CN116502139A - Radiation source signal individual identification method based on integrated countermeasure migration - Google Patents
Radiation source signal individual identification method based on integrated countermeasure migration Download PDFInfo
- Publication number
- CN116502139A CN116502139A CN202310478591.2A CN202310478591A CN116502139A CN 116502139 A CN116502139 A CN 116502139A CN 202310478591 A CN202310478591 A CN 202310478591A CN 116502139 A CN116502139 A CN 116502139A
- Authority
- CN
- China
- Prior art keywords
- domain
- source
- representing
- radiation source
- classifier
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000005855 radiation Effects 0.000 title claims abstract description 40
- 238000013508 migration Methods 0.000 title claims abstract description 24
- 230000005012 migration Effects 0.000 title claims abstract description 21
- 238000005070 sampling Methods 0.000 claims description 22
- 238000013528 artificial neural network Methods 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000003042 antagnostic effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/259—Fusion by voting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/094—Adversarial learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/096—Transfer learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Evolutionary Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Semiconductor Lasers (AREA)
Abstract
A radiation source signal individual identification method based on integration and migration countermeasure relates to a radiation source signal individual identification method. The method aims to solve the problems of low identification accuracy and poor model generalization of the traditional individual identification method. The invention adopts an integrated anti-migration method to identify the radiation source individuals, solves the problem of difficult marking of the radiation source signal data, and improves the robustness and generalization capability of the model. The invention belongs to the technical field of signal identification.
Description
Technical Field
The invention relates to a radiation source signal individual identification method, and belongs to the technical field of signal identification.
Background
The radiation source identification technology is one of key technologies in the fields of communication reconnaissance and electronic countermeasure, and is also the leading direction of the field of electronic warfare at home and abroad. It identifies the individual radiation source device by feature extraction of the captured electromagnetic signals. In the civilian field, radiation source identification plays an important role in the fields of spectrum management and communication signal identification. In the face of increasingly complex and changeable electromagnetic environments, the traditional radiation source identification method has more and more limitations, and the problems of poor individual identification effect, poor cross-environment identification stability and the like.
Deep learning is a branch of machine learning, takes an artificial neural network as a framework, and is an algorithm for performing characterization learning on data. The complex mapping is realized by connecting the multi-layer nodes and nonlinear activation, any distribution can be fitted theoretically, and a better identification effect can be realized in a cooperation scene. In a non-cooperative scenario, the recognition model also needs to be updated according to changes in the radiation source signal, but requires a lot of training marker data and expensive training costs, in the face of rapid updates of various electromagnetic devices. The thinking of transfer learning is to solve the problem, and the method utilizes a data set or a model of a source domain to design a corresponding algorithm to assist a target domain in learning and knowledge acquisition on the premise of maximally utilizing the knowledge of the source domain. However, the migration learning also faces the problem of poor robustness and generalization, so we need to enhance the robustness with the countermeasure method and the generalization with the integration method.
Disclosure of Invention
The invention aims to solve the problems of low identification accuracy and poor model generalization of the traditional individual identification method, and further provides a radiation source signal individual identification method based on integrated countermeasure migration.
The technical scheme adopted by the invention for solving the problems is as follows: the specific steps of the invention are as follows:
step one, selecting a radiation source signal under one signal-to-noise ratio as a source domain, and selecting a radiation source signal under another signal-to-noise ratio as a target domain;
step two, self-sampling m samples of the marked source domain and the unmarked target domain to generate n new data sets, wherein the sample proportion of the source domain and the target domain is kept consistent;
step three, respectively inputting the data sets into a domain countermeasure neural network for training to obtain n weak classifiers;
voting the predicted result by adopting a voting method to obtain a final predicted result of the integrated model;
calculating the accuracy of the integrated model on the target domain test set;
and step six, inputting the target domain test set into the integrated model to perform individual identification and classification of the radiation source.
Further, in the first step, noise with different signal to noise ratios is added to the radiation source signals of the source domain and the target domain, and the calculation formula of the signal to noise ratio is as follows:
in the formula (1), the unit of SNR is dB, S represents the power of a signal, and N represents the power of noise.
