CN117236448B - Radar intention reasoning and model training method based on time sequence knowledge graph - Google Patents
Radar intention reasoning and model training method based on time sequence knowledge graph Download PDFInfo
- Publication number
- CN117236448B CN117236448B CN202311515617.2A CN202311515617A CN117236448B CN 117236448 B CN117236448 B CN 117236448B CN 202311515617 A CN202311515617 A CN 202311515617A CN 117236448 B CN117236448 B CN 117236448B
- Authority
- CN
- China
- Prior art keywords
- radar
- time sequence
- layer
- graph
- knowledge graph
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000012549 training Methods 0.000 title claims abstract description 19
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 12
- 230000004913 activation Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000010030 laminating Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000005096 rolling process Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 abstract description 4
- 230000007774 longterm Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Abstract
The invention discloses a radar intention reasoning and model training method based on a time sequence knowledge graph, and belongs to the technical field of electronic countermeasure. Firstly, acquiring and preprocessing radar time sequence characteristic data; then constructing a knowledge graph as a data set by utilizing the characteristic data of the radar time sequence; then constructing a radar intention reasoning model based on the TemporalGCN; and finally, inputting part of the data set into the radar intention reasoning model to obtain a predicted value, and adjusting the model parameters through training until the obtained predicted value is closest to an actual value. The invention effectively utilizes the time sequence characteristics of the signals to acquire the dynamic evolution characteristics of the time sequence diagram so as to complete the intention reasoning of the multifunctional radar.
Description
Technical Field
The invention belongs to the technical field of electronic countermeasure, and particularly relates to a radar intention reasoning and model training method based on a time sequence knowledge graph.
Background
Aiming at more and more working modes and characteristic parameters of the novel radar, pulse description word forms used by the traditional radar intention recognition method are not comprehensive enough, and too dependent on priori information, and a matching template is relatively fixed, so that recognition of the novel radar working mode is not enough. Meanwhile, the multifunctional radar has the advantages of multiple parameters, flexible change, internal regularity of change of the working mode, and higher requirement on updating the priori knowledge base for the identification of the radar working mode.
The knowledge graph can construct a priori knowledge base, so that the data can be visualized, stored and utilized efficiently. Meanwhile, aiming at the characteristics that the parameters of the radiation source signals are increased and the changes are more flexible, the knowledge graph can store radar behavior knowledge and identify working modes by utilizing knowledge representation and reasoning technology, and the implicit information in the knowledge graph is deduced. In the prior art, the research on the knowledge graph is concentrated on static knowledge reasoning, the time sequence information of the knowledge graph is ignored, the knowledge graph cannot be dynamically updated, and the evolution process of the knowledge graph in time is difficult to realize by utilizing the dynamic information.
Disclosure of Invention
Aiming at the defects or improvement demands of the prior art, the invention provides a radar intention reasoning and model training method based on a time sequence knowledge graph, and aims to solve the technical problem that the prior radar identification method based on the knowledge graph ignores time sequence information.
To achieve the above object, in a first aspect, the present invention provides a method for training a radar intention inference model based on a time-series knowledge graph, the method comprising:
collecting and preprocessing radar time sequence characteristic data;
constructing a knowledge graph as a data set by utilizing the characteristic data of the radar time sequence;
constructing a radar intention reasoning model based on the TemporalGCN;
and inputting a part of data set into the radar intention reasoning model to obtain a predicted value, and adjusting the model parameters through training until the obtained predicted value is closest to an actual value.
Preferably, the knowledge graph is constructed by utilizing the characteristic data of the radar time sequence as a data set, specifically:
characterizing the radar time sequence characteristic data in a form of triples, wherein the triples comprise a head entity, a tail entity and a relation; the head entity describes radar operation modes including a VS mode, a TWS mode, a TAS mode and a STT mode; the relation is used for describing the relation between the head entity and the tail entity and comprises a mode type, a frequency view number, a bandwidth, pulse repetition frequency, a back-light, a duty ratio, a modulation mode, the number of pulses on each CPI and a pulse width; the tail entity is used to describe a specific numerical value of the relationship.
Preferably, the TemporalGCN includes a plurality of TemporalGCN layers, each TemporalGCN layer including two sequential convolution layers and a graph convolution network layer, the graph convolution network layer being located between the two sequential convolution layers.
