CN115987694B - Multi-domain federation-based device privacy protection method, system and device - Google Patents
Multi-domain federation-based device privacy protection method, system and device Download PDFInfo
- Publication number
- CN115987694B CN115987694B CN202310272999.4A CN202310272999A CN115987694B CN 115987694 B CN115987694 B CN 115987694B CN 202310272999 A CN202310272999 A CN 202310272999A CN 115987694 B CN115987694 B CN 115987694B
- Authority
- CN
- China
- Prior art keywords
- domain
- model
- equipment
- parameters
- internet
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
- Computer And Data Communications (AREA)
Abstract
The embodiment provides a device privacy protection method, system and device based on multi-domain federation. According to the method, the device and the system, the three-layer data sharing structure of the core device, the domain center device and the intra-domain internet of things device is constructed by dividing the internet of things devices in the whole domain, and the three-layer data sharing structure is matched to realize target detection or abnormality detection in an application scene on the premise of protecting the privacy of each internet of things device in the application scene.
Description
Technical Field
The application relates to the internet of things security technology, in particular to a device privacy protection method, system and device based on multi-domain federation.
Background
In the Internet of things, the number of the Internet of things devices is huge, the device types are numerous, and the data types acquired by the Internet of things devices of different device types are inconsistent. The data types herein are, for example, images, video, sound, infrared, vibration, light signals, GPS information, industrial data, etc. For example, the method is applied to traffic scenes, and a plurality of internet of things devices of different device types are deployed at a traffic intersection so as to respectively collect data of different data types such as images, videos, audios, GPS information and the like through the deployed internet of things devices.
The occurrence of any event in any application scene can be related to the data stored by a plurality of internet of things devices in the application scene, and the data collected by each internet of things device are different in event analysis application, but have the same meaning. Based on this, in order to realize target detection or anomaly detection in a certain application scenario, it is necessary to rely on data in multiple internet of things devices in the application scenario, which may involve private data or data that is not wanted to be shared in many internet of things devices, where the data is not shared, and may affect target detection or anomaly detection in the application scenario.
Disclosure of Invention
The embodiment of the application provides a device privacy protection method, a device privacy protection system and a device based on multi-domain federation, so that target detection or anomaly detection in an application scene is realized on the premise of protecting the privacy of each Internet of things device in the application scene.
The embodiment of the application provides a device privacy protection method based on multi-domain federation, which is applied to core devices of a target application scene and comprises the following steps:
sharing the model information of the federal model among the current domains to the domain center equipment of each domain; the Internet of things equipment in the target application scene is divided into at least one domain, and any domain is provided with corresponding domain center equipment; the model information includes: current model structure, current model parameters, current model state; the current model state is a first state or a second state, the first state represents model convergence, and the first state is used for indicating each domain center device to generate a target domain model based on the current model structure and the current model parameters and sharing the target domain model to the Internet of things devices in the domain for target detection or anomaly detection; the second state indicates that the model is not converged and is used for indicating that the model is continuously trained;
Receiving comprehensive parameters reported by the central equipment in each domain; the comprehensive parameters reported by any domain center equipment are determined based on training parameters trained by all Internet of things equipment in the domain, the training parameters trained by any Internet of things equipment are obtained by training a domain model when the current model state is a second state, and the domain model is a model corresponding to the domain determined based on the current model structure and the current model parameters;
determining inter-domain federation model parameters based on the comprehensive parameters reported by the domain center devices, re-determining a current inter-domain federation model according to the inter-domain federation model parameters and the current model structure, determining the current model state of the current inter-domain federation model according to the inter-domain federation model parameters, and returning the model information of the current inter-domain federation model to the domain center devices of the domains.
The embodiment of the application provides a device privacy protection method based on multi-domain federation, which is applied to domain center devices of any domain in a target application scene, wherein the Internet of things device in the target application scene is divided into at least one domain; the method comprises the following steps:
receiving model information of a federation model between current domains shared by core equipment of the target application scene;
When the current model state carried by the model information is a second state representing that the model is not converged, determining a domain model corresponding to the domain based on a current model structure and current model parameters carried by the model information, sharing the domain model to Internet of things equipment in the domain, training the domain model by the Internet of things equipment based on existing training data, and obtaining corresponding training parameters after training is completed; the method comprises the steps of receiving training parameters reported by Internet of things equipment in a local domain, generating comprehensive parameters based on the training parameters reported by the Internet of things equipment in the local domain, reporting the comprehensive parameters to the core equipment, determining inter-domain federal model parameters by the core equipment based on the comprehensive parameters reported by the domain center equipment, re-determining a current inter-domain federal model according to the inter-domain federal model parameters and the current model structure, determining the current model state of the current inter-domain federal model according to the inter-domain federal model parameters, and sharing model information of the current inter-domain federal model to the domain center equipment in each domain;
and when the current model state carried by the model information is a first state representing model convergence, determining a target intra-domain model corresponding to the local domain based on the current model structure and the current model parameters carried by the model information, and sharing the target intra-domain model to the Internet of things equipment in the local domain for target detection or anomaly detection.
The embodiment provides a device privacy protection system based on multi-domain federation, which comprises: the method comprises the steps of dividing internet of things equipment in a target application scene into at least one domain;
the core device performing the steps of the first method as above;
the domain center apparatus performs the steps in the second method as above.
