CN108875812A - A kind of driving behavior classification method based on branch's convolutional neural networks - Google Patents
A kind of driving behavior classification method based on branch's convolutional neural networks Download PDFInfo
- Publication number
- CN108875812A CN108875812A CN201810557923.5A CN201810557923A CN108875812A CN 108875812 A CN108875812 A CN 108875812A CN 201810557923 A CN201810557923 A CN 201810557923A CN 108875812 A CN108875812 A CN 108875812A
- Authority
- CN
- China
- Prior art keywords
- branch
- neural networks
- convolutional neural
- driving behavior
- layer
- 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
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
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
-
- 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
- G06N3/08—Learning methods
Abstract
The present invention relates to a kind of driving behavior classification methods based on branch's convolutional neural networks, including the following contents：1) the trunk convolutional neural networks based on construction handle test sample, obtain the preliminary classification result of driving behavior；2) it is required according to the subdivision of driving behavior, branch's convolutional neural networks based on construction handle test sample again, realize the subdivision of driving behavior.It does not need manually to extract the characteristic parameter about driver's type the present invention is based on the classification method of the driving behavior of deep learning, and the present invention can reduce calculation amount according to the requirement of actual classification, and enough effectively characteristic parameters about driver's type can be extracted, it is accurate to complete driving behavior classification.
Description
Technical field
The present invention relates to a kind of driving behavior classification method, more particularly to a kind of based on branch's convolutional neural networks
Driving behavior classification method belongs to road safety studying technological domain in intelligent transportation system.
Background technique
The driving behavior extreme influence of driver the safety of road, most traffic accidents be all the mistake by driver
Caused by behavior.Different drivers has a different driving habits, therefore identifies that the type of driver is one and important grinds
Study carefully direction.
Traditional driving behavior classification method is mostly used clustering method and neural network method, clustering method generally require people
Work extracts the characteristic parameter of characterization driver's type, this causes the preparation work for needing a large amount of profession early period that can just obtain properly
Parameter set.And common neural network method calculation amount is very big or extracts less than characteristic parameter effective enough, thus shadow
Ring the effect of classification.
Summary of the invention
In view of the above-mentioned problems, the object of the present invention is to provide one kind can effectively extract driver's type feature parameter in turn
The accurate driving behavior classification method based on branch's convolutional neural networks for completing driving behavior classification.
To achieve the above object, the present invention takes following technical scheme：A kind of driving based on branch's convolutional neural networks
Member's behavior classification method, including the following contents：
1) the trunk convolutional neural networks based on construction handle test sample, obtain preliminary point of driving behavior
Class result；
2) according to the subdivision of driving behavior require, branch's convolutional neural networks based on construction to test sample again into
Row processing, realizes the subdivision of driving behavior.
Further, the detailed process of the step 1) is：
1.1) concrete type of driver, including A are primarily determined1Type, A2Type ..., AkType；
1.2) training sample is selected：(x1,y1),(x2,y2),…,(xn,yn), wherein xiIndicate the feature of driving behavior
Parameter, yiIndicate the type of driving behavior；
1.3) construction trunk convolutional neural networks：
1.4) test sample is handled using trunk convolutional neural networks, establishes objective function
In formula, f (xi,θ1) indicate trunk neural network model, θ1For the parameter of trunk neural network, using boarding steps
Degree descent method finds out the θ in objective function1, obtain the sorting machine of driving behavior preliminary classification.
Further, the detailed process of the step 1.3) is：
1.3.1) construction input layer：For determining the neuron number of input layer, the neuron number of input layer is according to driving
The characteristic parameter for the person of sailing is determined；
1.3.2) construction convolutional layer：Constructing convolutional layer is determined to the size and step-length of convolution kernel, according to input number
The size of convolution kernel is determined according to the size of scale and the type of data；
1.3.3 pond layer) is constructed：Constructing pond layer is the determination completed to pond size and step-length and pond type；
1.3.4 full articulamentum) is constructed；
1.3.5 single convolutional layer, single pond layer and single full articulamentum) is arranged to test training sample；
1.3.6) increase a convolutional layer and increase a pond layer, training sample is tested again, if increased
After adding convolutional layer and pond layer, the precision incrementss of model are not more than 0.01, it is considered that Current Situation of Neural Network has reached closely
It is seemingly optimal, otherwise, continues growing convolutional layer and pond layer tests training sample, until finding optimal neural network conduct
Trunk convolutional neural networks.