Preferably, the time sequence convolution layer is constructed based on a 1D-CNN-GLU model, and the time sequence convolution layer is inputThe method comprises the following steps:
wherein,represent the firsttA graph representation of the moment knowledge graph; />，/>Is the time step of the time sequence convolution layer; />Representing the real number domain; />Is the length of the feature vector; />The number of nodes in the graph;
through the size ofDepth->1D-CNN operations along the time dimension, followed by GLU activation:
wherein,；/>and->Is a learning parameter; />Multiplying the matrix corresponding elements +.>Is a nonlinear activation function.
Preferably, the graph convolution network layer is configured to update a corresponding knowledge graph output by a previous time sequence convolution layer, where the updating is represented by:
wherein,and->Respectively represent +.>Layer and->Laminating layer of layer drawing>Is->Weight matrix of layer,/>Nonlinear activation function>Is a degree matrix->Is an adjacency matrix; upper energizer->Indicating that each node has added information about itself.
Preferably byThe function evaluates the difference between the predicted value and the actual value:
wherein,representation->The actual value of the moment, the subscript indicates the timing; />Representing the predicted value; by minimizingFunction training is carried out to adjust radar intention reasoning model parameters; />Representing the square of the norm.
Preferably, the pretreatment includes normalization treatment, specifically using the following formula:
wherein,is the characteristic data after normalization; />Is the collected characteristic data; />Is the minimum value of the feature data; />Is the maximum value of the characteristic data.
In a second aspect, the present invention provides a method for reasoning radar intention based on a time sequence knowledge graph, the method comprising:
acquiring and preprocessing time sequence characteristic data of a radar to be inferred in real time;
inputting the time sequence characteristic data into a radar intention reasoning model;
the model outputs the intention of the radar to be inferred;
the radar intent inference model is trained in accordance with the method of any one of the first aspects.
In a third aspect, the present invention provides an electronic device comprising: a memory for storing a program; a processor for executing a memory-stored program, the processor being for performing any of the methods described in the first aspect when the memory-stored program is executed.
In a fourth aspect, the present invention provides a storage medium storing a computer program which, when run on a processor, causes the processor to perform the method described in any of the first aspects.
In general, the above technical solutions conceived by the present invention have the following beneficial effects compared with the prior art:
the knowledge graph nodes are affected by neighborhood structure information, and characteristic information of the nodes also dynamically evolves continuously along with the development of time. The invention combines the advantages of 1D-CNN in time sequence extraction by utilizing the advantages of the knowledge graph in characteristic association, expresses the working mode and the characteristic parameters thereof in the form of the knowledge graph, and utilizes the signal time sequence characteristics to acquire the dynamic evolution characteristics of the time sequence graph so as to complete the intention reasoning of the multifunctional radar; by comparison experiments with other existing predictive models, it can be found that: traditional predictive models may perform well in short-term inference predictions, but their long-term inference predictions are not accurate due to error accumulation and lack of data correlation; the prediction method can give consideration to accurate prediction of short term and medium and long term, and has obvious advantages in the aspect of radar intention prediction.
Drawings
FIG. 1 is a flow chart of a radar intent inference method in an embodiment of the present invention;
FIG. 2 is a schematic diagram of a radar intent inference model based on TemporalGCN in an embodiment of the present invention;
FIG. 3 is a schematic diagram of a time-series convolution layer of a radar intent inference model in an embodiment of the present invention;
FIG. 4 is a graph showing the trend of the Loss curve during training of the radar intent inference model in an embodiment of the present invention;
FIG. 5 is a graph showing the trend of Acc curve in the training process of the radar intention inference model in the embodiment of the invention;
FIG. 6 is a graph of predictive performance versus various radar intent inference models in an embodiment of the invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
In embodiments of the invention, words such as "exemplary" or "such as" are used to mean serving as an example, instance, or illustration. Any embodiment or design described herein as "exemplary" or "e.g." in an embodiment should not be taken as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "such as" is intended to present related concepts in a concrete fashion.
Next, the technical scheme provided in the embodiment of the present invention is described.
As shown in fig. 1, an embodiment of the present invention includes the steps of:
s1, acquiring time sequence characteristic data and carrying out normalization processing;
performing minimum-maximum normalization processing on the acquired characteristics of the switching time sequence of the working mode of the multifunctional radar to accelerate the convergence of the model and eliminate the influence of singular data, so that the preprocessed data is limited on the [0,1] interval;
wherein,is the characteristic data after normalization; />Is the collected characteristic data; />Is the minimum value of the feature data; />Is the maximum value of the characteristic data.