The embodiment provides a device privacy protection device based on multi-domain federation, the device is applied to core devices of a target application scene, and the device comprises:
the inter-domain sharing unit is used for sharing the model information of the current inter-domain federation model to the domain center equipment of each domain; the Internet of things equipment in the target application scene is divided into at least one domain, and any domain is provided with corresponding domain center equipment; the model information includes at least: current model structure, current model parameters, current model state; the current model state is a first state or a second state, the first state represents model convergence, and the first state is used for indicating each domain center device to generate a target domain model based on the current model structure and the current model parameters and sharing the target domain model to the Internet of things devices in the domain for target detection or anomaly detection; the second state indicates that the model is not converged and is used for indicating that the model is continuously trained;
The parameter optimization unit is used for receiving the comprehensive parameters reported by the central equipment in each domain; the comprehensive parameters reported by any domain center equipment are determined based on training parameters trained by all Internet of things equipment in the domain, the training parameters trained by any Internet of things equipment are obtained by training a domain model when the current model state is a second state, and the domain model is a model corresponding to the domain determined based on the current model structure and the current model parameters; the method comprises the steps of,
determining inter-domain federation model parameters based on the comprehensive parameters reported by the inter-domain central equipment, re-determining a current inter-domain federation model according to the inter-domain federation model parameters and the current model structure, determining the current model state of the current inter-domain federation model according to the inter-domain federation model parameters, and triggering an inter-domain sharing unit to continuously execute the step of sharing the model information of the current inter-domain federation model to the inter-domain central equipment of each domain.
The embodiment provides a device privacy protection device based on multi-domain federation, which is applied to domain center devices of any domain in a target application scene, wherein the Internet of things devices in the target application scene are divided into at least one domain; the device comprises:
The receiving unit is used for receiving the model information of the federation model between the current domains shared by the core equipment of the target application scene;
the processing unit is used for determining a domain model corresponding to the local domain based on a current model structure and current model parameters carried by the model information when the current model state carried by the model information is a second state representing that the model is not converged, sharing the domain model to the Internet of things equipment in the local domain, so that the Internet of things equipment trains the domain model based on the existing training data and obtains corresponding training parameters after training is completed; the method comprises the steps of receiving training parameters reported by Internet of things equipment in a local domain, generating comprehensive parameters based on the training parameters reported by the Internet of things equipment in the local domain, reporting the comprehensive parameters to the core equipment, determining inter-domain federal model parameters by the core equipment based on the comprehensive parameters reported by the domain center equipment, re-determining a current inter-domain federal model according to the inter-domain federal model parameters and the current model structure, determining the current model state of the current inter-domain federal model according to the inter-domain federal model parameters, and sharing model information of the current inter-domain federal model to the domain center equipment in each domain; the method comprises the steps of,
And when the current model state carried by the model information is a first state representing model convergence, determining a target intra-domain model corresponding to the local domain based on the current model structure and the current model parameters carried by the model information, and sharing the target intra-domain model to the Internet of things equipment in the local domain for target detection or anomaly detection.
The present embodiment provides an electronic device, which is characterized in that the electronic device includes: a processor and a machine-readable storage medium having stored thereon computer instructions which when executed by the processor perform the steps of any of the methods described above.
According to the technical scheme, the three-layer data sharing structure of the core equipment, the domain center equipment and the intra-domain Internet of things equipment is constructed by dividing the Internet of things equipment in the whole domain, the three-layer data sharing structure is matched based on the three-layer data sharing structure, so that the fact that the intra-domain Internet of things equipment does not report private data any more is guaranteed through a transverse federal mode, model parameters are reported to the domain center equipment, the domain center equipment fuses model parameters reported by the intra-domain Internet of things equipment through the transverse federal mode to obtain comprehensive parameters and report the comprehensive parameters to the core equipment, the core equipment fuses the comprehensive parameters reported by the domain center equipment to train out an optimal inter-domain federal model, and then the domain center equipment obtains an optimal target intra-domain model in the domain based on the optimal inter-domain federal model to perform target detection or anomaly detection. The method and the device realize target detection or anomaly detection in the application scene on the premise of protecting the privacy of each Internet of things device in the application scene;
Furthermore, the trained optimal inter-domain federation model is obtained by fusing contributions of all the whole domain Internet of things devices, so that the mining event characteristics can be deeper than those of the common model, the overall data analysis capability is greatly improved on the premise of protecting the data privacy of all the Internet of things devices, and the accuracy of target detection or anomaly detection of the target intra-domain model is improved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure.
FIG. 1 is a flow chart of a method provided in an embodiment of the present application;
fig. 2 is a scene structure diagram provided in an embodiment of the present application;
FIG. 3 is a flowchart of another method provided by an embodiment of the present application;
fig. 4 is a device structure diagram provided in an embodiment of the present application;
FIG. 5 is a block diagram of another device according to an embodiment of the present application;
fig. 6 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present application as detailed in the accompanying claims.
The terminology used in the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used in this application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
In order to better understand the technical solutions provided by the embodiments of the present application and make the above objects, features and advantages of the embodiments of the present application more obvious, the technical solutions in the embodiments of the present application are described in further detail below with reference to the accompanying drawings.
Referring to fig. 1, fig. 1 is a flowchart of a method provided in an embodiment of the present application. The application scenario involved in the flow is denoted as a target application scenario. In order to implement the method provided in this embodiment, the present embodiment needs to perform domain division on the internet of things devices in the target application scene, such as a video camera, a sound receiver, a GPS, and the like, to obtain at least one domain (Field), denoted as { Field1, field2, field3 … Field n }. Optionally, each domain includes at least one internet of things device in the target application scenario.
As an embodiment, the domains may be divided according to data types, for example, internet of things devices with the same data type of the collected data are divided in the same domain, and internet of things devices with different data types of the collected data are divided in different domains.
In this embodiment, each domain is managed by the corresponding domain center apparatus. As an embodiment, the domain center device of any domain may be an internet of things device in the domain, or may be another device independent of the domain, or may even be a virtual character. The domain center device of any domain is used for taking charge of fusion calculation of shared data of all internet of things devices in the domain, and will be described hereinafter, which is not repeated here.
In this embodiment, the target application scenario also involves a core device (centrmain). The core device is responsible for fusion calculation of all shared data between domains, as will be described below. As an embodiment, the core device may be an internet of things device in the target application scenario, or may be another device independent of the target application scenario, or may even be a virtual character. Fig. 2 illustrates a target application scenario as described above.
As shown in fig. 1, the process may include the steps of:
and step 101, sharing the model information of the federation model among the current domains in the target application scene to the domain center equipment of each domain.