In the second step, m samples are sampled by self for the marked source domain and the unmarked target domain to generate n new data sets, wherein the n new data sets are obtained by replaced random sampling; for an original data set of m samples, randomly selecting one sample at a time to be placed into a sampling set, then reapplying the sample into the original data set, then randomly sampling the next sample until the number of the sampling set reaches m, thus constructing a data set, and generating n data sets by continuously repeating the process.
Further, in the third step, the data set is input into the domain countermeasure neural network for training, and the gradient inversion layer is added into the feature extractor and the domain discriminator to realize the countermeasure effect, so that the total loss of the countermeasure migration network is composed of two parts: loss of label predictor and domain arbiter, the domain against the neural network total objective function is:
in the formula (2), E represents an expected value of an objective function, W represents a source domain feature extractor parameter, V represents a source domain and target domain classifier parameter, b represents a bias term of a feature extractor of a source domain, c represents a bias term of a classifier of the source domain and the target domain, u represents a domain classifier parameter, z represents a domain classifier bias term, i represents an ith sample of a current batch, and N represents a sum of a source domain data set and a target domain data set in a training set;
wherein the parameters of the tag predictor are updated by minimizing the objective function, and the parameters of the domain arbiter are updated by maximizing the objective function:
in the formulas (3) and (4),representing the optimal parameters of the source domain feature extractor, +.>Representing optimal parameters of source domain and target domain classifiers, < ->Optimal bias term for feature extractor representing source domain,/->Optimal bias term of classifier representing source domain and target domain,/for the classifier>Representing domain classifier optimal parameters,/->Representing domain classifier optimal bias terms, w representing source domain feature extractor parameters, v representing source domain and target domain classifier parameters, b representing bias terms of feature extractor of source domain, c representing bias terms of classifier of source domain and target domain, u representing domain classifier parameters, z representing domain classifier bias terms.
Further, in the fourth step, the voting method is adopted to vote on the predicted result, a final predicted result of the integrated model is obtained, and the majority voting method, namely minority obeys majority, is adopted.
The beneficial effects of the invention are as follows:
1. the method reduces the difference between the source domain and the target domain by adopting the countermeasure migration, enhances the robustness of the model and reduces the labeling cost;
2. the invention adopts integrated countermeasure migration, and can obtain the generalization performance remarkably superior to that of a single learner by combining a plurality of learners, thereby improving the generalization capability of the model;
3. the invention adopts an integrated anti-migration method to identify the radiation source individuals, solves the problem of difficult marking of the radiation source signal data, and improves the robustness and generalization capability of the model.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a schematic diagram of a self-service sampling method;
fig. 3 is a schematic diagram of a domain antagonistic neural network.
Detailed Description
The first embodiment is as follows: referring to fig. 1 to 3, a radiation source signal individual identification method based on integrated anti-migration according to the present embodiment is implemented by the following steps:
step one, selecting a radiation source signal under one signal-to-noise ratio as a source domain, and selecting a radiation source signal under another signal-to-noise ratio as a target domain;
step two, self-sampling m samples of the marked source domain and the unmarked target domain to generate n new data sets, wherein the sample proportion of the source domain and the target domain is kept consistent;
step three, respectively inputting the data sets into a domain countermeasure neural network for training to obtain n weak classifiers;
voting the predicted result by adopting a voting method to obtain a final predicted result of the integrated model;
calculating the accuracy of the integrated model on the target domain test set;
and step six, inputting the target domain test set into the integrated model to perform individual identification and classification of the radiation source.
The second embodiment is as follows: referring to fig. 1 to 3, in the step one of the method for identifying an individual radiation source signal based on integration and migration countermeasure according to the present embodiment, noise with different signal to noise ratios is added to radiation source signals of a source domain and a target domain, and a calculation formula of the signal to noise ratio is as follows:
in the formula (1), the unit of SNR is dB, S represents the power of a signal, and N represents the power of noise.
The samples in each data set in this embodiment may be repeated, or some samples in the original data set may not be pumped at all, and the sample distribution in each data set may be different.