Further, the detailed process of the step 2) is：
2.1) branch's convolutional neural networks are constructed：
2.2) test sample is trained using branch's convolutional neural networks again, establishes objective functionWherein, g (x 'j,θ2) be branch's convolutional neural networks model, θ2For branch's convolutional neural networks
Parameter finds out the neural network parameter θ in objective function using stochastic gradient descent method2, obtain point of driving behavior subdivision
Class machine.
Further, the detailed process of the step 2.1) is：Add on trunk convolutional neural networks several layers therein
Upper several layers form branch's convolutional neural networks, then from determining A1,…,AkA is extracted in classqClass is subdivided into B1, B2..., Bm
Class determines that the output of u layers to v layers in trunk neural network is made with the Characteristics of Drivers ' Behavior newly chosen according to the type of subdivision
For the input of branch's neural network, new training sample is (x1′,y1′),(x′2,y′2),…,(x′m,y′m), wherein x 'jTable
Show the new cybernetics control number of driver, y 'jThe behavior type for indicating driver's subdivision is arranged branch's neural network single
Pond layer, single convolutional layer and single full articulamentum, then training sample is tested by branch's neural network, increase every time
One convolutional layer, a pond layer, until determining that optimal neural network is made when the precision incrementss of model are less than or equal to 0.01
For branch's convolutional neural networks.
The invention adopts the above technical scheme, which has the following advantages：1, the present invention is based on the driving of deep learning
The classification method of member's behavior does not need manually to extract the characteristic parameter about driver's type.2, the present invention can be according to practical point
The requirement of class reduces calculation amount, and can extract the effective characteristic parameter about driver's type enough, accurate to complete to drive
The person's of sailing behavior classification.3, the present invention can neatly add driving behavior under the premise of not changing entire neural network structure
Type.The present invention can be widely applied in intelligent transportation system.
Detailed description of the invention
Fig. 1 is the flow diagram of driving behavior classification method of the present invention.
Specific embodiment
Come to carry out detailed description to the present invention below in conjunction with the drawings and specific embodiments.It should be appreciated, however, that attached drawing
It has been provided only and has more fully understood the present invention, they should not be interpreted as limitation of the present invention.
Since the construction of convolutional neural networks depends on specific problem, so the method provided in the embodiment of the present invention exists
It is to modify as the case may be to the parameter of convolutional neural networks in practical application.It is provided by the invention to be based on branch
The driving behavior classification method of convolutional neural networks, including the following contents：
1, the trunk convolutional neural networks based on construction handle test sample, obtain preliminary point of driving behavior
Class result.
1) concrete type of driver, including A are primarily determined1Type, A2Type ..., AkType.
2) training sample is selected：(x1,y1),(x2,y2),…,(xn,yn), wherein xiIndicate the feature ginseng of driving behavior
Number, yiIndicate the type of driving behavior.
3) trunk convolutional neural networks are constructed：
3.1) construction input layer：For determining the neuron number of input layer, the neuron number of input layer is main here
It is to be determined according to the characteristic parameter of driver.
3.2) convolutional layer is constructed：Construction convolutional layer is mainly determined the size and step-length of convolution kernel, according to input
The type of the size of data scale and data determines the size of convolution kernel, and the size of the present embodiment convolution kernel is 2 to multiply 2, step-length
It is 1.
3.3) pond layer is constructed：Constructing pond layer is the determination completed to pond size and step-length and pond type, root
The size of pond size is determined according to upper one layer of scale, the pond of the present embodiment multiplies 2 having a size of 2, and step-length is according to pond
Size takes 2, and pond type takes maximum pond.
3.4) full articulamentum is constructed：The main number for considering neuron, the neuron of each neuron and preceding layer
It is connected, to form full articulamentum.In addition to full articulamentum, the activation primitive of other layers of trunk convolutional neural networks is all used
The activation primitive of ReLU function, full articulamentum uses Softmax function.
