CN111723997A - Automatic generation method of urban major traffic accident data sample based on GAN - Google Patents
Automatic generation method of urban major traffic accident data sample based on GAN Download PDFInfo
- Publication number
- CN111723997A CN111723997A CN202010595276.4A CN202010595276A CN111723997A CN 111723997 A CN111723997 A CN 111723997A CN 202010595276 A CN202010595276 A CN 202010595276A CN 111723997 A CN111723997 A CN 111723997A
- Authority
- CN
- China
- Prior art keywords
- network
- generating
- data
- max
- generated
- 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.)
- Pending
Links
- 206010039203 Road traffic accident Diseases 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 26
- 230000002068 genetic effect Effects 0.000 claims abstract description 16
- 239000013598 vector Substances 0.000 claims description 14
- 210000002569 neuron Anatomy 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 9
- 230000035772 mutation Effects 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 3
- 230000002123 temporal effect Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 2
- 238000013136 deep learning model Methods 0.000 abstract description 3
- 230000007614 genetic variation Effects 0.000 abstract description 3
- 230000003042 antagnostic effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
-
- 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
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Analytical Chemistry (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Strategic Management (AREA)
- Computing Systems (AREA)
- Chemical & Material Sciences (AREA)
- Molecular Biology (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Development Economics (AREA)
- Genetics & Genomics (AREA)
- Game Theory and Decision Science (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a method for automatically generating urban major traffic accident data samples based on GAN, which aims to automatically generate the urban major traffic accident data samples, thereby providing a large amount of data for learning and training a deep learning model, improving the accuracy of the model for predicting traffic accidents and further recovering a large amount of loss caused by the traffic accidents. The genetic variation-antagonistic generation network in the present invention is mainly composed of two parts, one is a generation network and the other is an identification network. The invention uses the genetic variation algorithm for generating the false data sample in combination with the generation network for the first time, and simultaneously, the anti-generation network is also used for generating the serious traffic accident sample for the first time.
Description
Technical Field
The invention relates to a method for generating urban major traffic accident data samples based on a genetic variation-confrontation generation network, and belongs to the technical field of intelligent traffic application.
Background
Major traffic accidents, particularly car accidents, are a huge problem all over the world, with about 130 million people dying from car traffic accidents and about 5 million people injured during car traffic accidents each year all over the world. There are large data based machine learning techniques to predict traffic accidents ahead of time to minimize losses. As is well known, the deep learning technology needs a large number of traffic accident samples, and the more the traffic accident data, the higher the accuracy of the model. However, the number of the major traffic accidents occurring and recorded is not large, certain privacy is involved, and meanwhile, compared with a negative sample (i.e. no major traffic accident occurs), the number of the positive samples (the occurring major traffic accident) is too small, the positive samples belong to unbalanced samples, so that the training precision of the model is greatly influenced, how to increase the number of the major traffic accident samples and provide the samples for the model to learn is how to increase, and therefore, the prediction accuracy of the major traffic accident is improved, which is a very important problem. Therefore, the method has important value and significance for automatically generating the urban major traffic accident data sample through the artificial intelligence algorithm.
Disclosure of Invention
The purpose of the invention is: and the urban major traffic accident data samples are automatically generated, so that a large amount of data is provided for learning and training the deep learning model, the accuracy of the model for predicting the traffic accidents is improved, and a large amount of loss caused by the traffic accidents is saved.
