CN112257774A - Target detection method, device, equipment and storage medium based on federal learning - Google Patents
Target detection method, device, equipment and storage medium based on federal learning Download PDFInfo
- Publication number
- CN112257774A CN112257774A CN202011125718.5A CN202011125718A CN112257774A CN 112257774 A CN112257774 A CN 112257774A CN 202011125718 A CN202011125718 A CN 202011125718A CN 112257774 A CN112257774 A CN 112257774A
- Authority
- CN
- China
- Prior art keywords
- target detection
- detection model
- model
- lightweight
- data set
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 261
- 238000013138 pruning Methods 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 37
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 230000006870 function Effects 0.000 claims description 45
- 238000004590 computer program Methods 0.000 claims description 11
- 238000000034 method Methods 0.000 claims description 11
- 238000012935 Averaging Methods 0.000 claims description 4
- 230000004927 fusion Effects 0.000 claims description 4
- 238000000638 solvent extraction Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 abstract description 5
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 239000000779 smoke Substances 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- 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
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to an artificial intelligence technology, and discloses a target detection method based on federal learning, which comprises the following steps: pruning the initial target detection model by adopting random weight to obtain a lightweight target detection model; training the lightweight target detection model based on a local data set to obtain a plurality of corresponding model gradient parameters, and sending the model gradient parameters to a server; receiving a global gradient parameter obtained by fusing the model gradient parameter by the server according to a federal average algorithm; updating the initial target detection model by using the global gradient parameters, and returning to the training step until a preset termination condition is met to obtain a target detection model; and carrying out target detection on the image to be detected by using the target detection model. In addition, the invention also relates to a block chain technology, and the local data set can be stored in the block chain node. The invention can improve the detection speed and the detection accuracy of target detection.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a target detection method and device based on federal learning, electronic equipment and a computer readable storage medium.
Background
The object identification is an important field of computer vision development, and can be applied to various technical schemes such as image recognition, object positioning, target detection and the like according to different tasks. Among them, object detection is widely studied as a basis for applications such as face recognition, case segmentation, and the like.
The existing target detection method mainly adopts a multilayer convolutional neural network structure to generate a candidate region, and then regression and classification are carried out, but the identification speed of the method is very low, and the efficiency is low; meanwhile, in some service scenarios, some service data need to be kept secret, so that the amount of collected sample data is small, and the detection accuracy is low.
Disclosure of Invention
The invention provides a target detection method, a target detection device, electronic equipment and a computer-readable storage medium based on federal learning, and mainly aims to provide a target detection method which is rapid and high in accuracy.
In order to achieve the above object, the present invention provides a target detection method based on federal learning, which comprises:
carrying out pruning treatment on the pre-constructed initial target detection model by adopting random weight to obtain a lightweight target detection model;
training the lightweight target detection model based on a local data set to obtain a plurality of corresponding model gradient parameters, and sending the model gradient parameters to a server;
receiving a global gradient parameter obtained by fusing the model gradient parameter by the server according to a federal average algorithm;
updating the initial target detection model by using the global gradient parameters, and returning to the step of training the lightweight target detection model based on the local data set until the lightweight target detection model meets a preset termination condition to obtain a trained target detection model;
and carrying out target detection on the image to be detected by using the target detection model.
Optionally, the pruning processing on the pre-constructed initial target detection model by using the random weight includes:
acquiring the weight of each convolution layer in the initial target detection model;
determining the priority of each convolution layer according to the weight;
pruning is carried out according to the priority of each convolution layer to obtain a pruned initial target detection model;
calculating a target function of the pruned initial target detection model;
and returning to the step of determining the priority of each convolution layer according to the weight, and stopping pruning until the target function meets the preset stop condition.
Optionally, the pruning according to the priority of each convolutional layer includes:
comparing the priority of each convolution layer with a preset threshold;
and when the priority is smaller than the threshold value, deleting the convolution layer corresponding to the priority in the initial target detection model.
Optionally, the training the lightweight target detection model based on the local data set to obtain a plurality of corresponding model gradient parameters includes:
inputting the local data set into the lightweight target detection model to obtain an output result of the lightweight target detection model;
calculating a loss value of the output result by using a preset loss function;
and adjusting parameters of the lightweight target detection model according to the loss value until the loss value is kept unchanged, and determining the parameters of the lightweight target detection model as model gradient parameters.
Optionally, the inputting the local data set to the lightweight target detection model to obtain an output result of the lightweight target detection model includes:
uniformly partitioning the local data set to obtain a plurality of feature blocks;
extracting feature data in each feature block, and generating a plurality of prediction frames according to the feature data;
and performing target recognition on the plurality of prediction frames, and outputting a prediction target class and a prediction probability corresponding to the prediction target class.
Optionally, the calculating a loss value of the output result by using a preset loss function includes:
calculating a frame coordinate position error, a prediction accuracy error and a prediction category error according to the output result;
and calculating to obtain a loss value according to the frame coordinate position error, the prediction accuracy error and the prediction category error by using a preset loss function.
Optionally, the federated averaging algorithm includes:
where W is a global gradient parameter, WkIs the model gradient parameter and p is the total number of participants.
In order to solve the above problem, the present invention further provides a target detection apparatus based on federal learning, including:
the pruning module is used for carrying out pruning treatment on the pre-constructed initial target detection model by adopting random weight to obtain a lightweight target detection model;
the local training module is used for training the lightweight target detection model based on a local data set to obtain a plurality of corresponding model gradient parameters and sending the model gradient parameters to a server;
the parameter fusion module is used for receiving a global gradient parameter obtained by fusing the model gradient parameter by the server according to a federal average algorithm;
a parameter updating module, configured to update the initial target detection model by using the global gradient parameter, and return to the above step of training the lightweight target detection model based on the local data set until the lightweight target detection model meets a preset termination condition, so as to obtain a trained target detection model;
and the detection module is used for carrying out target detection on the image to be detected by using the target detection model.
In order to solve the above problem, the present invention also provides an electronic device, including:
a memory storing at least one computer program; and
a processor executing a computer program stored in the memory to implement any of the above federal learning based goal detection methods.
In order to solve the above problem, the present invention further provides a computer-readable storage medium including a storage data area and a storage program area, the storage data area storing created data, the storage program area storing a computer program; wherein the computer program, when executed by a processor, implements any of the above federated learning-based goal detection methods.