3.5) single convolutional layer, single pond layer and single full articulamentum is arranged to test training sample；
3.6) increase a convolutional layer and increase a pond layer, training sample is tested again, if increased
After convolutional layer and pond layer, the precision incrementss of model are not more than 0.01, then it is considered that Current Situation of Neural Network has reached
Otherwise near-optimization continues growing convolutional layer and pond layer tests training sample, until finding optimal neural network,
If the number of plies of this optimal neural network is k1(such as Fig. 1), this model is as trunk convolutional neural networks to the affiliated type of driver
Judged.
4) test sample is trained using trunk convolutional neural networks, establishes objective function
In formula, f (xi,θ1) indicate trunk neural network model, θ1For the parameter of trunk neural network, using boarding steps
Degree descent method finds out the θ in objective function1, obtain the sorting machine of driving behavior preliminary classification.
2, it is required according to the subdivision of driving behavior, constructs branch's convolutional neural networks, and branch's convolution based on construction
Neural network handles test sample again, realizes the subdivision of driving behavior.
2.1) construction process of branch's convolutional neural networks is：
Branch's convolutional neural networks are formed plus several layers on trunk convolutional neural networks several layers therein, then in the past
The A that face determines1,…,AkA is extracted in classqClass is subdivided into B1, B2..., BmClass determines trunk nerve net according to the type of subdivision
Input of u layers to v layers of the Characteristics of Drivers ' Behavior for exporting and newly choosing as branch's neural network on network, sets above here
New training sample be (x '1,y′1),(x′2,y′2),…,(x′m,y′m), wherein x 'jIndicate the new behavioural characteristic of driver
Parameter, y 'jIndicate the behavior type of driver's subdivision.Single pond layer is arranged for branch's neural network, single convolutional layer with
Single full articulamentum, then branch's neural network is tested, increase a pond layer, a full articulamentum, Zhi Daomo every time
When the precision incrementss of type are less than or equal to 0.01, optimal neural network is determined, if the number of plies of this optimal neural network is k2As
Branch's convolutional neural networks.
2.2) test sample is trained using branch's convolutional neural networks again, establishes objective functionWherein, g (x 'j,θ2) be branch's convolutional neural networks model, θ2For branch's convolutional neural networks
Parameter finds out the neural network parameter in objective function using stochastic gradient descent method, obtains the classification of driving behavior subdivision
Machine.
Below by specific embodiment to the driving behavior classification method of the invention based on branch's convolutional neural networks
Feasibility be described in detail, detailed process is：
1, the construction of trunk convolutional neural networks：
Select training sample：(x1,y1),(x2,y2),…,(xn,yn), input xiFor：The age of driver, gender, speed
And braking state, export yiFor：Conservative, plain edition, radical type, some in dangerous type.
The construction of input layer：Input size is 35x128；
The construction of convolutional layer 1：32 cores, each size 35x5, step-length 1；
The construction of pond layer 1：Pond size is 1X2, with maximum pond；
The construction of convolutional layer 2：64 cores, each core size is 1X3, step-length 1；
The construction of pond layer 2：Pond size is 1X2, with maximum pond；
The construction of full articulamentum：128 neurons, excitation function are Sigmoid；
The construction of output layer：Softmax is as output；
Driver can be divided into (conservative and plain edition) by using the construction of trunk convolutional neural networks and (swashed
Into type and dangerous type) two classes, single pond layer, single convolutional layer are preset, single full articulamentum is made a preliminary test, increased with this
Add pond layer and convolutional layer, until the precision incrementss of model are not more than 0.01, if the number of plies of this optimal neural network is k3, structure
It makes to obtain trunk convolutional neural networks, test sample is trained using trunk convolutional neural networks, establishes objective functionDriver's type feature parameter in objective function is found out using stochastic gradient descent method, is obtained preliminary
Driving behavior classification results.