In order to achieve the above object, the technical solution of the present invention is to provide an automatic generation method of an urban major traffic accident data sample based on GAN, which is characterized by comprising the following steps:
step 1, defining urban traffic accident characteristics as four dimensions: weather characteristic, time characteristic, road characteristic and vehicle type characteristic, and carrying out unique hot coding on the characteristics of each accident sample in the real and important traffic accident sample set to obtain the characteristicsEigenvectors as a real data sample library X ═ { X1,x2,…,xi,…,xm},xiRepresenting the ith feature vector;
step 2, generating an initial fake data set Z ═ { Z ═ Z1,z2,…,zi,…,zm},ziA feature vector representing an ith false significant traffic accident sample;
step 3, establishing a generating network and an identifying network, wherein the generating network and the identifying network are neural networks, and the generating network and the identifying network are as follows: the number of output layer neurons and the number of input layer neurons of the generated network are the same; the input layer of the identification network has real data and data generated by the generation network, the output layer of the identification network has a neuron, and the output result is a value; then initializing parameters of the generation network and the authentication network, the parameters of the generation network being thetadAnd the parameter of the authentication network is thetag;
step 4, generating a false data set Z ═ { Z ═ from step 21,z2,…,zi,…,zmSampling n samples, inputting the samples into a generating network, and calculating a result generated by the generating network at the moment:
and step 5, expressing an objective function V of the authentication network as follows:
in the formula, D (x)i) Is the output value of the real data;generating an output value of false data generated by the network in the authentication network;
the weight parameters are updated by a gradient ascent method with the goal of maximizing the objective function V: where η denotes a Learning Rate (Learning Rate), which represents the length of each parameter update step,representing the parameter thetagA derivative of (a);
step 6, generating a false data set Z ═ { Z ═ from step 21,z2,…,zi,…,zmN samples are sampled, and the result generated by the network at that time is calculatedAccording to the objective function of the generated networkThe weight parameters are updated by a gradient-ascending method:in the formula (I), the compound is shown in the specification,representing the parameter thetagA derivative of (a);
and 7, repeating the steps 4 to 6 until the generation network and the identification network are converged, obtaining the trained confrontation generation network, and generating the major traffic accident data after the confrontation generation network is trained.
Preferably, in step 1: the time characteristics comprise whether the time is morning and evening peak time, whether the time is a commute day, whether the time is a holiday, and which month in a year; the road characteristics include a distance to a nearest intersection, a road width, and a road tortuosity.
Preferably, in step 2, when the false data set Z is generated, the exploration rate a% is set, and then, a% of data in the initial false data set Z is generated according to a mutation algorithm in a genetic algorithm, and (100-a)% of data is completely randomly generated.
Preferably, the mutation algorithm in the genetic algorithm comprises the steps of:
step 2.1, respectively counting the mean value, the minimum value and the maximum value of the four dimensional characteristics in the real data sample library X to obtain the mean value mu of the weather characteristic characteristics1Minimum value min1Max, max1(ii) a Mean value mu of the temporal characteristic2Minimum value min2Max, max2(ii) a Mean value mu of road characteristic features3Minimum value min3Max, max3(ii) a Mean value mu of characteristic features of type of vehicle4Minimum value min4Max, max4;
Step 2.2, a mutation algorithm in the genetic algorithm: and (3) according to a Gaussian distribution rule and a 3 sigma theorem, obtaining the mean value mu and the variance (max-min)/3 of 60% of data in the generated false data set Z according to the mean value, the minimum value and the maximum value of the four dimensional characteristics obtained in the step 2.1.
Preferably, in step 3, the generating network and the identifying network are neural networks with hidden layers of 5 layers and each hidden layer contains 4 neurons.
The invention discloses a method for generating urban major traffic accident data samples based on a genetic variation-confrontation generation network (GAN), and aims to automatically generate the urban major traffic accident data samples, so that a large amount of data is provided for learning and training a deep learning model, the accuracy of the model for predicting traffic accidents is improved, and a large amount of loss caused by the traffic accidents is saved. The method defines the urban traffic accident characteristics as four dimensions: the method comprises the steps of obtaining a characteristic vector (vector) as a real data sample library after one-hot coding (one-hot) is carried out on the characteristics of each accident sample according to weather characteristics, time characteristics, road characteristics and vehicle type characteristics. The genetic variation-confrontation generation network mainly comprises two parts, namely a generation network and an identification network, wherein the genetic variation-generation network (Generator) is adopted to generate a false serious traffic accident sample, the identification network (Discrimentor) is used to score the real data and the generated data, the result is fed back to the generation network (Generator), the generated data is improved by the generation network (Generator), and then the identification network is used to score, so that a good generation sample can be obtained after continuous game updating. The method is characterized in that a genetic variation algorithm is firstly used for generating false data samples by combining with a generation network, and meanwhile, the countergeneration network is also firstly used for generating serious traffic accident samples.
Drawings
FIG. 1 is a flow chart of the present invention;
fig. 2 is a schematic diagram of an authentication network.
Detailed Description
The invention will be further illustrated with reference to the following specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.