The embodiment of the invention adopts the random weight to carry out pruning treatment on the pre-constructed initial target detection model, optimizes the model with the multilayer network structure into a lightweight model, reduces the calculated amount and improves the detection speed of the model; meanwhile, the lightweight target detection model is trained based on the federal learning technology, so that the secrecy of a local data set is guaranteed, the training data set of the model is expanded, and the accuracy of the lightweight target detection model is improved. Therefore, the target detection method, the target detection device, the electronic equipment and the computer readable storage medium based on the federal learning, which are provided by the invention, can improve the detection speed and the detection accuracy of target detection.
Drawings
Fig. 1 is a schematic flow chart of a target detection method based on federal learning according to an embodiment of the present invention;
fig. 2 is a schematic flow chart of a pruning method for random weights according to an embodiment of the present invention;
FIG. 3 is a schematic flow chart of a model training method based on federated learning according to an embodiment of the present invention;
fig. 4 is a schematic block diagram of an object detection apparatus based on federal learning according to an embodiment of the present invention;
fig. 5 is a schematic internal structural diagram of an electronic device implementing a target detection method based on federal learning according to an embodiment of the present invention;
the implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The embodiment of the application provides a target detection method based on federal learning. The execution subject of the target detection method based on federal learning includes, but is not limited to, at least one of electronic devices, such as a server and a terminal, which can be configured to execute the method provided by the embodiments of the present application. In other words, the target detection method based on federal learning may be performed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like.
Referring to fig. 1, a flow diagram of a target detection method based on federated learning according to an embodiment of the present invention is provided, where the target detection method based on federated learning includes:
and S1, pruning the pre-constructed initial target detection model by adopting random weight to obtain a lightweight target detection model.
The initial target detection model in the embodiment of the invention is a detection model based on a convolutional neural network.
Optionally, before pruning the pre-constructed initial target detection model by using the random weight, the embodiment of the present invention further includes: and acquiring sample data, inputting the sample data into the initial target detection model, and determining output data of each convolution layer in the initial target detection model.
In detail, referring to fig. 2, the pruning process of the pre-constructed initial target detection model using the random weights includes:
s10, acquiring the weight of each convolution layer in the initial target detection model;
s11, determining the priority of each convolution layer according to the weight;
s12, pruning according to the priority of each convolution layer to obtain a pruned initial target detection model;
s13, calculating an objective function of the pruned initial object detection model;
s14, judging whether the target function meets a preset stop condition;
and S15, stopping pruning to obtain the lightweight target detection model.
When the target function does not meet the preset stop condition, returning to the step S11, and continuing to execute the step of determining the priority of the convolutional layer according to the weight; and if the target function meets the preset stop condition, executing the step S15, and stopping pruning to obtain the lightweight target detection model.
The weight is a different value given to each convolutional layer in advance, and in this embodiment, the value of the weight ranges from 0 to 1. The stop condition may be that the value of the objective function no longer decreases.
Optionally, the embodiment of the present invention determines the priority of each convolutional layer by using the following formula:
Wj＝gatej×outputj-1
where j is the number of layers of the convolutional layer, WjIs the priority, gate, of convolutional layer jjIs the weight, output, of convolutional layer jj-1Is the upper output of convolutional layer j, i.e., the input data of convolutional layer j. Further, said each according toPruning the convolutional layers according to the priority, comprising:
comparing the priority of each convolution layer with a preset threshold;
when the priority is larger than or equal to the threshold, reserving the convolutional layer corresponding to the priority;
and when the priority is smaller than the threshold value, deleting the convolution layer corresponding to the priority in the initial target detection model.
In the embodiment of the invention, a variable global threshold lambda is preset, and when the priority W of a convolutional layer j isjAnd when the lambda is less than or equal to lambda, deleting the convolution layer j in the initial target detection model. The size of the global threshold λ is related to the total number of layers in the initial target detection model, and when the total number of layers in the initial target detection model is decreased, λ is also decreased accordingly.
Further, the embodiment of the present invention calculates the objective function using the following formula:
wherein, f (x)i(ii) a W, α) is the output of the initial target detection model, xiIs sample data, W is the parameter of the initial target detection model, alpha is the structural representation of the initial target detection model, N is the total number of sample data, y isiRepresenting sample data xiCorresponding to the actual label, ω is a balance parameter, which is a constant, M is the number of convolution layers in the initial target detection model, gatejIs the weight of convolutional layer j.
After pruning, the calculation speed of the initial target detection model is improved, but the accuracy of the model is also reduced, in order to ensure the accuracy of the model, the pruning effect of the initial target detection model is detected by using an objective function, and the detection speed and the detection accuracy of the lightweight target detection model are ensured.
The initial target detection model comprises a plurality of convolution layers, pooling layers and full-connection layers, each layer comprises a large number of weight parameters, and a large number of calculations are needed in the actual use process, so that the identification speed is low.
S2, training the lightweight target detection model based on the local data set to obtain a plurality of corresponding model gradient parameters, and sending the model gradient parameters to the server.
The embodiment of the invention uses a federal learning algorithm to train the lightweight target detection model. The federated learning algorithm is jointly trained using multiple participants, where the participants may be clients, own their local data sets, and train the machine learning model using only their own data sets.
The local data set is a data set which is stored locally by each participant, has the same characteristics with the data sets of other participants, but has different data contents.
Optionally, before training the lightweight target detection model, the embodiment of the present invention further includes initializing a local data set, specifically including:
annotating a target object identification frame within the local data set;
and adding a category label of the target object in the labeled data set.
In detail, each participant firstly defines the category of the target object, there may be a plurality of categories, each participant marks the target object in the local data set with a recognition box, and then determines the category of the target object, and adds a category label to the target object, thereby facilitating the subsequent training.
In detail, referring to fig. 3, the training the lightweight target detection model based on the local data set to obtain a plurality of corresponding model gradient parameters includes:
s20, inputting the local data set into the lightweight target detection model to obtain an output result of the lightweight target detection model;
s21, calculating a loss value of the output result by using a preset loss function;
s22, judging whether the loss value is converged;
s23, adjusting the parameters of the lightweight target detection model according to the loss values, and returning to the step S20;
and S24, determining the parameters of the lightweight target detection model as model gradient parameters.
When the judgment result of the step S22 is that the loss value is not converged, executing step S23, adjusting the parameters of the lightweight target detection model according to the loss value, and returning to the step S20; if the loss value is converged, step S24 is executed to determine the parameters of the lightweight target detection model as model gradient parameters.
In the embodiment of the present invention, the loss value convergence means that the loss value is maintained unchanged. And the adjusting of the parameters of the lightweight target detection model according to the loss values is to perform back propagation on the lightweight target detection model according to the loss values, calculate the gradient of each convolution layer, calculate the update amount according to the gradient and use the original parameters minus the update amount as new parameters.
