the time-frequency domain image data and the interference type category name into the image encoder and text encoder of the fine-tuned CLIP model respectively, extract the image features and category text features respectively, and calculate the similarity between the image features and each category text feature, and take the text with the largest similarity as the best matching description, and this text type is the output interference type.

[0030] In a third aspect, a computer device comprises: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the small sample radar interference identification method enhanced by synthetic pseudo-text as described in the first aspect of the present invention are implemented.

[0031] In a fourth aspect, a computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the small sample radar interference identification method enhanced by synthetic pseudo-text as described above.

[0032] Beneficial effects: In order to solve the problem of interference pattern recognition under the scarcity of existing radar interference sample data, the present invention proposes a small sample interference recognition method and system of synthetic pseudo text data set, which generates pseudo text data set of each interference type by constructing a generative large language model LLM, and performs time-frequency transformation on the interference signal to obtain time-frequency domain image data, then passes the text and image data through the corresponding encoders respectively to obtain the image-to-text data set, and obtains the inferential model after fine-tuning the CLIP model, converts the signal to be recognized into time-frequency domain image data, extracts features through the encoder together with the text of each potential category, calculates and selects the text with the greatest similarity, and realizes the recognition of interference type. The present invention adopts the method of generating pseudo text data set to replace the scarce radar interference sample data set, so as to realize the effective recognition of interference under the condition of scarce radar interference samples. Simulation results show that the average recognition accuracy of this method reaches 93.32% when there are only 5 samples of each interference type. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of the method of the present invention;

[0034] Figure 2 It is the time-frequency domain diagram of COMB interference pattern;

[0035] Figure 3 Feature maps of the text dataset before and after fine-tuning;

[0036] Figure 4 This is a graph of the test results in the form of a confusion matrix. DETAILED DESCRIPTION

[0037] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0038] The present invention is suitable for identifying interference patterns of radar signals when interference samples are scarce. The overall process of the identification method is shown in the attached figure. Figure 1 As shown, the specific steps include:

[0039] Step 1: Synthesize a pseudo-text dataset

[0040] The present invention uses generative large language models (LLM) to generate pseudo text data sets for each interference sample. s(t) represents the transmission waveform. Assuming that the jammer adopts DRFM (Digital RF Memory) jamming technology, the radar signal can be intercepted and modulated. c (·) represents the modulation function used by the DRFM jammer, where c = 1, 2, ..., C represents the type of interference. The interfered received signal can be expressed as:

[0041]

[0042] Where τ represents the time delay and n represents Gaussian white noise.

[0043] There are different forms of expression for different types of interference. As shown in Table 1, six classic DRFM interference formulas are given, namely C&I (Chopper&Intercept), COMB (comb spectrum interference), NCJ (Noise Convolution Jamming), FM (Frequency Modulation), SMSP (Smeared Spectrum) and ISRJ (Interrupted Sampling Repeater Jamming).

[0044] Table 1 Interference formula

[0045]

[0046] For a typical small sample learning task, only a few samples are available, and this set is used as the support set. A small sample learning task with a support set is called a kways and nshots task, where k represents the number of categories and n represents the number of samples in each category. The present invention selects a set containing 6 types of interference, with 5 samples in each category as the support set of 6ways 5shots for training.

[0047] The support set contains 6 types of interference, as shown in Table 1. The interference is generated using the formula in Table 1, and the parameters when generating the interference signal are shown in Table 2. Each class has 5 samples, making the support set 6ways 5shots.

[0048] Table 2 Interference parameter settings

[0049]

[0050] The present invention selects Llama-3-8b-instruct, Gemma-7b-it, Openchat-7b, and Mistral-7b-instruct as the LLMs used, and the temperature parameter is a random floating point number of 0 to 1. For the query process of each interference, a prompt template and a large language model are randomly selected from a prompt template list and a large language model list according to the temperature parameter, and corresponding pseudo texts are obtained to obtain a text description of each interference. Then, the response text is split and filtered to filter out sentences that are too long or too short, inaccurate descriptions, and irrelevant texts.

[0051] Specifically, the present invention first designs a prompt word template set, such as Do you know something about{}, Give me several features{}have, Give me features of spectrum of{}, etc. Then, random sampling is performed in the interference type category name and the prompt word template set to form a complete prompt word, and the prompt word is provided to a randomly selected large language model LLM, and pseudo text data is generated according to the response of the LLM. Next, the response text generated by the LLM model is split and cleaned to obtain high-quality pseudo text data. Among them, the splitting operation includes segmenting the text with separators such as ",., " to obtain short text. The cleaning operation includes removing texts with a number of less than 10 and greater than 100, texts with similar data already in the data set, and obviously irrelevant texts.