And a third specific embodiment: referring to fig. 1 to 3, a description is given of the present embodiment, in which m samples are self-sampled in the marked source domain and the unmarked target domain in the second step of the method for identifying the individual radiation source signal based on integration and migration countermeasure, n new data sets are generated, and the n new data sets are obtained by random sampling with replacement; for an original data set of m samples, randomly selecting one sample at a time to be placed into a sampling set, then reapplying the sample into the original data set, then randomly sampling the next sample until the number of the sampling set reaches m, thus constructing a data set, and generating n data sets by continuously repeating the process.
The specific embodiment IV is as follows: referring to fig. 1 to 3, a description is given of the present embodiment, in which in the third step of the method for identifying an individual radiation source signal based on integrated migration countermeasure, a data set is input into a domain countermeasure neural network for training, and a gradient inversion layer is added to a feature extractor and a domain discriminator to achieve the countermeasure effect, and the total loss of the migration countermeasure network is composed of two parts: loss of label predictor and domain arbiter, the domain against the neural network total objective function is:
in the formula (2), E represents an expected value of an objective function, W represents a source domain feature extractor parameter, V represents a source domain and target domain classifier parameter, b represents a bias term of a feature extractor of a source domain, c represents a bias term of a classifier of the source domain and the target domain, u represents a domain classifier parameter, z represents a domain classifier bias term, i represents an ith sample of a current batch, and N represents a sum of a source domain data set and a target domain data set in a training set;
wherein the parameters of the tag predictor are updated by minimizing the objective function, and the parameters of the domain arbiter are updated by maximizing the objective function:
equation (3)(4) In the process, the liquid crystal display device comprises a liquid crystal display device,representing the optimal parameters of the source domain feature extractor, +.>Representing optimal parameters of source domain and target domain classifiers, < ->Optimal bias term for feature extractor representing source domain,/->Optimal bias term of classifier representing source domain and target domain,/for the classifier>Representing domain classifier optimal parameters,/->Representing domain classifier optimal bias terms, w representing source domain feature extractor parameters, v representing source domain and target domain classifier parameters, b representing bias terms of feature extractor of source domain, c representing bias terms of classifier of source domain and target domain, u representing domain classifier parameters, z representing domain classifier bias terms.
Fifth embodiment: referring to fig. 1 to 3, a description is given of the present embodiment, in which in the fourth step of the method for identifying an individual radiation source signal based on integrated countermeasure migration, a voting method is used to vote on the predicted result, a final predicted result of the integrated model is obtained, and a majority voting method, that is, a minority-subject majority method, is used.
Principle of operation
The invention discloses an integrated migration-resistant radiation source signal individual identification method, which is characterized in that ADS-B signals acquired by an actual environment are adopted, additive Gaussian white noise with different signal to noise ratios is respectively added to a source domain and a target domain, and radiation source individual identification under different noise environments is carried out. Wherein the source domain samples are labeled and the target domain samples are unlabeled;
the invention adopts a self-help sampling method to self-help sample m samples of an active domain and an unlabeled target domain to generate n new data sets, wherein the n new data sets are obtained by replaced random sampling, as shown in figure 2. For an original data set of m samples, randomly selecting one sample at a time to be placed into a sampling set, then reapplying the sample into the original data set, then randomly sampling the next sample until the number of the sampling set reaches m, thus constructing a data set, and generating n data sets by continuously repeating the process. The samples in each data set may be repeated or some samples in the original data set may not be drawn at all, and the sample distribution in each data set may be different;
the invention respectively inputs the data sets into the domain countermeasure neural network for training, and the structural schematic diagram of the domain countermeasure neural network is shown in fig. 3. Consists of three parts: feature extractor, label predictor, domain discriminator. The feature extractor is used to map the data to a specific feature space, so that the domain arbiter cannot distinguish which domain the data comes from while the tag predictor can distinguish the class of data from the source domain. The tag predictor classifies data from the source domain to separate as correct tags as possible. The domain discriminator classifies the data of the feature space to separate as far as possible which domain the data comes from. Wherein the feature extractor and the domain discriminator are connected by a gradient inversion layer.
The present invention is not limited to the preferred embodiments, but is capable of modification and variation in detail, and other embodiments, such as those described above, of making various modifications and equivalents will fall within the spirit and scope of the present invention.