Further, the inputting the local data set to the lightweight target detection model to obtain an output result of the lightweight target detection model includes:
uniformly partitioning the local data set to obtain a plurality of feature blocks;
extracting feature data in each feature block, and generating a plurality of prediction frames according to the feature data;
and performing target recognition on the plurality of prediction frames, and outputting a prediction target class and a prediction probability corresponding to the prediction target class.
The local data set entering the lightweight target detection model is uniformly partitioned, the size of the partitioned blocks determines the fine granularity of picture detection, so that whether pre-labeled target objects exist in each frame or not is predicted by using a plurality of prediction bounding boxes in each feature block, and the category and the corresponding probability of the target objects are respectively predicted.
In detail, the calculating the loss value of the output result by using the preset loss function according to the embodiment of the present invention includes:
calculating a frame coordinate position error, a prediction accuracy error and a prediction category error according to the output result;
and calculating to obtain a loss value according to the frame coordinate position error, the prediction accuracy error and the prediction category error by using a preset loss function.
The preset loss function in the embodiment of the invention can be obtained by combining the frame coordinate position error, the prediction accuracy error and the prediction category error, and errors in multiple aspects of the output result can be synthesized by calculating the loss function to obtain a loss value.
Further, embodiments of the present invention calculate the loss function using the following formula:
in the above loss function, the part before the first plus sign represents the frame coordinate position prediction error including the target object, the part after the second plus sign represents the prediction accuracy error including the frame of the target object, the part after the third plus sign represents the prediction category error, further, M is the total number of feature blocks, H is the total number of predicted frames, x is the total number of predicted framesmiIs a vector representation of the prediction bounding box i in the mth feature block,
the position of the bounding box i is predicted for the mth feature block,
marking the actual position of the frame in the mth characteristic block;
indicating the prediction accuracy of the predicted bounding box i in the mth feature block,
representing the accuracy of the actual labeling of the bounding box in the mth feature block; x is the number ofmIs a vector representation of the feature block m, N is the total number of object classes, P (N)j)predRepresenting the prediction probability, P (N), of the target class jj)trueRepresenting the actual probability of the target class j.
And S3, receiving a global gradient parameter obtained by fusing the plurality of model gradient parameters by the server according to a federal average algorithm.
In detail, in the embodiment of the present invention, each participant sends the gradient model parameters obtained through local training to the server, and the server fuses the multiple model gradient parameters by using a federate average algorithm to obtain global gradient parameters, and encrypts and transmits the global gradient parameters to each participant, so that the security and privacy of data of each participant can be ensured.
In detail, the federal averaging algorithm in the embodiment of the present invention includes:
where W is a global gradient parameter, WkIs the model gradient parameter and p is the total number of participants.
And S4, updating the initial target detection model by using the global gradient parameters.
In the embodiment of the present invention, the global gradient parameter is transmitted to each of the participating parties, and each of the participating parties updates the model parameter of the initial target detection model by using the global gradient parameter, so that the original model parameter in the initial target detection model is replaced by the global gradient parameter.
And S5, judging whether the updated initial target detection model meets the preset training termination condition.
In the embodiment of the present invention, the preset termination condition is that the iteration number of the initial target detection model reaches a preset number. Obtaining iteration times after updating; when the iteration times do not reach the preset times, determining that the updated initial target detection model does not meet the preset termination condition; and when the iteration times reach preset times, determining that the updated initial target detection model meets preset termination conditions.
When the updated initial target detection model does not meet the preset termination condition, returning to the step S2;
and when the updated initial target detection model meets a preset termination condition, obtaining a trained target detection model.
According to the embodiment of the invention, the sample space is enlarged by performing joint training on a plurality of participants, so that the accuracy of the model is improved.
And S6, carrying out target detection on the image to be detected by using the target detection model.
In detail, the embodiment of the present invention may use the target detection model to detect an image to be detected according to a preset target object, and feed back a detection result to a user.
The target detection model in the embodiment of the invention is an end-to-end lightweight target detection model, has high detection speed, optimizes the accuracy of the model by adopting a joint training scheme, can realize quick and accurate target object detection, can be applied to dangerous conditions and timely response, and can quickly and accurately detect dangerous conditions such as flame, smoke and the like particularly in a factory with high security and protection requirements, thereby avoiding the occurrence of dangerous conditions.
The embodiment of the invention adopts the random weight to carry out pruning treatment on the pre-constructed initial target detection model, optimizes the model with the multilayer network structure into a lightweight model, reduces the calculated amount and improves the detection speed of the model; meanwhile, the lightweight target detection model is trained based on the federal learning technology, so that the secrecy of a local data set is guaranteed, the training data set of the model is expanded, and the accuracy of the lightweight target detection model is improved. Therefore, the target detection method, the target detection device and the computer readable storage medium based on the federal learning can improve the detection speed and the detection accuracy of target detection.
Fig. 4 is a schematic block diagram of the target detection apparatus based on federal learning according to the present invention.
The target detection device 100 based on federal learning of the present invention can be installed in an electronic device. According to the realized functions, the target detection device based on the federal learning can comprise a pruning module 101, a local training module 102, a parameter fusion module 103, a parameter updating module 104 and a detection module 105. The module of the present invention, which may also be referred to as a unit, refers to a series of computer program segments that can be executed by a processor of an electronic device and that can perform a fixed function, and that are stored in a memory of the electronic device.
In the present embodiment, the functions regarding the respective modules/units are as follows:
the pruning module 101 is configured to perform pruning processing on the pre-constructed initial target detection model by using random weights, so as to obtain a lightweight target detection model.
The initial target detection model in the embodiment of the invention is a detection model based on a convolutional neural network.
Optionally, before pruning the pre-constructed initial target detection model by using the random weight, the embodiment of the present invention further includes: and acquiring sample data, inputting the sample data into the initial target detection model, and determining output data of each convolution layer in the initial target detection model.
In detail, the pruning module 101 is specifically configured to:
acquiring the weight of each convolution layer in the initial target detection model;
determining the priority of each convolution layer according to the weight;
pruning is carried out according to the priority of each convolution layer to obtain a pruned initial target detection model;
calculating a target function of the pruned initial target detection model;
judging whether the target function meets a preset stop condition or not;
and stopping pruning to obtain the lightweight target detection model.
When the target function does not meet the preset stop condition, returning to the operation of determining the priority of each convolution layer according to the weight; or
And if the target function meets the preset stop condition, executing stop pruning to obtain the lightweight target detection model.