2, the construction of branch's convolutional neural networks
2.1) construction of branch's convolutional neural networks 1：
Pond layer is added on convolutional layer 1, then adds full articulamentum again.Further driver's type is subdivided into (conservative
Type) and (plain edition), determine that the output of the convolutional layer 1 in trunk neural network and the Characteristics of Drivers ' Behavior newly chosen are refreshing as branch
Single pond layer, single convolutional layer and single full articulamentum is arranged for branch's neural network in input through network, then to branch
Neural network is tested, and a pond layer, a convolutional layer, when the precision incrementss of model are less than or equal to 0.01 are successively increased
When, determine optimal neural network, the number of plies of this optimal neural network is 8 layers, using this branch's convolutional neural networks to test specimens
This progress is trained again, establishes objective functionIt is found out in objective function using stochastic gradient descent method
Neural network parameter obtains the sorting machine of driving behavior subdivision.
2.2) construction of branch's convolutional neural networks 2：
Pond layer is added on convolutional layer 2, then adds full articulamentum again.It is (radical that further driver's type is subdivided into
Type) and (dangerous type), determine the output of convolutional layer 2 in trunk neural network and the Characteristics of Drivers ' Behavior newly chosen as branch's nerve
Single pond layer, single convolutional layer and single full articulamentum is arranged for branch's neural network in the input of network, then to branch's mind
It is tested through network, successively increases a pond layer, a convolutional layer, when the precision incrementss of model are less than or equal to 0.01
When, determine optimal neural network, the number of plies of this optimal neural network is 6 layers, using this branch's convolutional neural networks to test specimens
This progress is trained again, establishes objective functionIt is found out in objective function using stochastic gradient descent method
Neural network parameter, obtain driving behavior subdivision sorting machine.
The various embodiments described above are merely to illustrate the present invention, and wherein the step of method may be changed, it is all
The equivalents and improvement carried out on the basis of technical solution of the present invention, should not exclude except protection scope of the present invention.
Claims (5)
1. a kind of driving behavior classification method based on branch's convolutional neural networks, it is characterised in that including the following contents：
1) the trunk convolutional neural networks based on construction handle test sample, obtain the preliminary classification knot of driving behavior
Fruit；
2) it is required according to the subdivision of driving behavior, branch's convolutional neural networks based on construction locate test sample again
Reason, realizes the subdivision of driving behavior.
2. the driving behavior classification method according to claim 1 based on branch's convolutional neural networks, which is characterized in that
The detailed process of the step 1) is：
1.1) concrete type of driver, including A are primarily determined1Type, A2Type ..., AkType；
1.2) training sample is selected：(x1,y1),(x2,y2),…,(xn,yn), wherein xiIndicate the characteristic parameter of driving behavior,
yiIndicate the type of driving behavior；
1.3) trunk convolutional neural networks are constructed：
1.4) test sample is handled using trunk convolutional neural networks, establishes objective function
In formula, f (xi,θ1) indicate trunk neural network model, θ1For the parameter of trunk neural network, using under stochastic gradient
Drop method finds out the θ in objective function1, obtain the sorting machine of driving behavior preliminary classification.
3. the driving behavior classification method according to claim 2 based on branch's convolutional neural networks, which is characterized in that
The detailed process of the step 1.3) is：
1.3.1) construction input layer：For determining the neuron number of input layer, the neuron number of input layer is according to driver
Characteristic parameter be determined；
1.3.2) construction convolutional layer：Constructing convolutional layer is determined to the size and step-length of convolution kernel, is advised according to input data
The type of the size of mould and data determines the size of convolution kernel；
1.3.3 pond layer) is constructed：Constructing pond layer is the determination completed to pond size and step-length and pond type；
1.3.4 full articulamentum) is constructed；
1.3.5 single convolutional layer, single pond layer and single full articulamentum) is arranged to test training sample；
1.3.6) increase a convolutional layer and increase a pond layer, training sample is tested again, if increasing volume
After lamination and pond layer, the precision incrementss of model are not more than 0.01, it is considered that Current Situation of Neural Network has reached approximation most
It is excellent, otherwise, continues growing convolutional layer and pond layer tests training sample, until finding optimal neural network as trunk
Convolutional neural networks.