Claims (5)
1. An integrated migration-resistant radiation source signal individual identification method is characterized by comprising the following steps of: the radiation source signal individual identification method based on integrated countermeasure migration is realized through the following steps:
step one, selecting a radiation source signal under one signal-to-noise ratio as a source domain, and selecting a radiation source signal under another signal-to-noise ratio as a target domain;
step two, self-sampling m samples of the marked source domain and the unmarked target domain to generate n new data sets, wherein the sample proportion of the source domain and the target domain is kept consistent;
step three, respectively inputting the data sets into a domain countermeasure neural network for training to obtain n weak classifiers;
voting the predicted result by adopting a voting method to obtain a final predicted result of the integrated model;
calculating the accuracy of the integrated model on the target domain test set;
and step six, inputting the target domain test set into the integrated model to perform individual identification and classification of the radiation source.
2. A method of identifying an individual radiation source signal based on integrated migration countermeasure according to claim 1, wherein: in the first step, noise with different signal to noise ratios is added to radiation source signals of a source domain and a target domain respectively, and a calculation formula of the signal to noise ratio is as follows:
in the formula (1), the unit of SNR is dB, S represents the power of a signal, and N represents the power of noise.
3. A method of identifying an individual radiation source signal based on integrated migration countermeasure according to claim 1, wherein: in the second step, self-help sampling m samples of the marked source domain and the unmarked target domain to generate n new data sets, wherein the n new data sets are obtained by random sampling with a put-back; for an original data set of m samples, randomly selecting one sample at a time to be placed into a sampling set, then reapplying the sample into the original data set, then randomly sampling the next sample until the number of the sampling set reaches m, thus constructing a data set, and generating n data sets by continuously repeating the process.
4. A method of identifying an individual radiation source signal based on integrated migration countermeasure according to claim 1, wherein: inputting the data set into a domain countermeasure neural network for training, and adding a gradient inversion layer into a feature extractor and a domain discriminator to realize countermeasure effect, wherein the total loss of the countermeasure migration network consists of two parts: loss of label predictor and domain arbiter, the domain against the neural network total objective function is:
in the formula (2), E represents an expected value of an objective function, W represents a source domain feature extractor parameter, V represents a source domain and target domain classifier parameter, b represents a bias term of a feature extractor of a source domain, c represents a bias term of a classifier of the source domain and the target domain, u represents a domain classifier parameter, z represents a domain classifier bias term, i represents an ith sample of a current batch, and N represents a sum of a source domain data set and a target domain data set in a training set;
wherein the parameters of the tag predictor are updated by minimizing the objective function, and the parameters of the domain arbiter are updated by maximizing the objective function:
in the formulas (3) and (4),representing the optimal parameters of the source domain feature extractor, +.>Representing optimal parameters of source domain and target domain classifiers, < ->Optimal bias term for feature extractor representing source domain,/->Optimal bias terms for classifiers representing source and target domains,representing domain classifier optimal parameters,/->Representing domain classifier optimal bias terms, w representing source domain feature extractor parameters, v representing source domain and target domain classifier parameters, b representing bias terms of feature extractor of source domain, c representing bias terms of classifier of source domain and target domain, u representing domain classifier parameters, z representing domain classifier bias terms.