The weight is a different value given to each convolutional layer in advance, and in this embodiment, the value of the weight ranges from 0 to 1. The stop condition may be that the value of the objective function no longer decreases.
Optionally, the embodiment of the present invention determines the priority of each convolutional layer by using the following formula:
Wj＝gatej×outputj-1
where j is the number of layers of the convolutional layer, WjIs the priority, gate, of convolutional layer jjIs the weight, output, of convolutional layer jj-1Is the upper output of convolutional layer j, i.e., the input data of convolutional layer j.
Further, the pruning according to the priority of each convolutional layer includes:
comparing the priority of each convolution layer with a preset threshold;
when the priority is larger than or equal to the threshold, reserving the convolutional layer corresponding to the priority;
and when the priority is smaller than the threshold value, deleting the convolution layer corresponding to the priority in the initial target detection model.
In the embodiment of the invention, a variable global threshold lambda is preset, and when the priority W of a convolutional layer j isjAnd when the lambda is less than or equal to lambda, deleting the convolution layer j in the initial target detection model. Wherein the size of the global threshold λ is related to the total number of layers in the initial target detection model when the initial target is detectedWhen the total number of layers in the detection model is reduced, the lambda is reduced.
Further, the embodiment of the present invention calculates the objective function using the following formula:
wherein, f (x)i(ii) a W, α) is the output of the initial target detection model, xiIs sample data, W is the parameter of the initial target detection model, alpha is the structural representation of the initial target detection model, N is the total number of sample data, y isiRepresenting sample data xiCorresponding to the actual label, ω is a balance parameter, which is a constant, M is the number of convolution layers in the initial target detection model, gatejIs the weight of convolutional layer j.
After pruning, the calculation speed of the initial target detection model is improved, but the accuracy of the model is also reduced, in order to ensure the accuracy of the model, the pruning effect of the initial target detection model is detected by using an objective function, and the detection speed and the detection accuracy of the lightweight target detection model are ensured.
The initial target detection model comprises a plurality of convolution layers, pooling layers and full-connection layers, each layer comprises a large number of weight parameters, and a large number of calculations are needed in the actual use process, so that the identification speed is low.
The local training module 102 is configured to train the lightweight target detection model based on a local data set to obtain a plurality of corresponding model gradient parameters, and send the model gradient parameters to a server.
The embodiment of the invention uses a federal learning algorithm to train the lightweight target detection model. The federated learning algorithm is jointly trained using multiple participants, where the participants may be clients, own their local data sets, and train the machine learning model using only their own data sets.
The local data set is a data set which is stored locally by each participant, has the same characteristics with the data sets of other participants, but has different data contents.
Optionally, before training the lightweight target detection model, the embodiment of the present invention further includes initializing a local data set, specifically including:
annotating a target object identification frame within the local data set;
and adding a category label of the target object in the labeled data set.
In detail, each participant firstly defines the category of the target object, there may be a plurality of categories, each participant marks the target object in the local data set with a recognition box, and then determines the category of the target object, and adds a category label to the target object, thereby facilitating the subsequent training.
In detail, the training the lightweight target detection model based on the local data set to obtain a plurality of corresponding model gradient parameters includes:
step A: inputting the local data set into the lightweight target detection model to obtain an output result of the lightweight target detection model;
and B: calculating a loss value of the output result by using a preset loss function;
and C: judging whether the loss value is converged;
step D: adjusting parameters of the lightweight target detection model according to the loss value, and returning to the step A;
step E: and determining parameters of the lightweight target detection model as model gradient parameters. When the judgment result in the step C is that the loss value is not converged, executing a step D: and adjusting parameters of the lightweight target detection model according to the loss value, and returning to the step A: (ii) a
If the loss value is converged, executing step E: and determining parameters of the lightweight target detection model as model gradient parameters.
In the embodiment of the present invention, the loss value convergence means that the loss value is maintained unchanged. And the adjusting of the parameters of the lightweight target detection model according to the loss values is to perform back propagation on the lightweight target detection model according to the loss values, calculate the gradient of each convolution layer, calculate the update amount according to the gradient and use the original parameters minus the update amount as new parameters.
Further, the inputting the local data set to the lightweight target detection model to obtain an output result of the lightweight target detection model includes:
uniformly partitioning the local data set to obtain a plurality of feature blocks;
extracting feature data in each feature block, and generating a plurality of prediction frames according to the feature data;
and performing target recognition on the plurality of prediction frames, and outputting a prediction target class and a prediction probability corresponding to the prediction target class.
The local data set entering the lightweight target detection model is uniformly partitioned, the size of the partitioned blocks determines the fine granularity of picture detection, so that whether pre-labeled target objects exist in each frame or not is predicted by using a plurality of prediction bounding boxes in each feature block, and the category and the corresponding probability of the target objects are respectively predicted. In detail, the calculating the loss value of the output result by using the preset loss function according to the embodiment of the present invention includes:
calculating a frame coordinate position error, a prediction accuracy error and a prediction category error according to the output result;
and calculating to obtain a loss value according to the frame coordinate position error, the prediction accuracy error and the prediction category error by using a preset loss function.
The preset loss function in the embodiment of the invention can be obtained by combining the frame coordinate position error, the prediction accuracy error and the prediction category error, and errors in multiple aspects of the output result can be synthesized by calculating the loss function to obtain a loss value.
Further, embodiments of the present invention calculate the loss function using the following formula:
in the above loss function, the part before the first plus sign represents the frame coordinate position prediction error including the target object, the part after the second plus sign represents the prediction accuracy error including the frame of the target object, the part after the third plus sign represents the prediction category error, further, M is the total number of feature blocks, H is the total number of predicted frames, x is the total number of predicted framesmiIs a vector representation of the prediction bounding box i in the mth feature block,
the position of the bounding box i is predicted for the mth feature block,
marking the actual position of the frame in the mth characteristic block;
indicating the prediction accuracy of the predicted bounding box i in the mth feature block,
representing the accuracy of the actual labeling of the bounding box in the mth feature block; x is the number ofmIs a vector representation of the feature block m, N is the total number of object classes, P (N)j)predRepresenting the prediction probability, P (N), of the target class jj)trueRepresenting the actual probability of the target class j.
The parameter fusion module 103 is configured to receive a global gradient parameter obtained by fusing the multiple model gradient parameters by the server according to a federal average algorithm.
In detail, in the embodiment of the present invention, each participant sends the gradient model parameters obtained through local training to the server, and the server fuses the multiple model gradient parameters by using a federate average algorithm to obtain global gradient parameters, and encrypts and transmits the global gradient parameters to each participant, so that the security and privacy of data of each participant can be ensured.