Initially, the core device builds an initialization inter-domain federation model under a target application scene, and the built initialization inter-domain federation model is recorded as a current inter-domain federation model under the target application scene. It should be noted that federation learning is a privacy protection method, and is mainly divided into a horizontal federation and a vertical federation, which are respectively used for processing different data bodies of the same data type and data sharing methods under the premise of privacy protection of different data types of the same data body.
As an embodiment, the core device may construct an initial inter-domain federation model according to the number of domains in the target application scenario. For example, the initial inter-domain federation model built is a weight-based logistic regression model: y=w1+w2+w2+w3+w3+ … +wm+wm+b. Wherein W1-WM, W1-WM, b are model parameters, M represents the sum of feature numbers of all domains, W1-WM represents feature weights (alternatively, W1-WM may be randomly generated in the initialization process and take a value in the range of 0-1), W1-WM represents feature data values, and b represents the bias of the regression equation; for another example, the initial inter-domain federal model constructed is a feature stitching based deep learning model: the deep learning model consists of sub-models corresponding to each domain. Model parameters in the deep learning model can be assigned according to actual requirements in the initial stage.
Initially, the current model state of the current inter-domain federation model is a second state representing that the model does not converge, the second state indicating that each participant device needs to continue to participate in model training to train the optimal inter-domain federation model.
When not initially, the current inter-domain federal model is determined by the core device and the core devices in each domain, and corresponding model information is described in step 103 below, which is not repeated herein.
In this embodiment, the model information of the current inter-domain federation model at least includes: current model structure, current model parameters, current model state. The current model state may be the first state in addition to the second state described above, and will be described below, which is not repeated here.
And 102, receiving comprehensive parameters reported by the central equipment in each domain.
After the model information of the current inter-domain federation model is shared to the domain center devices of the domains, as shown in fig. 3, if the domain center device determines that the current inter-domain federation model is not converged based on the current model state in the model information (at this time, the model state of the current inter-domain federation model is a second state indicating that the model is not converged), the intra-domain model is determined according to the current model structure, the current model parameters and the characteristic parameters specified by the local domain in the model information.
As an embodiment, in this embodiment, there are many ways for the domain center device to determine the domain model according to the current model structure, the current model parameters and the specified characteristic parameters of the domain in the model information, for example, based on the current model structure and the current model parameters, the current inter-domain federation model is recovered, and the domain model corresponding to the domain is disassembled from the current inter-domain federation model. Here, if the current inter-domain federation model is a weight-based logistic regression model: y=w1+w1+w2+w3+w3+ … +wm+wm+b, then the specified feature parameter (also referred to as feature data value) Wi of the local domain can be found from the logistic regression model, an algorithm model related to the specified feature parameter of the local domain is selected from the logistic regression model based on the specified feature parameter of the local domain, for example, if the feature parameters specified in the domain are W1, W2, and W3 … Wi, the algorithm model related to the feature parameters specified in the domain is w1×w1+w2+w3×w3+ … +wi+wi. For another example, if the current inter-domain federal model is a deep learning model, sub-models corresponding to the domains are set in the deep learning model. The sub-model corresponding to the local domain is directly extracted from the deep learning model, and the sub-model corresponding to the local domain is used as the intra-domain model of the local domain.
After determining the domain model corresponding to the domain, the domain center device shares the domain model with each Internet of things device in the domain, so that each Internet of things device in the domain performs model training based on the domain model and the existing training data, and trained model parameters (recorded as training parameters) are obtained when the model training is finished.
In specific implementation, each internet of things device in the local domain performs model training based on the intra-domain model and the existing training data, and stable model parameters (such as feature weights, biases and the like, collectively referred to as training parameters) are obtained after training is completed. And then, the Internet of things equipment sends the training parameters to the domain center equipment.
And when the domain center equipment receives the training parameters sent by the appointed number of the Internet of things equipment, homomorphic calculation is carried out on each received training parameter to obtain the comprehensive parameters, the comprehensive parameters are reported to the core equipment, and finally the core equipment receives the comprehensive parameters reported by the domain center equipment.
As an embodiment, the sending, by the internet of things device, the training parameter to the domain center device may include: and homomorphic encryption public keys generated by the domain center equipment of the domain are adopted to homomorphic encrypt the training parameters, and ciphertext is sent to the domain center equipment.
Likewise, the reporting, by the domain center device, the comprehensive parameters to the core device may include: and homomorphic encryption is carried out on the comprehensive parameters by utilizing the homomorphic encryption public key corresponding to the domain generated by the core equipment, and the ciphertext is sent to the core equipment.
In this embodiment, homomorphic encryption is a common cryptographic technique that ensures that homomorphic encrypted data is processed to produce an output, which is decrypted in the same way as the output produced by processing unencrypted original data in the same way. In specific implementation, the present embodiment is not limited to homomorphic encryption.
And 103, determining inter-domain federation model parameters based on the comprehensive parameters reported by the central equipment in each domain, re-determining a current inter-domain federation model according to the inter-domain federation model parameters and the current model structure, determining the current model state of the current inter-domain federation model according to the inter-domain federation model parameters, and returning to the step 101.
In this embodiment, after the core device receives the integrated parameters reported by the core device in each domain, it determines inter-domain federal model parameters based on the integrated parameters reported by the core device in each domain.
As described above, the overall parameters reported by the core device in each domain are encrypted by the homomorphic encryption public key corresponding to the domain, and referring to the homomorphic encryption characteristic, in this step 103, the core device determines, based on the overall parameters reported by the core device in each domain, the inter-domain federal model parameters may include: homomorphism calculation is performed on the comprehensive parameters reported by the central equipment in each domain to obtain inter-domain federal model parameters (denoted as P_x= { W1, W2, W3 … WM, b } (x represents the round)). The homomorphism calculation is one of the bases of federal learning, is a special encryption calculation method, and allows the ciphertext to be processed to obtain a result which is still encrypted, namely the result obtained by directly processing the ciphertext and encrypting the processing result after processing the plaintext is the same, so that the homomorphism of the data is ensured, and the embodiment is not particularly limited.