The invention provides a method for intelligently and automatically generating an urban accident data sample. The method defines the urban traffic accident characteristics as four dimensions: weather characteristics, time characteristics (whether the time is morning and evening rush hour, whether the time is commuting day, whether the time is holiday, which month in a year), road characteristics (distance from the nearest intersection, road width and road tortuosity) and vehicle type characteristics, and a characteristic vector (vector) is obtained after the characteristic of each accident sample is subjected to one-hot coding (one-hot) and is used as a real data sample library.
The genetic variation-countermeasure generation network mainly comprises two parts, one part is a generation network, the other part is a countermeasure network, a false major traffic accident sample is generated by adopting the genetic variation-generation network (Generator), the identification network (Discrimentor) is used for scoring the real data and the generated data, the result is fed back to the generation network (Generator), the generated data is improved by the generation network (Generator), and then the identification network is used for scoring, so that a good generation sample can be obtained after continuous game updating.
The invention firstly uses a genetic variation algorithm to be combined with a generation network to generate a fake data sample, and simultaneously uses an antagonistic generation network to be also used for automatically generating a serious traffic accident sample for the first time, and the method specifically comprises the following steps:
step 1, defining urban traffic accident characteristics as four dimensions: weather characteristics, time characteristics (whether it is morning and evening rush hour, whether it is commuting day, whether it is holiday, which month of the year), road characteristics (distance to nearest intersection, road width, road tortuosity), vehicle type characteristics. Performing one-hot coding (one-hot) on the characteristics of each accident sample in the real major traffic accident sample set to obtain a characteristic vector, and using the characteristic vector as a real data sample library X ═ X1,x2,…,xi,…,xm},xiRepresenting the ith feature vector.
Step 2. generating an initial fake data set Z ═ { Z ═ Z1,z2,…,zi,…,zm},ziFeature vectors representing the ith false significant traffic accident sample: if the exploration rate (exploration rate) is set, for example, equal to 60%, then the algorithm for generating 60% of the data in the initial fake data set Z generates according to the mutation algorithm in the genetic algorithm, and 40% of the data generates completely randomly.
The diversity algorithm in the genetic algorithm is implemented as follows:
step 2.1, respectively counting the mean value, the minimum value and the maximum value of the four dimensional characteristics in the real data sample library X to obtain the mean value mu of the weather characteristic characteristics1Minimum value min1Max, max1(ii) a Mean value mu of the temporal characteristic2Minimum value min2Max, max2(ii) a Mean value mu of road characteristic features3Minimum value min3Max, max3(ii) a Mean value mu of characteristic features of type of vehicle4Minimum value min4Max, max4;
Step 2.2 mutation algorithm in genetic algorithm: and (3) according to a Gaussian distribution rule and a 3 sigma theorem, obtaining the mean value mu and the variance (max-min)/3 of 60% of data in the generated false data set Z according to the mean value, the minimum value and the maximum value of the four dimensional characteristics obtained in the step 2.1.
Step 3. GenerationThe network and the discrimination network are respectively a neural network with hidden layers of 5 layers and each hidden layer containing 4 neurons. The number of output layer neurons and input layer neurons of the generated network is the same. The input layer of the discrimination network has real data and data generated by the generation network, the output layer of the discrimination network has a neuron, and the output result is a value, as shown in fig. 2. The training network updates the weight parameters of the generating network and the identifying network. Initializing parameters of a generating network and an identifying network, the parameters of the generating network being thetadAnd the parameter for authenticating the network is thetag.
Step 4. from the false data set Z ═ { Z ] generated in step 21,z2,…,zi,…,zmN samples are sampled and input into the generating network, and the generating network at that time is calculated (theta)g0) The generated result is as follows:
and step 5, expressing an objective function V of the authentication network as follows:
in the formula, D (x)i) Is the output value of the real data, and the larger the value is, the better the value is for the authentication network;is the output value of the false data generated by the generation network in the authentication network, and the lower the value is, the better the value is for the authentication network;
so the goal is to maximize the objective function V, the weight parameters are updated by a gradient-ascending method: where η denotes a Learning Rate (Learning Rate), which represents the length of each parameter update step,representing the parameter thetagThe derivative of (c).