4. the driving behavior classification method according to claim 3 based on branch's convolutional neural networks, which is characterized in that
The detailed process of the step 2) is：
2.1) branch's convolutional neural networks are constructed：
2.2) test sample is trained using branch's convolutional neural networks again, establishes objective functionWherein, g (x 'j,θ2) be branch's convolutional neural networks model, θ2For branch's convolutional neural networks
Parameter finds out the neural network parameter θ in objective function using stochastic gradient descent method2, obtain point of driving behavior subdivision
Class machine.
5. the driving behavior classification method according to claim 4 based on branch's convolutional neural networks, which is characterized in that
The detailed process of the step 2.1) is：
Branch's convolutional neural networks are formed plus several layers on trunk convolutional neural networks several layers therein, then from determining
A1,…,AkA is extracted in classqClass is subdivided into B1, B2..., BmClass determines the u in trunk neural network according to the type of subdivision
For layer to v layer of output and the newly input of the Characteristics of Drivers ' Behavior chosen as branch's neural network, new training sample is (x '1,
y′1),(x′2,y′2),…,(x′m,y′m), wherein x 'jIndicate the new cybernetics control number of driver, y 'jIndicate that driver is thin
Single pond layer, single convolutional layer and single full articulamentum is arranged for branch's neural network in the behavior type divided, then by dividing
Branch neural network tests training sample, then increases a convolutional layer, a pond layer, until the precision of model every time
When incrementss are less than or equal to 0.01, determine optimal neural network as branch's convolutional neural networks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810557923.5A CN108875812B (en) | 2018-06-01 | 2018-06-01 | Driver behavior classification method based on branch convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810557923.5A CN108875812B (en) | 2018-06-01 | 2018-06-01 | Driver behavior classification method based on branch convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108875812A true CN108875812A (en) | 2018-11-23 |
| CN108875812B CN108875812B (en) | 2021-10-22 |
Family
ID=64335363
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810557923.5A Active CN108875812B (en) | 2018-06-01 | 2018-06-01 | Driver behavior classification method based on branch convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108875812B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109544550A (en) * | 2018-12-05 | 2019-03-29 | 易必祥 | Based on CT images intelligent measurement and recognition methods and system |
| CN109543627A (en) * | 2018-11-27 | 2019-03-29 | 西安电子科技大学 | A kind of method, apparatus and computer equipment judging driving behavior classification |
| CN110321970A (en) * | 2019-07-11 | 2019-10-11 | 山东领能电子科技有限公司 | A kind of fine-grained objective classification method of multiple features based on branch neural network |
| CN110322509A (en) * | 2019-06-26 | 2019-10-11 | 重庆邮电大学 | Object localization method, system and computer equipment based on level Class Activation figure |
| CN116092059A (en) * | 2022-11-30 | 2023-05-09 | 南京通力峰达软件科技有限公司 | Neural network-based vehicle networking user driving behavior recognition method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106056071A (en) * | 2016-05-30 | 2016-10-26 | 北京智芯原动科技有限公司 | Method and device for detection of driver' behavior of making call |
| US20170032222A1 (en) * | 2015-07-30 | 2017-02-02 | Xerox Corporation | Cross-trained convolutional neural networks using multimodal images |
| CN106682602A (en) * | 2016-12-16 | 2017-05-17 | 深圳市华尊科技股份有限公司 | Driver behavior identification method and terminal |
-
2018
- 2018-06-01 CN CN201810557923.5A patent/CN108875812B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170032222A1 (en) * | 2015-07-30 | 2017-02-02 | Xerox Corporation | Cross-trained convolutional neural networks using multimodal images |
| CN106056071A (en) * | 2016-05-30 | 2016-10-26 | 北京智芯原动科技有限公司 | Method and device for detection of driver' behavior of making call |
| CN106682602A (en) * | 2016-12-16 | 2017-05-17 | 深圳市华尊科技股份有限公司 | Driver behavior identification method and terminal |
Non-Patent Citations (3)
| Title |
|---|