After obtaining the inter-domain federation model parameters, the current inter-domain federation model can be redetermined according to the inter-domain federation model parameters and the current model structure, for example, the existing parameters under the current model structure can be correspondingly updated to the inter-domain federation model parameters at the moment, and the updated model can be recorded as the current inter-domain federation model; for another example, the current model structure may be slightly adjusted according to actual requirements, and then the parameters under the adjusted model structure are correspondingly updated to inter-domain federal model parameters, the updated model may be recorded as the current inter-domain federal model, and the like, and the embodiment is not particularly limited.
Further, in this embodiment, the current model state of the current inter-domain federal model may be further determined according to inter-domain federal model parameters, for example, whether a model loss value (loss) meets a set requirement is determined, if yes, the current model state of the current inter-domain federal model is determined to be a first state, the first state represents model convergence, and is used for indicating each domain center device to generate a target domain model based at least on the current model structure and the current model parameters and share the target domain model with the internet of things device in the domain, where the target domain model is used for target detection or anomaly detection; if not, determining the current model state of the federal model among the current domains as the second state; for another example, whether the inter-domain federation model parameter and the model parameter of the last inter-domain federation model satisfy an approximation condition (for example, an error of the corresponding parameter is within a set range) is determined, if yes, the current model state of the current inter-domain federation model is determined to be the first state, and if not, the current model state of the current inter-domain federation model is determined to be the second state. Etc., the present embodiment is not particularly limited.
It can be seen that, in this embodiment, if the current model state of the current inter-domain federation model is the first state, after the above step is returned to share the model information of the current inter-domain federation model to the domain center devices of each domain, on the premise that the current model state is found to be the first state, each domain center device generates a model in the target domain based on the current model structure and the current model parameters, and shares the model in the target domain to the internet of things device in the current domain to perform target detection or anomaly detection. Because the latest parameters of the current inter-domain federation model are obtained after the contribution training parameters of all the Internet of things devices in the whole domain are fused, the event characteristics can be mined in a deeper level compared with the common model, and therefore the model capacity is optimized on the premise of protecting the data privacy of all the Internet of things devices.
Thus, the flow shown in fig. 1 is completed.
As can be seen from the flow shown in fig. 1, in this embodiment, by dividing the internet of things devices in the whole domain, a three-layer data sharing structure of the core device, the domain center device and the intra-domain internet of things device is constructed, and based on the three-layer data sharing structure, the internet of things devices in each domain are matched to ensure that the internet of things devices in each domain no longer report private data in a horizontal federal manner, but report model parameters to the domain center device, the domain center device fuses model parameters reported by the internet of things devices in each domain in the horizontal federal manner to obtain comprehensive parameters and report the comprehensive parameters to the core device, and then the core device fuses the comprehensive parameters reported by the domain center devices to train out an optimal inter-domain federal model, so that the domain center devices obtain an optimal target intra-domain model in the domain based on the optimal inter-domain federal model to perform target detection or anomaly detection. The method and the device realize target detection or anomaly detection in the application scene on the premise of protecting the privacy of each Internet of things device in the application scene;
Furthermore, the trained optimal inter-domain federation model is obtained by fusing contributions of all the whole domain Internet of things devices, so that the mining event characteristics can be deeper than those of the common model, the overall data analysis capability is greatly improved on the premise of protecting the data privacy of all the Internet of things devices, and the accuracy of target detection or anomaly detection of the target intra-domain model is improved.
In this embodiment, migration of the target inter-domain federation model may also be supported, for example, the target inter-domain federation model is shared to a core device of another application scenario that satisfies a similar condition to the target application scenario, so that the core device shares the target inter-domain federation model to a domain center device of each domain, and an internet of things device in each domain generates a target intra-domain model based on the target inter-domain federation model. Through the sharing, model training resources can be saved, and the model utilization rate is greatly improved. Here, the target inter-domain federation model refers to an inter-domain federation model in which the current model state is the first state.
The method provided by the embodiment of the application is described below in terms of a center-of-field device:
referring to fig. 3, fig. 3 is another flowchart provided in an embodiment of the present application. The method is applied to the domain center device of any domain, as shown in fig. 3, and comprises the following steps:
This step 302 is performed when the current model state carried by the model information is a second state indicating that the model is not converged. Once the current model state carried by the model information is the second state indicating that the model is not converged, this means that the current model structure and the current model parameters in the model information are not optimal currently and further training is required, so as to describe in step 302, the domain center device determines the domain model corresponding to the domain based on the current model structure and the current model parameters in the model information, shares the domain model to the internet of things device in the domain, and trains the domain model by the internet of things device based on the existing training data and obtains the corresponding training parameters after the training is completed.
As an embodiment, before this step 302, the domain center device sends, for each internet of things device in the domain, a homomorphic encryption public key corresponding to the internet of things device; based on this, applied to this step 302, the training parameters reported by any one of the internet of things devices in the present domain are encrypted by the homomorphic encryption public key corresponding to the internet of things device, and correspondingly, the generating the comprehensive parameters based on the training parameters reported by each of the internet of things devices in the present domain may include: and homomorphic calculation is carried out on training parameters reported by all the Internet of things equipment to obtain the comprehensive parameters.
And 303, determining a target domain model corresponding to the local domain based on the current model structure and the current model parameters carried by the model information, and sharing the target domain model to the Internet of things equipment in the local domain for target detection or anomaly detection.
This step 303 is performed when the current model state carried by the model information is the first state representing model convergence. Once the current model state carried by the model information is the first state representing model convergence, the current model structure and the current model parameter in the model information are currently optimal, and the target intra-domain model corresponding to the local domain can be determined directly according to the current model structure and the current model parameter in the model information and shared for the Internet of things equipment in the local domain to perform target detection or anomaly detection. Here, the manner of determining the target intra-domain model is similar to the manner of determining the intra-domain model described above, and the description of this embodiment is omitted.
Thus, the flow shown in fig. 3 is completed.