Step 6. from the false data set Z ═ { Z ] generated in step 21,z2,…,zi,…,zmN samples are sampled, and the net generated at this time is calculated (theta)g) The result of the generationAccording to the objective function of the generated networkThe weight parameters are updated by a gradient-ascending method: in the formula (I), the compound is shown in the specification,representing the parameter thetagThe derivative of (c).
And 7, repeating the steps 4 to 6 until the generation network and the identification network are converged, obtaining the trained confrontation generation network, and generating the data of the major traffic accident after the confrontation generation network is trained.
Claims (5)
1. A method for automatically generating an urban major traffic accident data sample based on GAN is characterized by comprising the following steps:
step 1, defining urban traffic accident characteristics as four dimensions: the method comprises the steps of carrying out unique hot coding on the characteristics of each accident sample in a real major traffic accident sample set by using weather characteristics, time characteristics, road characteristics and vehicle type characteristics to obtain a characteristic vector, and using the characteristic vector as a real data sample library X ═ X1,x2,...,xi,...,xm},xiRepresenting the ith feature vector;
step 2Generating an initial fake data set Z ═ { Z ═ Z1,z2,...,zi,...,zm},ziA feature vector representing an ith false significant traffic accident sample;
step 3, establishing a generating network and an identifying network, wherein the generating network and the identifying network are neural networks, and the generating network and the identifying network are as follows: the number of output layer neurons and the number of input layer neurons of the generated network are the same; the input layer of the identification network has real data and data generated by the generation network, the output layer of the identification network has a neuron, and the output result is a value; then initializing parameters of the generation network and the authentication network, the parameters of the generation network being thetadAnd the parameter of the authentication network is thetag;
step 4, generating a false data set Z ═ { Z ═ from step 21,z2,...,zi,...,zmSampling n samples, inputting the samples into a generating network, and calculating a result generated by the generating network at the moment:
and step 5, expressing an objective function V of the authentication network as follows:
in the formula, D (x)i) Is the output value of the real data;generating an output value of false data generated by the network in the authentication network;
the weight parameters are updated by a gradient ascent method with the goal of maximizing the objective function V: in the formula, η represents a learning rate (bear)A ning Rate), representing the length of each parameter update step,representing the parameter thetagA derivative of (a);
step 6, generating a false data set Z ═ { Z ═ from step 21,z2,...,zi,...,zmN samples are sampled, and the result generated by the network at that time is calculatedAccording to the objective function of the generated networkThe weight parameters are updated by a gradient-ascending method:in the formula (I), the compound is shown in the specification,representing the parameter thetagA derivative of (a);
and 7, repeating the steps 4 to 6 until the generation network and the identification network are converged, obtaining the trained confrontation generation network, and generating the major traffic accident data after the confrontation generation network is trained.
2. The method for automatically generating the GAN-based urban mass transit accident data sample according to claim 1, wherein in the step 1: the time characteristics comprise whether the time is morning and evening peak time, whether the time is a commute day, whether the time is a holiday, and which month in a year; the road characteristics include a distance to a nearest intersection, a road width, and a road tortuosity.
3. The method as claimed in claim 1, wherein in the step 2, when the false data set Z is generated, the exploration rate a% is set, and then a% of data in the initial false data set Z is generated according to a variation algorithm in a genetic algorithm, and (100-a)% of data is completely randomly generated.
4. The method for automatically generating the GAN-based urban mass transit accident data sample as claimed in claim 1, wherein the mutation algorithm in the genetic algorithm comprises the following steps:
step 2.1, respectively counting the mean value, the minimum value and the maximum value of the four dimensional characteristics in the real data sample library X to obtain the mean value mu of the weather characteristic characteristics1Minimum value min1Max, max1(ii) a Mean value mu of the temporal characteristic2Minimum value min2Max, max2(ii) a Mean value mu of road characteristic features3Minimum value min3Max, max3(ii) a Mean value mu of characteristic features of type of vehicle4Minimum value min4Max, max4;
Step 2.2, a mutation algorithm in the genetic algorithm: and (3) according to a Gaussian distribution rule and a 3 sigma theorem, obtaining the mean value mu and the variance (max-min)/3 of 60% of data in the generated false data set Z according to the mean value, the minimum value and the maximum value of the four dimensional characteristics obtained in the step 2.1.