S2, constructing a knowledge graph as a data set;
characterizing the normalized time sequence features in a form of a triplet (head, tail), wherein the head represents a head entity in the triplet, such as a VS working mode, a TAS working mode, etc.; tail represents the specific value of the characteristics such as frequency apparent number, bandwidth and duty cycle in the triplet; together, head and tail form a collection of entities; the relation represents the relation between the head and tail entities in the triples, such as frequency vision number, bandwidth, duty ratio and the like, and forms a relation set; the nodes in the same working mode are also connected. And constructing an intention reasoning knowledge graph spectrogram of a plurality of time steps. See in particular table 1:
TABLE 1
S3, constructing a time sequence convolution layer;
in order to capture the intention information of the radar timing diagram, a model structure of 1D-CNN-GLU is used on the time axis; parallel and controllable training through a layered evolution structure of layered representation; as shown in fig. 2, the sequential convolution layer inputs are:
wherein,represent the firsttA graph representation of the moment knowledge graph; />，/>Is the time step of the time sequence convolution layer; />Representing the real number domain; />Is the length of the feature vector; />The number of nodes in the graph;
through the size ofDepth->1D-CNN operations along the time dimension, followed by GLU activation:
wherein,；/>and->Is a learning parameter; />Multiplying the matrix corresponding elements +.>Is a nonlinear activation function;
s4, building a graph rolling network layer;
the graph convolution network layer characterizes the relation of each working mode node of each graph, and the layer performs aggregation operation on each time sequence graph to continuously update the representation of each node, thereby completing the update of one time step time sequence graph; the expression is as follows:
wherein,and->Respectively represent +.>Layer and->Laminating layer of layer drawing>Is->Weight matrix of layer,/>Nonlinear activation function>Is a degree matrix->Is an adjacency matrix; upper energizer->Information indicating that each node has added itself to it;
s5, constructing a TemporalGCN network;
the TemporalGCN network consists of a plurality of TemporalGCN layers, and each TemporalGCN layer consists of two time sequence convolution layers and a graph convolution network layer positioned between the time sequence convolution layers; finally, a full connection layer is connected;
as shown in FIG. 3, for each TemporalGCN layer, the thTemporalGCN layer inputOutput is +.>Then by graph-rolling the network layer +.>Corresponding->Updating the knowledge graph, and outputting ++again through a time sequence convolution layer>The method comprises the steps of carrying out a first treatment on the surface of the Similarly, a TemporalGCN layer is adopted and then is sent into a fully connected network to obtain a final predicted value +.>；
S6, performing network training;
sending a part of the data set constructed in S2 into a TemporalGCN network to obtain a predicted value which is as close to an actual value as possible, and passingThe function evaluates the difference between the predicted value and the actual value:
wherein,representing an actual value, and subscripts representing timing marks; />Representing the predicted value; by minimizing +.>Function training is carried out to adjust radar intention reasoning model parameters, so that the reasoning effect of the network is optimal; />Representing the square of the norm;
s7, intention reasoning;
and (3) inputting the rest data sets constructed in the step (S2) into a network, and recording an output result of the network, namely a prediction result of the TemporalGCN network model.
The reasoning performance of the model of the invention is verified by a comparison experiment:
constructing a prediction sequence pre-len with the length of 5 by adopting an S2 method, inputting 3600 training sets with the length seq-len of 65 into a radar intention reasoning model for training, and inputting 900 seq-len 65 test sets for testing after obtaining an optimal radar intention reasoning model; as shown in fig. 4 and 5, the Loss and Acc curves eventually stabilize as the iteration number increases for the TemporalGCN-based radar intent inference model.
Based on the data set, comparing the radar intention inference model based on TemporalGCN with the traditional inference prediction models GRU and LSTM in performance:
as can be seen from fig. 6, for the data set, when the input sequence is 65, the accuracy of the radar intention reasoning model reasoning prediction based on the TemporalGCN is optimal; the model provided by the invention obtains the best performance in the evaluation index.
At the same time, traditional model models may perform well in short-term inference predictions, but their long-term inference predictions are not accurate due to error accumulation and lack of data correlation.
The invention also realizes a radar intention reasoning method based on the time sequence knowledge graph.
The method comprises the following steps:
acquiring and preprocessing time sequence characteristic data of a radar to be inferred in real time;
inputting the time sequence characteristic data into a radar intention reasoning model;
the model outputs the intention of the radar to be inferred;
the radar intention inference model is trained according to the method in the above embodiment, and will not be described herein.