As can be seen from the flow shown in fig. 3, in this embodiment, by dividing the internet of things devices in the whole domain, a three-layer data sharing structure of the core device, the domain center device and the intra-domain internet of things device is constructed, and based on the three-layer data sharing structure, the internet of things devices in each domain are matched to ensure that the internet of things devices in each domain no longer report private data in a horizontal federal manner, but report model parameters to the domain center device, the domain center device fuses model parameters reported by the internet of things devices in each domain in the horizontal federal manner to obtain comprehensive parameters and report the comprehensive parameters to the core device, and then the core device fuses the comprehensive parameters reported by the domain center devices to train out an optimal inter-domain federal model, so that the domain center devices obtain an optimal target intra-domain model in the domain based on the optimal inter-domain federal model to perform target detection or anomaly detection. The method and the device realize target detection or anomaly detection in the application scene on the premise of protecting the privacy of each Internet of things device in the application scene;
furthermore, the trained optimal inter-domain federation model is obtained by fusing contributions of all the whole domain Internet of things devices, so that the mining event characteristics can be deeper than those of the common model, the overall data analysis capability is greatly improved on the premise of protecting the data privacy of all the Internet of things devices, and the accuracy of target detection or anomaly detection of the target intra-domain model is improved.
The method provided by the embodiment of the application is described above, and the system and the device provided by the embodiment of the application are described below:
the embodiment of the application provides a device privacy protection system based on multi-domain federation. The system may include: the method comprises the steps of dividing internet of things equipment in a target application scene into at least one domain;
wherein the core device performs the steps in the flow shown in fig. 1; the domain center apparatus performs the steps in the flow shown in fig. 3.
The embodiment also provides a device privacy protection device based on multi-domain federation, which is specifically shown in fig. 4. Referring to fig. 4, fig. 4 is a block diagram of an apparatus according to an embodiment of the present application. As shown in fig. 4, the apparatus may include:
the inter-domain sharing unit is used for sharing the model information of the current inter-domain federation model to the domain center equipment of each domain; the Internet of things equipment in the target application scene is divided into at least one domain, and any domain is provided with corresponding domain center equipment; the model information includes at least: current model structure, current model parameters, current model state; the current model state is a first state or a second state, the first state represents model convergence, and the first state is used for indicating each domain center device to generate a target domain model based on the current model structure and the current model parameters and sharing the target domain model to the Internet of things devices in the domain for target detection or anomaly detection; the second state indicates that the model is not converged and is used for indicating that the model is continuously trained;
The parameter optimization unit is used for receiving the comprehensive parameters reported by the central equipment in each domain; the comprehensive parameters reported by any domain center equipment are determined based on training parameters trained by all Internet of things equipment in the domain, the training parameters trained by any Internet of things equipment are obtained by training a domain model when the current model state is a second state, and the domain model is a model corresponding to the domain determined based on the current model structure and the current model parameters; the method comprises the steps of,
determining inter-domain federation model parameters based on the comprehensive parameters reported by the inter-domain central equipment, re-determining a current inter-domain federation model according to the inter-domain federation model parameters and the current model structure, determining the current model state of the current inter-domain federation model according to the inter-domain federation model parameters, and triggering an inter-domain sharing unit to continuously execute the step of sharing the model information of the current inter-domain federation model to the inter-domain central equipment of each domain.
Optionally, the data types of the data collected by the internet of things devices in the same domain are the same;
the inter-domain sharing unit further sends the homomorphic encryption public key corresponding to each domain to domain center equipment of the domain;
The comprehensive parameters reported by the domain center equipment of any domain are encrypted by the homomorphic encryption public key corresponding to the domain; the parameter optimization unit determines inter-domain federal model parameters based on the comprehensive parameters reported by the central equipment in each domain, and the inter-domain federal model parameters comprise: homomorphic calculation is carried out on the comprehensive parameters reported by the central equipment in each domain to obtain the inter-domain federal model parameters;
the inter-domain sharing unit further shares the target inter-domain federation model to core equipment of other application scenes meeting similar conditions with the target application scene, so that the core equipment shares the target inter-domain federation model to domain center equipment of each domain, and the Internet of things equipment in each domain generates a target intra-domain model based on the target inter-domain federation model. Here, the target inter-domain federation model refers to an inter-domain federation model in which the current model state is the first state.
The structural description of the apparatus shown in fig. 4 is thus completed.
Referring to fig. 5, fig. 5 is a schematic diagram of another apparatus according to an embodiment of the present application. The device is applied to domain center equipment of any domain in a target application scene, and the Internet of things equipment in the target application scene is divided into at least one domain; the device comprises:
The receiving unit is used for receiving the model information of the federation model between the current domains shared by the core equipment of the target application scene;
the processing unit is used for determining a domain model corresponding to the local domain based on a current model structure and current model parameters carried by the model information when the current model state carried by the model information is a second state representing that the model is not converged, sharing the domain model to the Internet of things equipment in the local domain, so that the Internet of things equipment trains the domain model based on the existing training data and obtains corresponding training parameters after training is completed; the method comprises the steps of receiving training parameters reported by Internet of things equipment in a local domain, generating comprehensive parameters based on the training parameters reported by the Internet of things equipment in the local domain, reporting the comprehensive parameters to the core equipment, determining inter-domain federal model parameters by the core equipment based on the comprehensive parameters reported by the domain center equipment, re-determining a current inter-domain federal model according to the inter-domain federal model parameters and the current model structure, determining the current model state of the current inter-domain federal model according to the inter-domain federal model parameters, and sharing model information of the current inter-domain federal model to the domain center equipment in each domain; the method comprises the steps of,
And when the current model state carried by the model information is a first state representing model convergence, determining a target intra-domain model corresponding to the local domain based on the current model structure and the current model parameters carried by the model information, and sharing the target intra-domain model to the Internet of things equipment in the local domain for target detection or anomaly detection.
Optionally, the processing unit further sends the homomorphic encryption public key corresponding to each piece of internet of things equipment in the local domain to the piece of internet of things equipment;
training parameters reported by any Internet of things equipment in the domain are encrypted by homomorphic encryption public keys corresponding to the Internet of things equipment; the processing unit generates comprehensive parameters based on training parameters reported by all the Internet of things devices in the local area, and the comprehensive parameters comprise: and homomorphic calculation is carried out on training parameters reported by all the Internet of things equipment to obtain the comprehensive parameters.