5. The method according to claim 1, wherein in step 3, the generating network and the discriminating network are neural networks with hidden layers of 5 layers and each hidden layer contains 4 neurons.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010595276.4A CN111723997A (en) | 2020-06-23 | 2020-06-23 | Automatic generation method of urban major traffic accident data sample based on GAN |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010595276.4A CN111723997A (en) | 2020-06-23 | 2020-06-23 | Automatic generation method of urban major traffic accident data sample based on GAN |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111723997A true CN111723997A (en) | 2020-09-29 |
Family
ID=72569191
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010595276.4A Pending CN111723997A (en) | 2020-06-23 | 2020-06-23 | Automatic generation method of urban major traffic accident data sample based on GAN |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111723997A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112529114A (en) * | 2021-01-13 | 2021-03-19 | 北京云真信科技有限公司 | Target information identification method based on GAN, electronic device and medium |
CN113434459A (en) * | 2021-06-30 | 2021-09-24 | 电子科技大学 | Network-on-chip task mapping method based on generation of countermeasure network |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108074244A (en) * | 2017-09-07 | 2018-05-25 | 汉鼎宇佑互联网股份有限公司 | A kind of safe city wagon flow statistical method for merging deep learning and Background difference |
WO2018122585A1 (en) * | 2016-12-30 | 2018-07-05 | 同济大学 | Method for urban road traffic incident detecting based on floating-car data |
CN109493599A (en) * | 2018-11-16 | 2019-03-19 | 南京航空航天大学 | A kind of Short-time Traffic Flow Forecasting Methods based on production confrontation network |
US20190114511A1 (en) * | 2017-10-16 | 2019-04-18 | Illumina, Inc. | Deep Learning-Based Techniques for Training Deep Convolutional Neural Networks |
WO2019145018A1 (en) * | 2018-01-23 | 2019-08-01 | Siemens Aktiengesellschaft | System, device and method for detecting abnormal traffic events in a geographical location |
DE102018201290A1 (en) * | 2018-01-29 | 2019-08-01 | Osram Gmbh | METHOD FOR PRODUCING A COMPRESSED VERSION OF AN ARTIFICIAL NEURONAL NETWORK AND DEVICE AND STORAGE MEDIUM |
US20190273510A1 (en) * | 2018-03-01 | 2019-09-05 | Crowdstrike, Inc. | Classification of source data by neural network processing |
US20190325264A1 (en) * | 2018-04-24 | 2019-10-24 | Here Global B.V. | Machine learning a feature detector using synthetic training data |
CN110458370A (en) * | 2019-08-19 | 2019-11-15 | 南京航空航天大学 | A kind of urban roads safety prediction technique and system based on open data |
CN111079640A (en) * | 2019-12-09 | 2020-04-28 | 合肥工业大学 | Vehicle type identification method and system based on automatic amplification sample |
CN111231983A (en) * | 2020-01-14 | 2020-06-05 | 浙江吉利汽车研究院有限公司 | Vehicle control method, device and equipment based on traffic accident memory network |
-
2020
- 2020-06-23 CN CN202010595276.4A patent/CN111723997A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018122585A1 (en) * | 2016-12-30 | 2018-07-05 | 同济大学 | Method for urban road traffic incident detecting based on floating-car data |
CN108074244A (en) * | 2017-09-07 | 2018-05-25 | 汉鼎宇佑互联网股份有限公司 | A kind of safe city wagon flow statistical method for merging deep learning and Background difference |
US20190114511A1 (en) * | 2017-10-16 | 2019-04-18 | Illumina, Inc. | Deep Learning-Based Techniques for Training Deep Convolutional Neural Networks |
WO2019145018A1 (en) * | 2018-01-23 | 2019-08-01 | Siemens Aktiengesellschaft | System, device and method for detecting abnormal traffic events in a geographical location |
DE102018201290A1 (en) * | 2018-01-29 | 2019-08-01 | Osram Gmbh | METHOD FOR PRODUCING A COMPRESSED VERSION OF AN ARTIFICIAL NEURONAL NETWORK AND DEVICE AND STORAGE MEDIUM |
US20190273510A1 (en) * | 2018-03-01 | 2019-09-05 | Crowdstrike, Inc. | Classification of source data by neural network processing |
US20190325264A1 (en) * | 2018-04-24 | 2019-10-24 | Here Global B.V. | Machine learning a feature detector using synthetic training data |