[0052] Examples of high-quality pseudo-text datasets for each class are shown in Table 3.

[0053] Table 3 Pseudo text data examples

[0054]

[0055] It should be understood that the pseudo text data generated in this embodiment is English text, but this is only for example purposes and does not mean that only English text can be generated. In actual applications, pseudo text in Chinese or other languages ​​can be generated based on the prompt template and the large language model.

[0056] Step 2: Identification model fine-tuning

[0057] The present invention uses the Contrastive Language-Image Pre-Training (CLIP) model as the recognition model. For the CLIP model, the data set is in the form of image-text pairs. Based on the support set and pseudo-text data set of each interference type, image-text pairs are generated by random sampling. In order to convert the interference signal into an image, the present invention uses short-time Fourier transform (STFT) to obtain the time-frequency domain image data of the interference signal. The training set has a total of 35 samples, including 5 uninterrupted samples and 5 samples of each of the 6 interference styles. Figure 2 An example diagram of COMB interference pattern is given.

[0058] The backbone image encoder of the CLIP model selected by the present invention is Vit-B / 16. In the fine-tuning process, according to the forward process of CLIP, both image and text data must pass through the FSRRM model, including an image encoder and a text encoder, to perform image feature extraction and text feature extraction respectively; then the feature similarity between the image and the text is calculated. During the training process, CLIP learns the embedding vector by maximizing the similarity with the corresponding image and text description while minimizing the similarity with other unmatched pairs. Through contrastive learning, the image and text are mapped to the same embedding space; the loss function is calculated, the model parameters are back-propagated and updated to reach the required number of training rounds. This process is called model fine-tuning (Tuning). Among them, represents the features of a small batch of images and texts. b represents the batch number, and d represents the feature dimension. The similarity between the image and the text is:

[0059]

[0060] In order to obtain a better model training effect, the present invention adds two losses based on the CLIP model loss function, namely, the triple loss of image features and the triple loss of text features. The total loss function can be represented by the sum of the following three parts: the triple loss of image features, the triple loss of text features, and the calculation of the log-likelihood ratio between the image and the text through the similarity matrix and the calculation of the cross entropy loss, which can be expressed as:

[0061]

[0062] The CrossEntropyLoss reflects the degree of alignment between the image and the text, and the two TripletMarginLoss reflect the degree of aggregation and distinguishability between different categories of images and different categories of text, respectively.

[0063]

[0064] The above formula is the calculation formula of TripletMarginLoss loss function, where x a is the anchor point sample, x p is a positive sample, x n is a negative sample, and α is the tolerance. This loss function improves the clustering effect of feature vectors by minimizing the distance between samples of the same category and maximizing the distance between samples of different categories.

[0065] Figure 3 The feature graphs of the text dataset before and after fine-tuning are given. It can be seen that by fine-tuning the CLIP model on the pseudo-text-image dataset, the same type of text data is more clustered in the latent space, and the different types of text data are more distinguishable. Therefore, the fine-tuning process enables the model to identify the type of radar interference by calculating and comparing the similarity between the time-frequency domain image of the interference and the descriptive text prompts of each latent class.

[0066] Step 3: Interference Identification

[0067] Based on the fine-tuned model, the input interference is identified:

[0068] First, the interference signal to be identified is converted into time-frequency domain image data through STFT, and then passed through the encoder together with the category name of each potential interference type. The image encoder and text encoder calculate the similarity between the time-frequency domain image data to be identified and each potential category, respectively, and compare the similarities. The text with the largest similarity is selected as the best matching description, and this text type is the output interference type.

[0069] The present invention performs recognition tests on a test data set containing 2000 samples in each category. Figure 4 The test results are shown in the form of confusion matrix. As can be seen from the figure, for the five types of NCJ, FM, ISRJ, C&I and COMB, the present invention has a high recognition accuracy. However, the recognition accuracy of SMSP is 72%, which is slightly lower than that of other interference categories. The reason is that Figure 3 It can be seen that the characteristics of SMSP are more scattered and closer to the undisturbed class compared to other classes.