The structural description of the apparatus shown in fig. 5 is thus completed.
Based on the same application concept as the above method, the embodiment of the present application further provides an electronic device, as shown in fig. 6, including: a processor and a machine-readable storage medium; the machine-readable storage medium stores machine-executable instructions executable by the processor; the processor is configured to execute machine-executable instructions to implement steps in the methods as described above in fig. 1 or 3.
Based on the same application concept as the above method, the embodiments of the present application further provide a machine-readable storage medium, where a number of computer instructions are stored, where the computer instructions can implement the method disclosed in the above example of the present application when executed by a processor.
By way of example, the machine-readable storage medium may be any electronic, magnetic, optical, or other physical storage device that can contain or store information, such as executable instructions, data, and the like. For example, a machine-readable storage medium may be: RAM (Radom Access Memory, random access memory), volatile memory, non-volatile memory, flash memory, a storage drive (e.g., hard drive), a solid state drive, any type of storage disk (e.g., optical disk, dvd, etc.), or a similar storage medium, or a combination thereof.
The system, apparatus, module or unit set forth in the above embodiments may be implemented in particular by a computer entity or by an article of manufacture having some functionality. A typical implementation device is a computer, which may be in the form of a personal computer, laptop computer, cellular telephone, camera phone, smart phone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or a combination of any of these devices.
For convenience of description, the above devices are described as being functionally divided into various units, respectively. Of course, the functions of each element may be implemented in one or more software and/or hardware elements when implemented in the present application.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Moreover, these computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and changes may be made to the present application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. which are within the spirit and principles of the present application are intended to be included within the scope of the claims of the present application.
Claims (12)
1. The device privacy protection method based on the multi-domain federation is characterized by being applied to core devices of a target application scene, wherein the core devices are Internet of things devices or virtual roles in the target application scene, and the method comprises the following steps:
sharing the model information of the federal model among the current domains to the domain center equipment of each domain; the method comprises the steps that Internet of things equipment in a target application scene is divided into at least one domain, any domain is provided with corresponding domain center equipment, and the Internet of things equipment with different data types of collected data is divided into different domains; the model information includes: current model structure, current model parameters, current model state; the current model state is a first state or a second state, the first state represents model convergence, and the first state is used for indicating each domain center device to generate a target domain model based on the current model structure and the current model parameters and sharing the target domain model to the Internet of things devices in the domain for target detection or anomaly detection; the second state indicates that the model is not converged and is used for indicating that the model is continuously trained;
receiving comprehensive parameters reported by the central equipment in each domain; the comprehensive parameters reported by the center equipment in any domain are determined based on training parameters trained by all the Internet of things equipment in the domain, the training parameters trained by the Internet of things equipment in any domain are obtained by training a domain model corresponding to the domain when the current model state is a second state, and the domain model is determined by the center equipment in the domain according to the current model structure shared by the core equipment, the current model parameters and the characteristic parameters appointed by the domain;
Determining inter-domain federation model parameters based on the comprehensive parameters reported by the domain center devices, re-determining a current inter-domain federation model according to the inter-domain federation model parameters and the current model structure, determining the current model state of the current inter-domain federation model according to the inter-domain federation model parameters, and returning the model information of the current inter-domain federation model to the domain center devices of the domains.
2. The method of claim 1, wherein data types of data collected by each internet of things device in the same domain are the same.
3. The method according to claim 1, characterized in that the method further comprises: for each domain, sending the homomorphic encryption public key corresponding to the domain to domain center equipment of the domain;
the comprehensive parameters reported by the domain center equipment of any domain are encrypted by the homomorphic encryption public key corresponding to the domain;
the determining the inter-domain federation model parameters based on the comprehensive parameters reported by the central equipment in each domain comprises the following steps: and homomorphic calculation is carried out on the comprehensive parameters reported by the central equipment in each domain to obtain the inter-domain federal model parameters.
4. The method according to claim 1, characterized in that the method further comprises:
Sharing the target inter-domain federation model to core equipment of other application scenes meeting similar conditions with the target application scene, so that the core equipment shares the target inter-domain federation model to domain center equipment of each domain, and generating a target intra-domain model based on the target inter-domain federation model by Internet of things equipment in each domain; the target inter-domain federation model refers to an inter-domain federation model with a current model state being a first state.
5. The device privacy protection method based on the multi-domain federation is characterized by being applied to domain center devices of any domain in a target application scene, wherein the Internet of things devices in the target application scene are divided into at least one domain, and the Internet of things devices with different data types of collected data are divided into different domains; the method comprises the following steps:
receiving model information of a federation model between current domains shared by core equipment of the target application scene; the core equipment is the Internet of things equipment in the target application scene or is a virtual role;
when the current model state carried by the model information is a second state representing that the model is not converged, determining a domain model corresponding to the domain according to the current model structure, the current model parameters and the characteristic parameters appointed by the domain shared by the core equipment, and sharing the domain model to the Internet of things equipment in the domain so as to train the domain model by the Internet of things equipment based on the existing training data and obtain corresponding training parameters after training is completed; the method comprises the steps of receiving training parameters reported by Internet of things equipment in a local domain, generating comprehensive parameters based on the training parameters reported by the Internet of things equipment in the local domain, reporting the comprehensive parameters to the core equipment, determining inter-domain federal model parameters by the core equipment based on the comprehensive parameters reported by the domain center equipment, re-determining a current inter-domain federal model according to the inter-domain federal model parameters and the current model structure, determining the current model state of the current inter-domain federal model according to the inter-domain federal model parameters, and sharing model information of the current inter-domain federal model to the domain center equipment in each domain;
And when the current model state carried by the model information is a first state representing model convergence, determining a target intra-domain model corresponding to the local domain based on the current model structure and the current model parameters carried by the model information, and sharing the target intra-domain model to the Internet of things equipment in the local domain for target detection or anomaly detection.