CN109493599A (en) * | 2018-11-16 | 2019-03-19 | 南京航空航天大学 | A kind of Short-time Traffic Flow Forecasting Methods based on production confrontation network |
CN110458370A (en) * | 2019-08-19 | 2019-11-15 | 南京航空航天大学 | A kind of urban roads safety prediction technique and system based on open data |
CN111079640A (en) * | 2019-12-09 | 2020-04-28 | 合肥工业大学 | Vehicle type identification method and system based on automatic amplification sample |
CN111231983A (en) * | 2020-01-14 | 2020-06-05 | 浙江吉利汽车研究院有限公司 | Vehicle control method, device and equipment based on traffic accident memory network |
Non-Patent Citations (4)
Title |
---|
杨昌东;余烨;徐珑刀;付源梓;路强;: "基于AT-PGGAN的增强数据车辆型号精细识别", 中国图象图形学报, no. 03, 16 March 2020 (2020-03-16), pages 179 - 190 * |
殷洪峰;周伟健;: "深度学习在江苏省高速公路交通事故预测中的应用", 中国交通信息化, no. 02, 15 February 2020 (2020-02-15), pages 131 - 135 * |
熊晓夏;陈龙;梁军;陈月霞;: "基于贝叶斯网络模型的道路交通事故链生成与演化研究", 公路交通科技, no. 05, 15 May 2018 (2018-05-15), pages 103 - 111 * |
郭清杨;: "基于生成对抗网络的对抗样本生成", 现代计算机, no. 07, 5 March 2020 (2020-03-05), pages 25 - 29 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112529114A (en) * | 2021-01-13 | 2021-03-19 | 北京云真信科技有限公司 | Target information identification method based on GAN, electronic device and medium |
CN113434459A (en) * | 2021-06-30 | 2021-09-24 | 电子科技大学 | Network-on-chip task mapping method based on generation of countermeasure network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110781838B (en) | Multi-mode track prediction method for pedestrians in complex scene | |
CN108922560B (en) | Urban noise identification method based on hybrid deep neural network model | |
CN110097755A (en) | Freeway traffic flow amount state identification method based on deep neural network | |
CN105096614B (en) | Newly-built crossing traffic flow Forecasting Methodology based on generation moldeed depth belief network | |
CN108765951B (en) | Express highway traffic state identification method based on bat algorithm support vector machine | |
CN109785618B (en) | Short-term traffic flow prediction method based on combinational logic | |
CN111260118A (en) | Vehicle networking traffic flow prediction method based on quantum particle swarm optimization strategy | |
CN108985515A (en) | A kind of new energy based on independent loops neural network goes out force prediction method and system | |
CN110555476B (en) | Intelligent vehicle lane change track prediction method suitable for man-machine hybrid driving environment | |
CN112732905B (en) | Traffic accident analysis, prevention and control method and system based on knowledge graph | |
CN111723997A (en) | Automatic generation method of urban major traffic accident data sample based on GAN | |
CN114202122A (en) | Urban traffic flow prediction method based on Markov cluster map attention network | |
CN113887643A (en) | New dialogue intention recognition method based on pseudo label self-training and source domain retraining | |
CN115662113A (en) | Signalized intersection people-vehicle game conflict risk assessment and early warning method | |
CN101740029A (en) | Three-particle cooperative optimization method applied to vector quantization-based speaker recognition | |
CN113673749A (en) | Vehicle and pedestrian map track prediction method and device | |
CN111907523A (en) | Vehicle following optimization control method based on fuzzy reasoning | |
CN116541755A (en) | Financial behavior pattern analysis and prediction method based on time sequence diagram representation learning | |
CN115270506A (en) | Method and system for predicting passing time of people going upstairs along stairs | |
CN112098869B (en) | Self-adaptive electric vehicle SOC estimation method based on big data | |
CN117592602A (en) | Subway passenger flow prediction method and system based on space-time diagram ordinary differential equation | |
CN114912719B (en) | Heterogeneous traffic individual trajectory collaborative prediction method based on graph neural network | |
CN115798629A (en) | Interpretable molecular property prediction method and system based on graph neural network | |
CN115796017A (en) | Interpretable traffic cognition method based on fuzzy theory | |
CN111028512B (en) | Real-time traffic prediction method and device based on sparse BP neural 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 |