Based on the method in the above embodiment, the embodiment of the invention provides an electronic device. The apparatus may include: a memory for storing a program and a processor for executing the program stored by the memory. Wherein the processor is adapted to perform the method described in the above embodiments when the program stored in the memory is executed.
Based on the method in the above embodiment, the embodiment of the present invention provides a storage medium storing a computer program, which when executed on a processor causes the processor to perform the method in the above embodiment.
It is to be appreciated that the processor in embodiments of the invention may be a central processing unit (centralprocessing unit, CPU), other general purpose processor, digital signal processor (digital signalprocessor, DSP), application specific integrated circuit (application specific integrated circuit, ASIC), field programmable gate array (field programmable gate array, FPGA) or other programmable logic device, transistor logic device, hardware components, or any combination thereof. The general purpose processor may be a microprocessor, but in the alternative, it may be any conventional processor.
The method steps in the embodiments of the present invention may be implemented by hardware, or may be implemented by executing software instructions by a processor. The software instructions may be comprised of corresponding software modules that may be stored in random access memory (random access memory, RAM), flash memory, read-only memory (ROM), programmable ROM (PROM), erasable programmable PROM (EPROM), electrically erasable programmable EPROM (EEPROM), registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, produces a flow or function in accordance with embodiments of the present invention, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in a storage medium or transmitted over the storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital Subscriber Line (DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). The storage media may be any available media that can be accessed by a computer or a data storage device such as a server, data center, or the like that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a Solid State Disk (SSD)), or the like.
It will be appreciated that the various numerical numbers referred to in the embodiments of the present invention are merely for ease of description and are not intended to limit the scope of the embodiments of the present invention.
It will be readily appreciated by those skilled in the art that the foregoing is merely a preferred embodiment of the invention and is not intended to limit the invention, but any modifications, equivalents, improvements or alternatives falling within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (6)
1. A training method of a radar intention reasoning model based on a time sequence knowledge graph is characterized by comprising the following steps:
collecting and preprocessing radar time sequence characteristic data;
constructing a knowledge graph as a data set by utilizing the characteristic data of the radar time sequence;
constructing a radar intention reasoning model based on the TemporalGCN;
inputting a part of data set into the radar intention reasoning model to obtain a predicted value, and adjusting the model parameters through training until the obtained predicted value is closest to an actual value;
the method comprises the steps of constructing a knowledge graph as a data set by utilizing radar time sequence characteristic data, wherein the knowledge graph is specifically as follows:
characterizing the radar time sequence characteristic data in a form of triples, wherein the triples comprise a head entity, a tail entity and a relation; the head entity describes radar operation modes including a VS mode, a TWS mode, a TAS mode and a STT mode; the relation is used for describing the relation between the head entity and the tail entity and comprises a mode type, a frequency view number, a bandwidth, pulse repetition frequency, a back-light, a duty ratio, a modulation mode, the number of pulses on each CPI and a pulse width; the tail entity is used for describing specific numerical values of the relation;
the TemporalGCN comprises a plurality of TemporalGCN layers, each TemporalGCN layer comprises two time sequence convolution layers and a graph convolution network layer, and the graph convolution network layer is positioned between the two time sequence convolution layers; finally, a full connection layer is connected; for each TemporalGCN layer, the thTemporalGCN layer input +.>Output is +.>Then by graph-rolling the network layer +.>Corresponding toUpdating the knowledge graph, and outputting ++again through a time sequence convolution layer>The method comprises the steps of carrying out a first treatment on the surface of the Similarly, a TemporalGCN layer is adopted and then is sent into a fully connected network to obtain a final predicted value +.>；
Constructing the time sequence convolution layer based on a 1D-CNN-GLU model, and inputting the time sequence convolution layerThe method comprises the following steps:
wherein,represent the firsttA graph representation of the moment knowledge graph; />，/>Is the time step of the time sequence convolution layer; />Representing the real number domain; />Is the length of the feature vector; />The number of nodes in the graph;
through the size ofDepth->1D-CNN operations along the time dimension, followed by GLU activation:
wherein,；/>and->Is a learning parameter; />Multiplying the matrix corresponding elements +.>Is a nonlinear activation function;
the graph convolution network layer is used for updating the corresponding knowledge graph outputted by the previous time sequence convolution layer, and the graph convolution network layer is expressed as follows:
wherein,and->Respectively represent +.>Layer and->Laminating layer of layer drawing>Is->Weight matrix of layer,/>Nonlinear activation function>Is a degree matrix->Is an adjacency matrix; upper energizer->Indicating that each node has added information about itself.