6. The method according to claim 5, characterized in that the method further comprises:
aiming at each piece of Internet of things equipment in the local area, sending the homomorphic encryption public key corresponding to the piece of Internet of things equipment;
training parameters reported by any Internet of things equipment in the domain are encrypted by homomorphic encryption public keys corresponding to the Internet of things equipment;
the generating of the comprehensive parameters based on the training parameters reported by the Internet of things devices in the local domain comprises the following steps: and homomorphic calculation is carried out on training parameters reported by all the Internet of things equipment to obtain the comprehensive parameters.
7. A multi-domain federal-based device privacy protection system, the system comprising: the method comprises the steps of dividing internet of things equipment in a target application scene into at least one domain;
The core device performing the operations of the method of any one of claims 1 to 4;
the domain center apparatus performs the operations in the method of any one of claims 5 to 6.
8. The utility model provides a device privacy protection device based on multi-domain federation which characterized in that, the device is applied to the core equipment of target application scene, core equipment is the thing networking equipment in the target application scene, perhaps is virtual role, and the device includes:
the inter-domain sharing unit is used for sharing the model information of the current inter-domain federation model to the domain center equipment of each domain; the method comprises the steps that Internet of things equipment in a target application scene is divided into at least one domain, the Internet of things equipment with different data types of collected data is divided into different domains, and any domain is provided with corresponding domain center equipment; the model information includes at least: current model structure, current model parameters, current model state; the current model state is a first state or a second state, the first state represents model convergence, and the first state is used for indicating each domain center device to generate a target domain model based on the current model structure and the current model parameters and sharing the target domain model to the Internet of things devices in the domain for target detection or anomaly detection; the second state indicates that the model is not converged and is used for indicating that the model is continuously trained;
The parameter optimization unit is used for receiving the comprehensive parameters reported by the central equipment in each domain; the comprehensive parameters reported by the center equipment in any domain are determined based on training parameters trained by all the Internet of things equipment in the domain, the training parameters trained by the Internet of things equipment in any domain are obtained by training a domain model corresponding to the domain when the current model state is a second state, and the domain model is determined by the center equipment in the domain according to the current model structure shared by the core equipment, the current model parameters and the characteristic parameters appointed by the domain; the method comprises the steps of,
determining inter-domain federation model parameters based on the comprehensive parameters reported by the inter-domain central equipment, re-determining a current inter-domain federation model according to the inter-domain federation model parameters and the current model structure, determining the current model state of the current inter-domain federation model according to the inter-domain federation model parameters, and triggering an inter-domain sharing unit to continuously execute the step of sharing the model information of the current inter-domain federation model to the inter-domain central equipment of each domain.
9. The apparatus of claim 8, wherein data types of data collected by each internet of things device in the same domain are the same;
The inter-domain sharing unit further sends the homomorphic encryption public key corresponding to each domain to domain center equipment of the domain;
the comprehensive parameters reported by the domain center equipment of any domain are encrypted by the homomorphic encryption public key corresponding to the domain; the parameter optimization unit determines inter-domain federal model parameters based on the comprehensive parameters reported by the central equipment in each domain, and the inter-domain federal model parameters comprise: homomorphic calculation is carried out on the comprehensive parameters reported by the central equipment in each domain to obtain the inter-domain federal model parameters;
the inter-domain sharing unit further shares the target inter-domain federation model to core equipment of other application scenes meeting similar conditions with the target application scene, so that the core equipment shares the target inter-domain federation model to domain center equipment of each domain, and the Internet of things equipment in each domain generates a target intra-domain model based on the target inter-domain federation model; the target inter-domain federation model refers to an inter-domain federation model with a current model state being a first state.
10. The device privacy protection device based on the multi-domain federation is characterized in that the device is applied to domain center devices of any domain in a target application scene, the Internet of things devices in the target application scene are divided into at least one domain, and the Internet of things devices with different data types of collected data are divided into different domains; the device comprises:
The receiving unit is used for receiving the model information of the federation model between the current domains shared by the core equipment of the target application scene; the core equipment is the Internet of things equipment in the target application scene or is a virtual role;
the processing unit is used for determining an intra-domain model corresponding to the local domain based on the current model structure, the current model parameters and the characteristic parameters appointed by the local domain which are shared according to the core equipment when the current model state carried by the model information is a second state representing that the model is not converged, sharing the intra-domain model to the Internet of things equipment in the local domain, so that the Internet of things equipment trains the intra-domain model based on the existing training data and obtains the corresponding training parameters after training is completed; the method comprises the steps of receiving training parameters reported by Internet of things equipment in a local domain, generating comprehensive parameters based on the training parameters reported by the Internet of things equipment in the local domain, reporting the comprehensive parameters to the core equipment, determining inter-domain federal model parameters by the core equipment based on the comprehensive parameters reported by the domain center equipment, re-determining a current inter-domain federal model according to the inter-domain federal model parameters and the current model structure, determining the current model state of the current inter-domain federal model according to the inter-domain federal model parameters, and sharing model information of the current inter-domain federal model to the domain center equipment in each domain; the method comprises the steps of,
And when the current model state carried by the model information is a first state representing model convergence, determining a target intra-domain model corresponding to the local domain based on the current model structure and the current model parameters carried by the model information, and sharing the target intra-domain model to the Internet of things equipment in the local domain for target detection or anomaly detection.
11. The apparatus of claim 10, wherein the processing unit further sends, for each internet of things device in the home domain, a homomorphic encryption public key corresponding to the internet of things device;
training parameters reported by any Internet of things equipment in the domain are encrypted by homomorphic encryption public keys corresponding to the Internet of things equipment; the processing unit generates comprehensive parameters based on training parameters reported by all the Internet of things devices in the local area, and the comprehensive parameters comprise: and homomorphic calculation is carried out on training parameters reported by all the Internet of things equipment to obtain the comprehensive parameters.