In detail, the federal averaging algorithm in the embodiment of the present invention includes:
where W is a global gradient parameter, WkIs the model gradient parameter and p is the total number of participants.
The parameter updating module 104 is configured to update the initial target detection model using the global gradient parameter;
and judging whether the updated initial target detection model meets a preset training termination condition or not.
In the embodiment of the present invention, the global gradient parameter is transmitted to each of the participating parties, and each of the participating parties updates the model parameter of the initial target detection model by using the global gradient parameter, so that the original model parameter in the initial target detection model is replaced by the global gradient parameter.
In the embodiment of the present invention, the preset termination condition is that the iteration number of the initial target detection model reaches a preset number. Obtaining iteration times after updating; when the iteration times do not reach the preset times, determining that the updated initial target detection model does not meet the preset termination condition; and when the iteration times reach preset times, determining that the updated initial target detection model meets preset termination conditions.
When the updated initial target detection model does not meet the preset termination condition, returning to the local training module 102;
and when the updated initial target detection model meets a preset termination condition, obtaining a trained target detection model.
And the preset termination condition is that the initial target detection model converges or the iteration frequency reaches a preset frequency.
According to the embodiment of the invention, the sample space is enlarged by performing joint training on a plurality of participants, so that the accuracy of the model is improved.
The detection module 105 is configured to perform target detection on an image to be detected by using the target detection model.
In detail, the embodiment of the present invention may use the target detection model to detect an image to be detected according to a preset target object, and feed back a detection result to a user.
The target detection model in the embodiment of the invention is an end-to-end lightweight target detection model, has high detection speed, optimizes the accuracy of the model by adopting a joint training scheme, can realize quick and accurate target object detection, can be applied to dangerous conditions and timely response, and can quickly and accurately detect dangerous conditions such as flame, smoke and the like particularly in a factory with high security and protection requirements, thereby avoiding the occurrence of dangerous conditions.
Fig. 5 is a schematic structural diagram of an electronic device for implementing the target detection method based on federal learning according to the present invention.
The electronic device 1 may comprise a processor 10, a memory 11 and a bus, and may further comprise a computer program, such as a federal learning based object detection program 12, stored in the memory 11 and operable on the processor 10.
The memory 11 includes at least one type of readable storage medium, which includes flash memory, removable hard disk, multimedia card, card-type memory (e.g., SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, etc. The memory 11 may in some embodiments be an internal storage unit of the electronic device 1, such as a removable hard disk of the electronic device 1. The memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in mobile hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, which are provided on the electronic device 1. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device 1. The memory 11 may be used not only to store application software installed in the electronic device 1 and various types of data, such as codes of the federal learning-based object detection program 12, but also to temporarily store data that has been output or will be output.
The processor 10 may be composed of an integrated circuit in some embodiments, for example, a single packaged integrated circuit, or may be composed of a plurality of integrated circuits packaged with the same or different functions, including one or more Central Processing Units (CPUs), microprocessors, digital Processing chips, graphics processors, and combinations of various control chips. The processor 10 is a Control Unit (Control Unit) of the electronic device, connects various components of the whole electronic device by using various interfaces and lines, and executes various functions and processes data of the electronic device 1 by running or executing programs or modules (for example, executing an object detection program based on federal learning, etc.) stored in the memory 11 and calling data stored in the memory 11.
The bus may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is arranged to enable connection communication between the memory 11 and at least one processor 10 or the like.
Fig. 5 only shows an electronic device with components, and it will be understood by a person skilled in the art that the structure shown in fig. 5 does not constitute a limitation of the electronic device 1, and may comprise fewer or more components than shown, or a combination of certain components, or a different arrangement of components.
For example, although not shown, the electronic device 1 may further include a power supply (such as a battery) for supplying power to each component, and preferably, the power supply may be logically connected to the at least one processor 10 through a power management device, so as to implement functions of charge management, discharge management, power consumption management, and the like through the power management device. The power supply may also include any component of one or more dc or ac power sources, recharging devices, power failure detection circuitry, power converters or inverters, power status indicators, and the like. The electronic device 1 may further include various sensors, a bluetooth module, a Wi-Fi module, and the like, which are not described herein again.
Further, the electronic device 1 may further include a network interface, and optionally, the network interface may include a wired interface and/or a wireless interface (such as a WI-FI interface, a bluetooth interface, etc.), which are generally used for establishing a communication connection between the electronic device 1 and other electronic devices.
Optionally, the electronic device 1 may further comprise a user interface, which may be a Display (Display), an input unit (such as a Keyboard), and optionally a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, or the like. The display, which may also be referred to as a display screen or display unit, is suitable for displaying information processed in the electronic device 1 and for displaying a visualized user interface, among other things.
It is to be understood that the described embodiments are for purposes of illustration only and that the scope of the appended claims is not limited to such structures.
The federal learning based target test program 12 stored in the memory 11 of the electronic device 1 is a combination of a plurality of computer programs, which when executed in the processor 10, can realize:
carrying out pruning treatment on the pre-constructed initial target detection model by adopting random weight to obtain a lightweight target detection model;
training the lightweight target detection model based on a local data set to obtain a plurality of corresponding model gradient parameters, and sending the model gradient parameters to a server;
receiving a global gradient parameter obtained by fusing the model gradient parameter by the server according to a federal average algorithm;
updating the initial target detection model by using the global gradient parameters, and returning to the step of training the lightweight target detection model based on the local data set until the lightweight target detection model meets a preset termination condition to obtain a trained target detection model;
and carrying out target detection on the image to be detected by using the target detection model.
Further, the integrated modules/units of the electronic device 1, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. The computer-readable medium may include: any entity or device capable of carrying said computer program code, recording medium, U-disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM).
Further, the computer usable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created according to the use of the blockchain node, and the like.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is only one logical functional division, and other divisions may be realized in practice.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional module.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof.
The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any accompanying claims should not be construed as limiting the claim concerned.
The block chain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.
Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. A plurality of units or means recited in the system claims may also be implemented by one unit or means in software or hardware. The terms second, etc. are used to denote names, but not any particular order.
Finally, it should be noted that the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims (10)
1. A target detection method based on federated learning is characterized in that the method is applied to the participants of federated learning and comprises the following steps:
carrying out pruning treatment on the pre-constructed initial target detection model by adopting random weight to obtain a lightweight target detection model;
training the lightweight target detection model based on a local data set to obtain a plurality of corresponding model gradient parameters, and sending the model gradient parameters to a server;
receiving a global gradient parameter obtained by fusing the model gradient parameter by the server according to a federal average algorithm;
updating the initial target detection model by using the global gradient parameters, and returning to the step of training the lightweight target detection model based on the local data set until the lightweight target detection model meets a preset termination condition to obtain a trained target detection model;
and carrying out target detection on the image to be detected by using the target detection model.