2. The method according to claim 1, characterized by the fact that byThe function evaluates the difference between the predicted value and the actual value:
wherein,representation->The actual value of the moment, the subscript indicates the timing; />Representing the predicted value; by minimizing +.>Function training is carried out to adjust radar intention reasoning model parameters; />Representing the square of the norm.
3. The method according to claim 1, wherein the pre-processing comprises a normalization process, in particular normalized by the formula:
wherein,is the characteristic data after normalization; />Is the collected characteristic data; />Is the minimum value of the feature data;is the maximum value of the characteristic data.
4. A method for radar intent inference based on a time-series knowledge graph, the method comprising:
acquiring and preprocessing time sequence characteristic data of a radar to be inferred in real time;
inputting the time sequence characteristic data into a radar intention reasoning model;
the model outputs the intention of the radar to be inferred;
the radar intent inference model is trained in accordance with the method of any one of claims 1-3.
5. An electronic device, comprising:
a memory for storing a program;
a processor for executing the program stored in the memory, which processor is adapted to perform the method of any one of claims 1-3 or to perform the method of claim 4 when the program stored in the memory is executed.
6. A storage medium storing a computer program, characterized in that the computer program, when run on a processor, causes the processor to perform the method of any one of claims 1-3 or to perform the method of claim 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311515617.2A CN117236448B (en) | 2023-11-15 | 2023-11-15 | Radar intention reasoning and model training method based on time sequence knowledge graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311515617.2A CN117236448B (en) | 2023-11-15 | 2023-11-15 | Radar intention reasoning and model training method based on time sequence knowledge graph |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117236448A CN117236448A (en) | 2023-12-15 |
| CN117236448B true CN117236448B (en) | 2024-02-09 |
Family
ID=89095307
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311515617.2A Active CN117236448B (en) | 2023-11-15 | 2023-11-15 | Radar intention reasoning and model training method based on time sequence knowledge graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117236448B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112036556A (en) * | 2020-11-06 | 2020-12-04 | 西南交通大学 | Target intention inversion method based on LSTM neural network |
| CN112543936A (en) * | 2020-10-29 | 2021-03-23 | 香港应用科技研究院有限公司 | Motion structure self-attention-seeking convolutional network for motion recognition |
| WO2021218424A1 (en) * | 2020-04-30 | 2021-11-04 | 江苏科技大学 | Rbf neural network-based method for sea surface wind speed inversion from marine radar image |
| CN113742491A (en) * | 2021-08-12 | 2021-12-03 | 上海熙业信息科技有限公司 | Representation learning-based time knowledge graph reasoning method |
| CN114048849A (en) * | 2021-08-31 | 2022-02-15 | 赵浩新 | Unsupervised representation learning method and device supporting time sequence social network diagram |
| CN115422373A (en) * | 2022-09-05 | 2022-12-02 | 中国人民解放军国防科技大学 | Knowledge graph and user intention task decomposition method for satellite task planning |
| CN115542318A (en) * | 2022-10-12 | 2022-12-30 | 南京航空航天大学 | Air-ground combined multi-domain detection system and method for unmanned aerial vehicle group target |
| CN115856811A (en) * | 2022-11-05 | 2023-03-28 | 西安电子工程研究所 | Micro Doppler feature target classification method based on deep learning |
| CN116682553A (en) * | 2023-08-02 | 2023-09-01 | 浙江大学 | Diagnosis recommendation system integrating knowledge and patient representation |
| CN116911377A (en) * | 2023-06-30 | 2023-10-20 | 中国电子科技集团公司第十研究所 | Radiation source individual identification method, equipment and medium based on transfer learning |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP4094194A1 (en) * | 2020-01-23 | 2022-11-30 | Umnai Limited | An explainable neural net architecture for multidimensional data |
| US20230230484A1 (en) * | 2022-01-18 | 2023-07-20 | The Regents Of The University Of California | Methods for spatio-temporal scene-graph embedding for autonomous vehicle applications |
| US20230305136A1 (en) * | 2022-03-25 | 2023-09-28 | Mitsubishi Electric Research Laboratories, Inc. | System and Method for Radar Object Recognition with Cross-Frame Temporal Relationality |
-
2023