12. An electronic device, comprising: a processor and a machine-readable storage medium having stored thereon computer instructions which, when executed by a processor, implement the steps in the method of any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310272999.4A CN115987694B (en) | 2023-03-20 | 2023-03-20 | Multi-domain federation-based device privacy protection method, system and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310272999.4A CN115987694B (en) | 2023-03-20 | 2023-03-20 | Multi-domain federation-based device privacy protection method, system and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115987694A CN115987694A (en) | 2023-04-18 |
| CN115987694B true CN115987694B (en) | 2023-06-27 |
Family
ID=85965249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310272999.4A Active CN115987694B (en) | 2023-03-20 | 2023-03-20 | Multi-domain federation-based device privacy protection method, system and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115987694B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114254386A (en) * | 2021-12-13 | 2022-03-29 | 北京理工大学 | Federated learning privacy protection system and method based on hierarchical aggregation and block chain |
| CN114465900A (en) * | 2022-03-01 | 2022-05-10 | 北京邮电大学 | Data sharing delay optimization method and device based on federal edge learning |
| CN115033780A (en) * | 2022-05-06 | 2022-09-09 | 中国人民大学 | Privacy protection cross-domain recommendation system based on federal learning |
| WO2022206498A1 (en) * | 2021-03-31 | 2022-10-06 | 华为技术有限公司 | Federated transfer learning-based model training method and computing nodes |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112434313A (en) * | 2020-11-11 | 2021-03-02 | 北京邮电大学 | Data sharing method, system, electronic device and storage medium |
| CN112583575B (en) * | 2020-12-04 | 2023-05-09 | 华侨大学 | Federal learning privacy protection method based on homomorphic encryption in Internet of vehicles |
| CN113435472A (en) * | 2021-05-24 | 2021-09-24 | 西安电子科技大学 | Vehicle-mounted computing power network user demand prediction method, system, device and medium |
| CN113420888B (en) * | 2021-06-03 | 2023-07-14 | 中国石油大学(华东) | Non-supervision federal learning method based on generalization domain self-adaption |
| CN114116198A (en) * | 2021-10-21 | 2022-03-01 | 西安电子科技大学 | Asynchronous federal learning method, system, equipment and terminal for mobile vehicle |
| CN115455476A (en) * | 2022-09-16 | 2022-12-09 | 中国电力科学研究院有限公司 | Longitudinal federal learning privacy protection method and system based on multi-key homomorphic encryption |
| CN115358487A (en) * | 2022-09-21 | 2022-11-18 | 国网河北省电力有限公司信息通信分公司 | Federal learning aggregation optimization system and method for power data sharing |
| CN115587381B (en) * | 2022-12-12 | 2023-04-07 | 四川大学华西医院 | Medical diagnosis model combined training method and system based on differential privacy |
-
2023
- 2023-03-20 CN CN202310272999.4A patent/CN115987694B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022206498A1 (en) * | 2021-03-31 | 2022-10-06 | 华为技术有限公司 | Federated transfer learning-based model training method and computing nodes |
| CN114254386A (en) * | 2021-12-13 | 2022-03-29 | 北京理工大学 | Federated learning privacy protection system and method based on hierarchical aggregation and block chain |
| CN114465900A (en) * | 2022-03-01 | 2022-05-10 | 北京邮电大学 | Data sharing delay optimization method and device based on federal edge learning |
| CN115033780A (en) * | 2022-05-06 | 2022-09-09 | 中国人民大学 | Privacy protection cross-domain recommendation system based on federal learning |
| CN115510316A (en) * | 2022-05-06 | 2022-12-23 | 中国人民大学 | Privacy protection cross-domain recommendation system based on federal representation learning |
Non-Patent Citations (3)
| Title |
|---|
| Federated Transfer Learning Based Cross-Domain Prediction for Smart Manufacturing;Kevin I-Kai Wang;《IEEE Transactions on Industrial Informatics》;全文 * |
| 联邦学习在保险行业的应用研究;王春凯;冯键;;《保险职业学院学报》(第01期);全文 * |
| 联邦学习安全与隐私保护研究综述;周俊;方国英;吴楠;;《西华大学学报(自然科学版)》(第04期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115987694A (en) | 2023-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109255247B (en) | Multi-party security calculation method and device and electronic equipment | |
| CN113765657B (en) | Key data processing method, device and server | |
| CN108632292B (en) | Data sharing method and system based on alliance chain | |
| CN114860735A (en) | Method and device for inquiring hiding trace | |
| CN111213170B (en) | Asset hosting method, storage medium, blockchain system and blockchain node | |
| CN111291411B (en) | Safe video anomaly detection system and method based on convolutional neural network | |
| CN107135408A (en) | A kind of method for authenticating and device of video flowing address | |
| CN113947211A (en) | Federal learning model training method and device, electronic equipment and storage medium | |
| CN114648130A (en) | Longitudinal federal learning method and device, electronic equipment and storage medium | |
| CN107105199A (en) | Smart home nurse method and system based on technology of Internet of things | |
| CN114169537B (en) | Federal learning method and system for longitudinal xgboost decision tree | |
| Lidkea et al. | Convolutional neural network framework for encrypted image classification in cloud-based ITS | |
| CN111669434B (en) | Method, system, device and equipment for establishing communication group | |
| CN111639367A (en) | Tree model-based two-party combined classification method, device, equipment and medium | |
| Cheng et al. | SecureAD: A secure video anomaly detection framework on convolutional neural network in edge computing environment | |
| CN112801307B (en) | Block chain-based federal learning method and device and computer equipment | |
| CN114500119A (en) | Block chain service calling method and device | |
| CN115987694B (en) | Multi-domain federation-based device privacy protection method, system and device | |
| CN116502732B (en) | Federal learning method and system based on trusted execution environment | |
| CN117034287A (en) | Multiparty joint modeling safety enhancement method based on privacy computing interconnection and interworking technology | |
| CN117082505A (en) | Data security sharing method, system and device, electronic equipment and storage medium | |
| CN111294196A (en) | Signal sending and receiving method and device, electronic equipment and storage medium | |
| CN114547684A (en) | Method and device for protecting multi-party joint training tree model of private data | |
| CN113657685A (en) | Federal model training method, device, equipment, storage medium and program | |
| Hernandez-Castro et al. | Cryptanalysis of ubiquitous computing systems |
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 |