2. The federal learning-based target detection method as claimed in claim 1, wherein the pruning of the pre-constructed initial target detection model using the random weights comprises:
acquiring the weight of each convolution layer in the initial target detection model;
determining the priority of each convolution layer according to the weight;
pruning is carried out according to the priority of each convolution layer to obtain a pruned initial target detection model;
calculating a target function of the pruned initial target detection model;
and returning to the step of determining the priority of each convolution layer according to the weight, and stopping pruning until the target function meets the preset stop condition.
3. The federal learning-based target detection method of claim 2, wherein the pruning in accordance with the priorities of the convolutional layers comprises:
comparing the priority of each convolution layer with a preset threshold;
and when the priority is smaller than the threshold value, deleting the convolution layer corresponding to the priority in the initial target detection model.
4. The federal learning-based target detection method of claim 1, wherein the training of the lightweight target detection model based on a local data set to obtain a corresponding plurality of model gradient parameters comprises:
inputting the local data set into the lightweight target detection model to obtain an output result of the lightweight target detection model;
calculating a loss value of the output result by using a preset loss function;
and adjusting parameters of the lightweight target detection model according to the loss value until the loss value is kept unchanged, and determining the parameters of the lightweight target detection model as model gradient parameters.
5. The federal learning based target detection method of claim 4, wherein the inputting the local data set into the lightweight target detection model to obtain the output result of the lightweight target detection model comprises:
uniformly partitioning the local data set to obtain a plurality of feature blocks;
extracting feature data in each feature block, and generating a plurality of prediction frames according to the feature data;
and performing target recognition on the plurality of prediction frames, and outputting a prediction target class and a prediction probability corresponding to the prediction target class.
6. The federal learning-based target detection method as claimed in claim 5, wherein the calculating of the loss value of the output result using a preset loss function comprises:
calculating a frame coordinate position error, a prediction accuracy error and a prediction category error according to the output result;
and calculating to obtain a loss value according to the frame coordinate position error, the prediction accuracy error and the prediction category error by using a preset loss function.
7. The federal learning-based target detection method as claimed in any one of claims 1 to 6, wherein the federal averaging algorithm includes:
where W is a global gradient parameter, WkIs the model gradient parameter and p is the total number of participants.
8. An object detection apparatus based on federal learning, the apparatus comprising:
the pruning module is used for carrying out pruning treatment on the pre-constructed initial target detection model by adopting random weight to obtain a lightweight target detection model;
the local training module is used for training the lightweight target detection model based on a local data set to obtain a plurality of corresponding model gradient parameters and sending the model gradient parameters to a server;
the parameter fusion module is used for receiving a global gradient parameter obtained by fusing the model gradient parameter by the server according to a federal average algorithm;
a parameter updating module, configured to update the initial target detection model by using the global gradient parameter, and return to the above step of training the lightweight target detection model based on the local data set until the lightweight target detection model meets a preset termination condition, so as to obtain a trained target detection model;
and the detection module is used for carrying out target detection on the image to be detected by using the target detection model.
9. An electronic device, characterized in that the electronic device comprises:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform a federal learning based target detection method as claimed in any one of claims 1 to 7.
10. A computer-readable storage medium comprising a storage data area storing created data and a storage program area storing a computer program; wherein the computer program, when executed by a processor, implements the federal learning based objective detection method as claimed in any one of claims 1 to 7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011125718.5A CN112257774B (en) | 2020-10-20 | 2020-10-20 | Target detection method, device, equipment and storage medium based on federal learning |
| PCT/CN2020/131986 WO2021189906A1 (en) | 2020-10-20 | 2020-11-27 | Target detection method and apparatus based on federated learning, and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011125718.5A CN112257774B (en) | 2020-10-20 | 2020-10-20 | Target detection method, device, equipment and storage medium based on federal learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112257774A true CN112257774A (en) | 2021-01-22 |
| CN112257774B CN112257774B (en) | 2024-03-15 |
Family
ID=74245443
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011125718.5A Active CN112257774B (en) | 2020-10-20 | 2020-10-20 | Target detection method, device, equipment and storage medium based on federal learning |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112257774B (en) |
| WO (1) | WO2021189906A1 (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113033652A (en) * | 2021-03-23 | 2021-06-25 | 电子科技大学 | Image recognition system and method based on block chain and federal learning |
| CN113139662A (en) * | 2021-04-23 | 2021-07-20 | 深圳市大数据研究院 | Global and local gradient processing method, device, equipment and medium for federal learning |
| CN113179244A (en) * | 2021-03-10 | 2021-07-27 | 上海大学 | Federal deep network behavior feature modeling method for industrial internet boundary safety |
| CN113344221A (en) * | 2021-05-10 | 2021-09-03 | 上海大学 | Federal learning method and system based on neural network architecture search |
| CN113364543A (en) * | 2021-06-03 | 2021-09-07 | 厦门大学 | Edge calculation model training method based on federal reinforcement learning |
| CN113379708A (en) * | 2021-02-26 | 2021-09-10 | 山东大学 | Air conditioner external unit appearance detection method and system based on federal learning |