| SHIYANG YAN；ET AL: "《Driver Behavior Recognition Based on Deep Convolutional Neural Networks》", 《2016 12TH INTERNATIONAL CONFERENCE ON NATURAL COMPUTATION, FUZZY SYSTEMS AND KNOWLEDGE DISCOVERY》 * |
| ZHUJIAN LIANG，ET AL: "《Unconstrained Facial Landmark Localization with Backbone-Branches Fully-Convolutional Networks》", 《ARXIV：1507.03409V1》 * |
| 刘鹏: "《融合面部表情和语音的驾驶员路怒症识别方法研究》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109543627A (en) * | 2018-11-27 | 2019-03-29 | 西安电子科技大学 | A kind of method, apparatus and computer equipment judging driving behavior classification |
| CN109543627B (en) * | 2018-11-27 | 2023-08-01 | 西安电子科技大学 | Method and device for judging driving behavior category and computer equipment |
| CN109544550A (en) * | 2018-12-05 | 2019-03-29 | 易必祥 | Based on CT images intelligent measurement and recognition methods and system |
| CN109544550B (en) * | 2018-12-05 | 2021-10-22 | 易必祥 | CT image-based intelligent detection and identification method and system |
| CN110322509A (en) * | 2019-06-26 | 2019-10-11 | 重庆邮电大学 | Object localization method, system and computer equipment based on level Class Activation figure |
| CN110321970A (en) * | 2019-07-11 | 2019-10-11 | 山东领能电子科技有限公司 | A kind of fine-grained objective classification method of multiple features based on branch neural network |
| CN116092059A (en) * | 2022-11-30 | 2023-05-09 | 南京通力峰达软件科技有限公司 | Neural network-based vehicle networking user driving behavior recognition method and system |
| CN116092059B (en) * | 2022-11-30 | 2023-10-20 | 南京通力峰达软件科技有限公司 | Neural network-based vehicle networking user driving behavior recognition method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108875812B (en) | 2021-10-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108875812A (en) | A kind of driving behavior classification method based on branch's convolutional neural networks | |
| CN106485251A (en) | Egg embryo classification based on deep learning | |
| CN108288035A (en) | The human motion recognition method of multichannel image Fusion Features based on deep learning | |
| CN106874956A (en) | The construction method of image classification convolutional neural networks structure | |
| CN109102014A (en) | The image classification method of class imbalance based on depth convolutional neural networks | |
| CN109543502A (en) | A kind of semantic segmentation method based on the multiple dimensioned neural network of depth | |
| CN106446942A (en) | Crop disease identification method based on incremental learning | |
| CN107247989A (en) | A kind of neural network training method and device | |
| CN108596274A (en) | Image classification method based on convolutional neural networks | |
| CN112308158A (en) | Multi-source field self-adaptive model and method based on partial feature alignment | |
| CN108615071A (en) | The method and device of model measurement | |
| CN106919951A (en) | A kind of Weakly supervised bilinearity deep learning method merged with vision based on click | |
| CN107292335A (en) | A kind of transmission line of electricity cloud data automatic classification method based on Random Forest model | |
| CN106875511A (en) | A kind of method for learning driving style based on own coding regularization network | |
| CN106570516A (en) | Obstacle recognition method using convolution neural network | |
| CN106651574A (en) | Personal credit assessment method and apparatus | |
| CN106529503A (en) | Method for recognizing face emotion by using integrated convolutional neural network | |
| CN107506350A (en) | A kind of method and apparatus of identification information | |
| CN108846512A (en) | Based on the water quality prediction method preferentially classified | |
| CN105088595B (en) | The online color matching apparatus of printing and dyeing based on neural network model and method | |
| CN108875788A (en) | A kind of SVM classifier parameter optimization method based on modified particle swarm optiziation | |
| CN106127240A (en) | A kind of classifying identification method of plant image collection based on nonlinear reconstruction model | |
| CN108985617A (en) | A kind of product manufacturing process dispatching method and system based on intelligence manufacture | |
| CN107945534A (en) | A kind of special bus method for predicting based on GMDH neutral nets | |
| CN109102498A (en) | A kind of method of cluster type nucleus segmentation in cervical smear image |
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 |