[0070] The present invention selects two typical small sample learning methods, matching network and prototype network, as the control group for comparison. As shown in Table 4, in NCJ, FM, C&I and COMB classes, the method proposed by the present invention shows better recognition performance. Although in SMSP and ISRJ classes, the method proposed by the present invention has no advantage over the comparison method, its accuracy is 2.73% and 0.9% lower than that of the matching network, respectively. However, on average, the recognition accuracy of each interference type is 93.32%, which is 4.15% and 5.94% higher than that of the prototype network and the matching network, respectively.

[0071] Table 4 Accuracy comparison results

[0072]

[0073] The present invention identifies the query signal by selecting a set of 6ways and 5shots as a support set; and selects text descriptions as supporting data to help establish consistency between interference and categories. First, the consistency between the signal and the text description is established; then, the recognition proof is performed by calculating the similarity between the input signal and the potential text description; finally, the best matching text description should reflect the interference type of the input signal. The simulation results show that the average recognition accuracy reaches 93.32% when there are only 5 samples of each interference type. Therefore, it can be said that the present invention can effectively identify interference under the condition that radar interference samples are scarce. In addition, the present invention utilizes the semantic generalization ability of the existing pre-trained model and uses text enhancement to perform fine-tuning to achieve radar interference style recognition under small sample conditions; compared with traditional methods, it has strong portability and the pre-trained model is replaceable; and the method generated by LLM replaces the manual collection of data, which has good simplicity and efficiency.

[0074] According to another embodiment of the present invention, a small sample radar interference identification system enhanced by synthetic pseudo-text is provided, comprising:

[0075] The pseudo text data generation module is used to generate a support set of k-classified n samples for radar interference types; using a generative large language model, according to the support set and a preset prompt template list for interference type query, randomly give prompt templates and temperature parameters, generate pseudo text corresponding to each interference type, and filter the pseudo text to obtain a pseudo text data set for each interference type;

[0076] The training set generation module is used to convert the interference signal into sample image data by using the time-frequency transformation method according to the support set; according to the sample image data and pseudo-text data set of each interference type, an image-text pair is generated by random sampling, and a total of k classified image-text pair data sets are generated to form a training set;

[0077] A model training module is used to input the training set into the image encoder and text encoder of the comparative language-image pre-training CLIP model, perform image feature extraction and text feature extraction respectively, calculate the feature similarity between the image and the text, calculate the cross entropy loss based on the similarity, and calculate the triplet loss of the image feature and the triplet loss of the text feature to form a total loss function, perform backpropagation on the model parameters according to the total loss function and update the parameters until the required number of training rounds is reached, so as to achieve fine-tuning of the CLIP model;

[0078] The interference recognition module is used to convert the interference signal to be identified into time-frequency domain image data by using the time-frequency transformation method, input the time-frequency domain image data and the interference type category name into the image encoder and text encoder of the fine-tuned CLIP model respectively, extract the image features and category text features respectively, and calculate the similarity between the image features and each category text feature, and take the text with the largest similarity as the best matching description, and this text type is the output interference type.

[0079] It should be understood that the synthetic pseudo-text enhanced small sample radar interference identification system provided in this embodiment can implement all the technical solutions in the above method embodiments, and the functions of its various functional modules can be specifically implemented according to the methods in the above method embodiments. The specific implementation process can refer to the relevant description in the above embodiments, which will not be repeated here.

[0080] The present invention also provides a computer device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the small sample radar interference identification method enhanced by synthetic pseudo-text as described above are implemented.

[0081] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the synthetic pseudo-text enhanced small sample radar interference identification method as described above are implemented.

[0082] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices (systems), computer equipment or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0083] The present invention is described with reference to a flowchart of a method according to an embodiment of the present invention. It should be understood that each process in the flowchart and a combination of processes in the flowchart can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A device that specifies functions in a process or multiple processes.

[0084] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A function specified in a process or multiple processes.

[0085] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 The steps of a specified function in a process or multiple processes.