- 2023-11-15 CN CN202311515617.2A patent/CN117236448B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021218424A1 (en) * | 2020-04-30 | 2021-11-04 | 江苏科技大学 | Rbf neural network-based method for sea surface wind speed inversion from marine radar image |
| CN112543936A (en) * | 2020-10-29 | 2021-03-23 | 香港应用科技研究院有限公司 | Motion structure self-attention-seeking convolutional network for motion recognition |
| CN112036556A (en) * | 2020-11-06 | 2020-12-04 | 西南交通大学 | Target intention inversion method based on LSTM neural network |
| CN113742491A (en) * | 2021-08-12 | 2021-12-03 | 上海熙业信息科技有限公司 | Representation learning-based time knowledge graph reasoning method |
| CN114048849A (en) * | 2021-08-31 | 2022-02-15 | 赵浩新 | Unsupervised representation learning method and device supporting time sequence social network diagram |
| CN115422373A (en) * | 2022-09-05 | 2022-12-02 | 中国人民解放军国防科技大学 | Knowledge graph and user intention task decomposition method for satellite task planning |
| CN115542318A (en) * | 2022-10-12 | 2022-12-30 | 南京航空航天大学 | Air-ground combined multi-domain detection system and method for unmanned aerial vehicle group target |
| CN115856811A (en) * | 2022-11-05 | 2023-03-28 | 西安电子工程研究所 | Micro Doppler feature target classification method based on deep learning |
| CN116911377A (en) * | 2023-06-30 | 2023-10-20 | 中国电子科技集团公司第十研究所 | Radiation source individual identification method, equipment and medium based on transfer learning |
| CN116682553A (en) * | 2023-08-02 | 2023-09-01 | 浙江大学 | Diagnosis recommendation system integrating knowledge and patient representation |
Non-Patent Citations (2)
| Title |
|---|
| TimeTraveler:Reinforcement Learning for Temporal Knowledge Graph Forecasting;Haohai Sun等;《arXiv》;第1-14页 * |
| 相控阵雷达工作模式识别与意图推理技术研究;田卫东;《中国优秀硕士学位论文全文数据库信息科技辑》(第07期);第I136-816页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117236448A (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11403479B2 (en) | Feedback signaling to facilitate data classification functionality of a spiking neural network | |
| US10387769B2 (en) | Hybrid memory cell unit and recurrent neural network including hybrid memory cell units | |
| US20190050710A1 (en) | Adaptive bit-width reduction for neural networks | |
| WO2019232851A1 (en) | Method and apparatus for training speech differentiation model, and computer device and storage medium | |
| WO2019072166A1 (en) | Semantic analysis method, device, and storage medium | |
| WO2019227574A1 (en) | Voice model training method, voice recognition method, device and equipment, and medium | |
| CN109242099B (en) | Training method and device of reinforcement learning network, training equipment and storage medium | |
| WO2021089012A1 (en) | Node classification method and apparatus for graph network model, and terminal device | |
| WO2021089013A1 (en) | Spatial graph convolutional network training method, electronic device and storage medium | |
| US20210081763A1 (en) | Electronic device and method for controlling the electronic device thereof | |
| CN112990444B (en) | Hybrid neural network training method, system, equipment and storage medium | |
| CN109191276A (en) | A kind of P2P network loan institutional risk appraisal procedure based on intensified learning | |
| CN110874635B (en) | Deep neural network model compression method and device | |
| WO2022012233A1 (en) | Method and computing apparatus for quantification calibration, and computer-readable storage medium | |
| CN113221758B (en) | GRU-NIN model-based underwater sound target identification method | |
| CN117236448B (en) | Radar intention reasoning and model training method based on time sequence knowledge graph | |
| Basterrech et al. | Evolutionary Echo State Network: A neuroevolutionary framework for time series prediction | |
| Qi et al. | Hyperparameter optimization of neural networks based on Q-learning | |
| KR20200000660A (en) | System and method for generating prediction model for real-time time-series data | |
| CN116911377A (en) | Radiation source individual identification method, equipment and medium based on transfer learning | |
| Zhang et al. | Generalized maximum correntropy-based echo state network for robust nonlinear system identification | |
| CN115169740A (en) | Sequence prediction method and system of pooled echo state network based on compressed sensing | |
| CN114897274A (en) | Method and system for improving time sequence prediction effect | |
| US20210374549A1 (en) | Meta-learned, evolution strategy black box optimization classifiers | |
| CN112951270B (en) | Voice fluency detection method and device and electronic equipment |
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 | ||
| GR01 | Patent grant |