| CN114169412A (en) * | 2021-11-23 | 2022-03-11 | 北京邮电大学 | Federal learning model training method for large-scale industrial chain privacy calculation |
| CN114241305A (en) * | 2021-12-07 | 2022-03-25 | 中原动力智能机器人有限公司 | Target detection model updating method and device |
| CN114339252A (en) * | 2021-12-31 | 2022-04-12 | 深圳大学 | Data compression method and device |
| CN114782738A (en) * | 2022-03-25 | 2022-07-22 | 苏州浪潮智能科技有限公司 | Federal learning safety detection method and device, electronic equipment and readable storage medium |
| CN114782758A (en) * | 2022-06-21 | 2022-07-22 | 平安科技（深圳）有限公司 | Image processing model training method, system, computer device and storage medium |
| CN116523081A (en) * | 2023-04-07 | 2023-08-01 | 花瓣云科技有限公司 | Data standardization method and device |
Families Citing this family (36)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112836819B (en) * | 2021-01-26 | 2023-07-25 | 北京奇艺世纪科技有限公司 | Neural network model generation method and device |
| CN113950046B (en) * | 2021-10-19 | 2022-05-03 | 北京工商大学 | Credible encryption positioning method for heterogeneous topological network based on federal learning |
| CN114187006A (en) * | 2021-11-03 | 2022-03-15 | 杭州未名信科科技有限公司 | Block chain supervision-based federal learning method |
| CN114049554A (en) * | 2021-11-09 | 2022-02-15 | 江苏科技大学 | Lawn obstacle detection method based on lightweight YOLOv5s model |
| CN114077901B (en) * | 2021-11-23 | 2024-05-24 | 山东大学 | User position prediction method based on clustering graph federation learning |
| CN114117926B (en) * | 2021-12-01 | 2024-05-14 | 南京富尔登科技发展有限公司 | Robot cooperative control algorithm based on federal learning |
| CN114373144B (en) * | 2021-12-08 | 2024-09-06 | 上海海洋大学 | Automatic identification method for circular identification points in high-speed video |
| CN114219147B (en) * | 2021-12-13 | 2024-06-07 | 南京富尔登科技发展有限公司 | Power distribution station fault prediction method based on federal learning |
| CN114332832A (en) * | 2021-12-31 | 2022-04-12 | 杭州电子科技大学 | Driver distraction behavior identification method based on federal learning |
| CN114553869B (en) * | 2022-02-11 | 2024-09-13 | 新智我来网络科技有限公司 | Method and device for determining resource contribution degree based on joint learning and electronic equipment |
| CN114547658B (en) * | 2022-03-02 | 2024-06-04 | 深圳前海微众银行股份有限公司 | Data processing method, device, equipment and computer readable storage medium |
| CN114707634B (en) * | 2022-03-09 | 2024-10-01 | 大连理工大学 | Federal learning convolutional neural network model aggregation method, device and storage medium |
| CN114595888A (en) * | 2022-03-10 | 2022-06-07 | 广西中烟工业有限责任公司 | Method and device for predicting process control parameters of tobacco shred production line |
| CN114841364B (en) * | 2022-04-14 | 2024-06-14 | 北京理工大学 | Federal learning method for meeting personalized local differential privacy requirements |
| CN114764601B (en) * | 2022-05-05 | 2024-01-30 | 北京瑞莱智慧科技有限公司 | Gradient data fusion method, device and storage medium |
| CN114998251B (en) * | 2022-05-30 | 2024-09-03 | 天津理工大学 | Air multi-vision platform ground anomaly detection method based on federal learning |
| CN115017351B (en) * | 2022-05-31 | 2024-05-10 | 杭州卷积云科技有限公司 | Lightweight industrial picture classification method and system based on federal small sample learning |
| CN115081014B (en) * | 2022-05-31 | 2024-07-26 | 西安翔迅科技有限责任公司 | Automatic labeling method for target detection labels based on federal learning |
| CN115049058B (en) * | 2022-08-17 | 2023-01-20 | 北京智芯微电子科技有限公司 | Compression method and device of topology recognition model, electronic equipment and medium |
| CN115694877B (en) * | 2022-08-30 | 2023-08-15 | 电子科技大学长三角研究院(衢州) | Space crowdsourcing task allocation method based on federal preference learning |
| CN115186937B (en) * | 2022-09-09 | 2022-11-22 | 闪捷信息科技有限公司 | Prediction model training and data prediction method and device based on multi-party data cooperation |
| CN115410190B (en) * | 2022-11-02 | 2023-04-14 | 平安银行股份有限公司 | License plate image corner detection model training method, license plate desensitization method and equipment thereof |
| CN115577858B (en) * | 2022-11-21 | 2023-04-07 | 山东能源数智云科技有限公司 | Block chain-based carbon emission prediction method and device and electronic equipment |
| CN115834516A (en) * | 2022-11-28 | 2023-03-21 | 超聚变数字技术有限公司 | Data processing method, equipment and server |
| CN116032431B (en) * | 2023-02-06 | 2023-09-12 | 哈尔滨工业大学（深圳） | Communication compression method for distributed learning and related equipment |
| CN116132184A (en) * | 2023-02-17 | 2023-05-16 | 中国工商银行股份有限公司 | Method, device and server for detecting distributed denial of service attack |
| CN116503420B (en) * | 2023-04-26 | 2024-05-14 | 佛山科学技术学院 | Image segmentation method based on federal learning and related equipment |
| CN116756764B (en) * | 2023-05-04 | 2024-06-04 | 浙江大学 | Model blocking aggregation privacy protection method for lithography hotspot detection |
| CN116596865B (en) * | 2023-05-05 | 2024-04-16 | 深圳市大数据研究院 | Defect detection method, defect detection system and robot |
| CN116258420B (en) * | 2023-05-11 | 2023-08-01 | 中南大学 | Product quality detection method, device, terminal equipment and medium |
| CN116432867A (en) * | 2023-06-09 | 2023-07-14 | 日照鲁光电子科技有限公司 | Diode preparation control optimization method and system |
| CN117237804B (en) * | 2023-09-15 | 2024-02-13 | 江苏三棱智慧物联发展股份有限公司 | Pyrotechnical recognition system and method based on federal learning |
| CN117688425B (en) * | 2023-12-07 | 2024-07-16 | 重庆大学 | Multi-task graph classification model construction method and system for Non-IID graph data |
| CN117575423B (en) * | 2024-01-10 | 2024-04-16 | 湖南工商大学 | Industrial product quality detection method based on federal learning system and related equipment |
| CN117579399B (en) * | 2024-01-17 | 2024-05-14 | 北京智芯微电子科技有限公司 | Training method and system of abnormal flow detection model and abnormal flow detection method |
| CN118095579B (en) * | 2024-04-26 | 2024-08-30 | 宁德时代新能源科技股份有限公司 | Method, device and system for determining process parameters, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110309847A (en) * | 2019-04-26 | 2019-10-08 | 深圳前海微众银行股份有限公司 | A kind of model compression method and device |
| CN111461291A (en) * | 2020-03-13 | 2020-07-28 | 西安科技大学 | Long-distance pipeline inspection method based on YO L Ov3 pruning network and deep learning defogging model |
| WO2020164270A1 (en) * | 2019-02-15 | 2020-08-20 | 平安科技（深圳）有限公司 | Deep-learning-based pedestrian detection method, system and apparatus, and storage medium |
| CN111598238A (en) * | 2020-07-17 | 2020-08-28 | 支付宝(杭州)信息技术有限公司 | Compression method and device of deep learning model |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109754105B (en) * | 2017-11-07 | 2024-01-05 | 华为技术有限公司 | Prediction method, terminal and server |
| CN111126627B (en) * | 2019-12-25 | 2023-07-04 | 四川新网银行股份有限公司 | Model training system based on separation index |
| CN111310932B (en) * | 2020-02-10 | 2024-09-27 | 深圳前海微众银行股份有限公司 | Method, device, equipment and readable storage medium for optimizing transverse federal learning system |
| CN111382706A (en) * | 2020-03-10 | 2020-07-07 | 深圳前海微众银行股份有限公司 | Prediction method and device based on federal learning, storage medium and remote sensing equipment |
-
2020
- 2020-10-20 CN CN202011125718.5A patent/CN112257774B/en active Active
- 2020-11-27 WO PCT/CN2020/131986 patent/WO2021189906A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020164270A1 (en) * | 2019-02-15 | 2020-08-20 | 平安科技（深圳）有限公司 | Deep-learning-based pedestrian detection method, system and apparatus, and storage medium |
| CN110309847A (en) * | 2019-04-26 | 2019-10-08 | 深圳前海微众银行股份有限公司 | A kind of model compression method and device |
| CN111461291A (en) * | 2020-03-13 | 2020-07-28 | 西安科技大学 | Long-distance pipeline inspection method based on YO L Ov3 pruning network and deep learning defogging model |
| CN111598238A (en) * | 2020-07-17 | 2020-08-28 | 支付宝(杭州)信息技术有限公司 | Compression method and device of deep learning model |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113379708A (en) * | 2021-02-26 | 2021-09-10 | 山东大学 | Air conditioner external unit appearance detection method and system based on federal learning |
| CN113179244A (en) * | 2021-03-10 | 2021-07-27 | 上海大学 | Federal deep network behavior feature modeling method for industrial internet boundary safety |
| CN113033652A (en) * | 2021-03-23 | 2021-06-25 | 电子科技大学 | Image recognition system and method based on block chain and federal learning |
| CN113139662B (en) * | 2021-04-23 | 2023-07-14 | 深圳市大数据研究院 | Global and local gradient processing method, device, equipment and medium for federal learning |
| CN113139662A (en) * | 2021-04-23 | 2021-07-20 | 深圳市大数据研究院 | Global and local gradient processing method, device, equipment and medium for federal learning |
| CN113344221A (en) * | 2021-05-10 | 2021-09-03 | 上海大学 | Federal learning method and system based on neural network architecture search |
| CN113364543A (en) * | 2021-06-03 | 2021-09-07 | 厦门大学 | Edge calculation model training method based on federal reinforcement learning |
| CN113364543B (en) * | 2021-06-03 | 2022-03-15 | 厦门大学 | Edge calculation model training method based on federal reinforcement learning |
| CN114169412A (en) * | 2021-11-23 | 2022-03-11 | 北京邮电大学 | Federal learning model training method for large-scale industrial chain privacy calculation |
| CN114169412B (en) * | 2021-11-23 | 2024-09-06 | 北京邮电大学 | Federal learning model training method for large-scale industry chain privacy calculation |
| CN114241305A (en) * | 2021-12-07 | 2022-03-25 | 中原动力智能机器人有限公司 | Target detection model updating method and device |
| CN114339252B (en) * | 2021-12-31 | 2023-10-31 | 深圳大学 | Data compression method and device |
| CN114339252A (en) * | 2021-12-31 | 2022-04-12 | 深圳大学 | Data compression method and device |
| CN114782738A (en) * | 2022-03-25 | 2022-07-22 | 苏州浪潮智能科技有限公司 | Federal learning safety detection method and device, electronic equipment and readable storage medium |
| CN114782738B (en) * | 2022-03-25 | 2024-06-25 | 苏州浪潮智能科技有限公司 | Federal learning security detection method and device, electronic equipment and readable storage medium |
| CN114782758A (en) * | 2022-06-21 | 2022-07-22 | 平安科技（深圳）有限公司 | Image processing model training method, system, computer device and storage medium |
| CN114782758B (en) * | 2022-06-21 | 2022-09-02 | 平安科技（深圳）有限公司 | Image processing model training method, system, computer device and storage medium |
| CN116523081A (en) * | 2023-04-07 | 2023-08-01 | 花瓣云科技有限公司 | Data standardization method and device |
| CN116523081B (en) * | 2023-04-07 | 2024-02-13 | 花瓣云科技有限公司 | Data standardization method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021189906A1 (en) | 2021-09-30 |
| CN112257774B (en) | 2024-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112257774A (en) | Target detection method, device, equipment and storage medium based on federal learning | |
| CN112446544A (en) | Traffic flow prediction model training method and device, electronic equipment and storage medium | |
| CN111814962A (en) | Method and device for acquiring parameters of recognition model, electronic equipment and storage medium | |
| CN112137591B (en) | Target object position detection method, device, equipment and medium based on video stream | |
| CN112396005A (en) | Biological characteristic image recognition method and device, electronic equipment and readable storage medium | |
| CN111783982B (en) | Method, device, equipment and medium for acquiring attack sample | |
| CN112801718A (en) | User behavior prediction method, device, equipment and medium | |
| CN113946690A (en) | Potential customer mining method and device, electronic equipment and storage medium | |
| CN112269875B (en) | Text classification method, device, electronic equipment and storage medium | |
| CN113516417A (en) | Service evaluation method and device based on intelligent modeling, electronic equipment and medium | |
| CN112990374A (en) | Image classification method, device, electronic equipment and medium | |
| CN114511038A (en) | False news detection method and device, electronic equipment and readable storage medium | |
| CN114781832A (en) | Course recommendation method and device, electronic equipment and storage medium | |
| CN114491047A (en) | Multi-label text classification method and device, electronic equipment and storage medium | |
| CN113157739A (en) | Cross-modal retrieval method and device, electronic equipment and storage medium | |
| CN113868529A (en) | Knowledge recommendation method and device, electronic equipment and readable storage medium | |
| CN112396547A (en) | Course recommendation method, device, equipment and medium based on unsupervised learning | |
| CN112288371A (en) | Customs clearance inspection method and device, electronic equipment and computer readable storage medium | |
| CN114494800A (en) | Prediction model training method and device, electronic equipment and storage medium | |
| CN114187489A (en) | Vehicle abnormal driving risk detection method and device, electronic equipment and storage medium | |
| CN113658002A (en) | Decision tree-based transaction result generation method and device, electronic equipment and medium | |
| CN113591881A (en) | Intention recognition method and device based on model fusion, electronic equipment and medium | |
| CN113627160A (en) | Text error correction method and device, electronic equipment and storage medium | |
| CN113313211A (en) | Text classification method and device, electronic equipment and storage medium | |
| CN115049836B (en) | Image segmentation method, device, equipment and storage medium |
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 |