Claims

1. A small sample radar interference recognition method enhanced by synthetic pseudo-text, characterized in that: The following steps are involved: A support set of k-classified n samples is generated for radar interference types. A generative large language model is used to randomly give prompt templates and temperature parameters based on the support set and a preset prompt template list for interference type queries, generate pseudo texts corresponding to each interference type, and filter the pseudo texts to obtain a pseudo text dataset for each interference type. According to the support set, the interference signal is converted into sample image data using the time-frequency transformation method; according to the sample image data and pseudo-text data set of each interference type, image-text pairs are generated by random sampling, and a total of k classified image-text pair data sets are generated to form a training set; The training set is input into the image encoder and text encoder of the comparative language-image pre-trained CLIP model, image feature extraction and text feature extraction are performed respectively, and the feature similarity between the image and the text is calculated. The cross entropy loss is calculated based on the similarity, and the triple loss of the image feature and the triple loss of the text feature are calculated to form a total loss function. The model parameters are back-propagated according to the total loss function and the parameters are updated until the required number of training rounds is reached to achieve fine-tuning of the CLIP model. The interference signal to be identified is converted into time-frequency domain image data using the time-frequency transformation method. The time-frequency domain image data and the interference type category name are respectively input into the image encoder and text encoder of the fine-tuned CLIP model to extract image features and category text features respectively, and the similarity between the image features and each category text feature is calculated. The text with the largest similarity is taken as the best matching description, and this text type is the output interference type.

2. The method according to claim 1, characterized in that The generative large language models include Llama-3-8b-instruct, Gemma-7b-it, Openchat-7b, and Mistral-7b-instruct, and the temperature parameter is a random floating point number between 0 and 1.

3. The method according to claim 1, characterized in that Using the generative large language model, according to the support set and the preset prompt template list for interference type query, the prompt template and temperature parameters are randomly given to generate pseudo text corresponding to each interference type, including: Determine, according to the support set, a list of prompt templates for interference type query, wherein the prompt template is a prompt statement for inquiring about the characteristics of a specified interference type; According to the selected generative large language model, a large language model list is formed; Execute the query. For each query process, randomly select the prompt template and the large language model from the prompt template list and the large language model list according to the temperature parameter to obtain the corresponding pseudo text.

4. The method according to claim 1, characterized in that: The pseudo text is filtered to obtain pseudo text data of each interference type, including: The query response text is split with the specified delimiter to obtain short texts. For short texts, sentences whose length is not within the specified range, irrelevant texts, and existing texts of the same type are removed to obtain a high-quality pseudo-text dataset for each interference type.

5. The method according to claim 1, characterized in that The feature similarity between image and text is expressed as: They represent the features of the image and text respectively, b represents the batch number, and d represents the feature dimension.

6. The method according to claim 5, characterized in that The total loss function is expressed as: TripletMarginLoss represents triplet loss, and CrossEntropyLoss represents cross entropy loss.

7. The method according to claim 1, characterized in that The time-frequency transform method adopts short-time Fourier transform.

8. A small sample radar interference recognition system enhanced by synthetic pseudo-text, characterized in that: include: The pseudo text data generation module is used to generate a support set of k-classified n samples for radar interference types; using a generative large language model, according to the support set and a preset prompt template list for interference type query, randomly give prompt templates and temperature parameters, generate pseudo text corresponding to each interference type, and filter the pseudo text to obtain a pseudo text data set for each interference type; The training set generation module is used to convert the interference signal into sample image data by using the time-frequency transformation method according to the support set; according to the sample image data and pseudo-text data set of each interference type, an image-text pair is generated by random sampling, and a total of k classified image-text pair data sets are generated to form a training set; A model training module is used to input the training set into the image encoder and text encoder of the comparative language-image pre-training CLIP model, perform image feature extraction and text feature extraction respectively, calculate the feature similarity between the image and the text, calculate the cross entropy loss based on the similarity, and calculate the triplet loss of the image feature and the triplet loss of the text feature to form a total loss function, perform backpropagation on the model parameters according to the total loss function and update the parameters until the required number of training rounds is reached, so as to achieve fine-tuning of the CLIP model; The interference recognition module is used to convert the interference signal to be identified into time-frequency domain image data by using the time-frequency transformation method, input the time-frequency domain image data and the interference type category name into the image encoder and text encoder of the fine-tuned CLIP model respectively, extract the image features and category text features respectively, and calculate the similarity between the image features and each category text feature, and take the text with the largest similarity as the best matching description, and this text type is the output interference type.

9. A computer device, characterized in that: include: one or more processors; Memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of small sample radar interference identification enhanced by synthetic pseudo-text as described in any one of claims 1-7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the synthetic pseudo-text enhanced small sample radar interference identification method according to any one of claims 1 to 7 are implemented.

Images