5. A method of identifying an individual radiation source signal based on integrated migration countermeasure according to claim 1, wherein: and step four, voting is carried out on the predicted result by adopting a voting method, a final predicted result of the integrated model is obtained, and a majority voting method, namely minority obeys majority, is adopted.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310478591.2A CN116502139B (en) | 2023-04-28 | Radiation source signal individual identification method based on integrated countermeasure migration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310478591.2A CN116502139B (en) | 2023-04-28 | Radiation source signal individual identification method based on integrated countermeasure migration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116502139A true CN116502139A (en) | 2023-07-28 |
| CN116502139B CN116502139B (en) | 2024-07-02 |
Family
ID=
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106815643A (en) * | 2017-01-18 | 2017-06-09 | 中北大学 | Infrared spectrum Model Transfer method based on random forest transfer learning |
| CN108508411A (en) * | 2018-03-22 | 2018-09-07 | 天津大学 | Passive radar external sort algorithm signal recognition method based on transfer learning |
| CN112308008A (en) * | 2020-11-12 | 2021-02-02 | 电子科技大学 | Radar radiation source individual identification method based on working mode open set of transfer learning |
| CN115169469A (en) * | 2022-07-18 | 2022-10-11 | 中国人民解放军国防科技大学 | Radiation source individual identification method based on depth sub-field self-adaption |
| CN115600150A (en) * | 2022-09-26 | 2023-01-13 | 郑州大学（Cn） | Multi-mode gearbox fault diagnosis method based on deep migration learning |
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106815643A (en) * | 2017-01-18 | 2017-06-09 | 中北大学 | Infrared spectrum Model Transfer method based on random forest transfer learning |
| CN108508411A (en) * | 2018-03-22 | 2018-09-07 | 天津大学 | Passive radar external sort algorithm signal recognition method based on transfer learning |
| CN112308008A (en) * | 2020-11-12 | 2021-02-02 | 电子科技大学 | Radar radiation source individual identification method based on working mode open set of transfer learning |
| CN115169469A (en) * | 2022-07-18 | 2022-10-11 | 中国人民解放军国防科技大学 | Radiation source individual identification method based on depth sub-field self-adaption |
| CN115600150A (en) * | 2022-09-26 | 2023-01-13 | 郑州大学（Cn） | Multi-mode gearbox fault diagnosis method based on deep migration learning |
Non-Patent Citations (3)
| Title |
|---|
| YAROSLAV GANIN等: "Domain-Adversarial Training of Neural Networks", 《ARXIV:1505.07818V4》, pages 1 - 31 * |
| 王天池 等: "迁移学习及其在通信辐射源个体识别中的应用", 《通信技术》, vol. 55, no. 3, pages 265 - 273 * |
| 王满喜 等: "基于数据增强与集成学习的小样本识别技术", 《移动通信》, vol. 46, no. 7, pages 2 - 9 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110414368A (en) | A kind of unsupervised pedestrian recognition methods again of knowledge based distillation | |
| CN109871875B (en) | Building change detection method based on deep learning | |
| CN114564982B (en) | Automatic identification method for radar signal modulation type | |
| CN113705446B (en) | Open set identification method for individual radiation source | |
| CN110728694B (en) | Long-time visual target tracking method based on continuous learning | |
| CN107491729B (en) | Handwritten digit recognition method based on cosine similarity activated convolutional neural network | |
| CN105303169B (en) | A kind of cell division identification method and its identification device based on slow feature | |
| CN111931814B (en) | Unsupervised countering domain adaptation method based on intra-class structure tightening constraint | |
| CN111428817A (en) | Defense method for resisting attack by radio signal identification | |
| CN112347970A (en) | Remote sensing image ground object identification method based on graph convolution neural network | |
| CN108805102A (en) | A kind of video caption detection and recognition methods and system based on deep learning | |
| CN109376763A (en) | Sample classification method, system and medium based on multisample reasoning neural network | |
| CN114638256A (en) | Transformer fault detection method and system based on sound wave signals and attention network | |
| CN110048978A (en) | A kind of signal modulate method | |
| CN109753962A (en) | Text filed processing method in natural scene image based on hybrid network | |
| CN102722717B (en) | Cell division identification method | |
| CN116502139B (en) | Radiation source signal individual identification method based on integrated countermeasure migration | |
| CN110909645B (en) | Crowd counting method based on semi-supervised manifold embedding | |
| CN116502139A (en) | Radiation source signal individual identification method based on integrated countermeasure migration | |
| Feng et al. | FCGCN: Feature Correlation Graph Convolution Network for Few-Shot Individual Identification | |
| Zhang et al. | Remote sensing image retrieval based on DenseNet model and CBAM | |
| Shao et al. | Attention GAN-based method for designing intelligent making system | |
| Xia et al. | A shadow detection of remote sensing images based on statistical texture features | |
| CN114724245B (en) | Incremental learning human body action recognition method based on CSI | |
| Ding et al. | Surgical workflow recognition using two-stream mixed convolution network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |