A deep learning data sharing method and device based on blockchain technology

Abstract

The embodiment of the application provides a kind of based on the depth learning data sharing method and device of blockchain technology, receiving the acquisition request including the identification information of requested depth learning data and target token number that demand end equipment sends, based on identification information and the smart contract of pre-setting, determine the reference token number corresponding to identification information, if target token number is greater than or equal to reference token number, then the depth learning data with the identification information of pre-stored is sent to demand end equipment, to make demand end equipment utilize received depth learning data and carry out model training.If target token number is greater than or equal to reference token number, it indicates that the token number provided when demand end equipment requests depth learning data satisfies the token number requirement set when publishing end equipment publishes the depth learning data, at this time, the depth learning data of its request is only sent to demand end equipment, to improve the data security when depth learning data is shared.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of deep learning, in particular to a deep learning data sharing method and device based on blockchain technology. BACKGROUND

[0002] A deep learning model is an intelligent model that analyzes and learns by establishing and simulating the human brain, and has been well applied in target detection and segmentation, behavior detection and recognition, speech recognition, etc. The deep learning model is trained using a large amount of sample data, and therefore, a device running the deep learning model is required to pre-store a large amount of sample data to ensure the accuracy of the deep learning model.

[0003] With the development of computer networks, multiple devices will be deployed in the same local area network. According to the traditional training requirements of the deep learning model, a large amount of sample data needs to be pre-stored on each device, resulting in a waste of a large amount of storage resources.

[0004] In order to solve the above problems, the corresponding method uses blockchain technology to share deep learning data among devices. Blockchain technology is a distributed database technology that can maintain continuously growing and tamper-proof data records by maintaining a chain structure of data blocks. As shown in Figure 1 Each device uploads deep learning data such as sample data used locally and deep learning models trained to the blockchain network. When any device needs to use deep learning data of other devices, it can obtain the requested deep learning data by sending a request to the blockchain network, thereby achieving sharing of deep learning data.

[0005] However, in the current deep learning data sharing method based on blockchain technology, any device can obtain deep learning data by sending a request when it has a demand, resulting in poor data security. SUMMARY

[0006] The purpose of the embodiments of the present application is to provide a deep learning data sharing method and device based on blockchain technology to improve the data security when sharing deep learning data. The specific technical solutions are as follows:

[0007] In a first aspect, the embodiments of the present application provide a deep learning data sharing method based on blockchain technology, applied to a blockchain network, which includes:

[0008] receiving an acquisition request sent by a demand end device, wherein the acquisition request includes identification information of the requested deep learning data and a target number of tokens;

[0009] Determine the reference token number corresponding to the identification information based on the identification information and a pre-set smart contract, wherein the smart contract is set by the publishing end device when publishing the deep learning data to the blockchain network, and the smart contract sets the corresponding relationship between the identification information of the deep learning data and the reference token number;

[0010] If the target token number is greater than or equal to the reference token number, send the pre-stored deep learning data with the identification information to the demand end device, so that the demand end device uses the deep learning data for model training.

[0011] Optionally, the smart contract also sets the corresponding relationship between the identification information of the deep learning data and the data key;

[0012] The method further comprises:

[0013] Determine the data key corresponding to the identification information based on the identification information and the smart contract;

[0014] If the target token number is greater than or equal to the reference token number, send the data key to the demand end device, so that the demand end device uses the data key to decrypt the deep learning data, and performs model training based on the decrypted deep learning data.

[0015] Optionally, the deep learning data includes sample data and / or a deep learning model;

[0016] The step of sending the pre-stored deep learning data with the identification information to the demand end device comprises:

[0017] Send the pre-stored sample data with the identification information to the demand end device, so that the demand end device uses the sample data to train the local deep learning model;

[0018] And / or,

[0019] Send the pre-stored deep learning model with the identification information to the demand end device, so that the demand end device uses the local sample data to train the deep learning model.

[0020] Optionally, after the step of sending the pre-stored deep learning data with the identification information to the demand end device, the method further comprises:

[0021] Obtain the trained deep learning data, the identification information of the trained deep learning data, and the reference token number sent by the demand end device, wherein the trained deep learning data includes a deep learning model that has completed training, and / or sample data used to train the deep learning model;

[0022] deploy the smart contract, and set a correspondence between the identification information and the benchmark token number in the smart contract based on the identification information and the benchmark token number.

[0023] Optionally, the method further includes:

[0024] obtaining a data key of the trained deep learning data;

[0025] setting a correspondence between the identification information and the data key in the smart contract.

[0026] Optionally, the method further includes:

[0027] If the target token number is greater than or equal to the benchmark token number, the tokens of the target token number are sent to the publishing end device.

[0028] In a second aspect, the embodiments of the present application provide a deep learning data sharing device based on a blockchain technology, applied to a blockchain network, the device comprising:

[0029] a receiving module configured to receive an obtaining request sent by a demand end device, wherein the obtaining request comprises identification information of requested deep learning data and a target token number;

[0030] a searching module configured to determine a benchmark token number corresponding to the identification information based on the identification information and a pre-set smart contract, wherein the smart contract is set by a publishing end device when publishing deep learning data to the blockchain network, and the smart contract sets a correspondence between identification information of deep learning data and a benchmark token number;

[0031] a sending module configured to send pre-stored deep learning data with the identification information to the demand end device if the target token number is greater than or equal to the benchmark token number, so that the demand end device uses the deep learning data for model training.

[0032] Optionally, the smart contract further sets a correspondence between identification information of deep learning data and a data key;

[0033] the searching module is further configured to determine a data key corresponding to the identification information based on the identification information and the smart contract;

[0034] the sending module is further configured to send the data key to the demand end device if the target token number is greater than or equal to the benchmark token number, so that the demand end device uses the data key to decrypt the deep learning data, and performs model training based on the decrypted deep learning data.

[0035] Optionally, the device further comprises:

[0036] The acquisition module is configured to acquire the trained deep learning data, identification information of the trained deep learning data, and a reference token number sent by the demand-side device, wherein the trained deep learning data comprises a deep learning model that has completed training and / or sample data used for training the deep learning model;

[0037] The deployment module is configured to deploy the smart contract and set a correspondence between the identification information and the reference token number in the smart contract based on the identification information and the reference token number.

[0038] Optionally, the acquisition module is further configured to acquire a data key of the trained deep learning data.

[0039] The deployment module is further configured to set a correspondence between the identification information and the data key in the smart contract.

[0040] The embodiments of the present application have the following beneficial effects:

[0041] The embodiments of the present application provide a deep learning data sharing method and device based on a blockchain technology, which are applied to a blockchain network, and the method comprises the following steps: receiving an acquisition request sent by a demand-side device, the acquisition request comprising identification information of requested deep learning data and a target token number; determining a reference token number corresponding to the identification information based on the identification information and a pre-set smart contract; if the target token number is greater than or equal to the reference token number, sending pre-stored deep learning data with the identification information to the demand-side device, so that the demand-side device performs model training by using the received deep learning data. When a publishing-side device publishes deep learning data, the smart contract is set on the blockchain network, and a correspondence between identification information of the deep learning data and a reference token number is set in the smart contract. After receiving the acquisition request, the reference token number corresponding to the identification information of the requested deep learning data is determined according to the smart contract, and the target token number sent by the demand-side device and the reference token number are compared. If the target token number is greater than or equal to the reference token number, it indicates that the token number provided by the demand-side device when requesting the deep learning data meets the token number requirement set by the publishing-side device when publishing the deep learning data. Therefore, the blockchain network sends the deep learning data to the demand-side device. That is, the blockchain network does not send the deep learning data requested by the demand-side device to the demand-side device as soon as the acquisition request is received, but sends the deep learning data requested by the demand-side device to the demand-side device only after determining that the target token number is greater than or equal to the reference token number, that is, when the token number provided by the demand-side device when requesting the deep learning data meets the token number requirement set by the publishing-side device when publishing the deep learning data. Therefore, the data security during deep learning data sharing is improved. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced. Obviously, the accompanying drawings in the following description only represent some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0043] Figure 1 A flowchart of a prior art deep learning data sharing method based on blockchain technology;

[0044] Figure 2 A flowchart of a deep learning data sharing method based on blockchain technology according to an embodiment of the present application;

[0045] Figure 3 A flowchart of a deep learning data sharing method based on blockchain technology according to another embodiment of the present application;

[0046] Figure 4 A flowchart of a deep learning data sharing method based on blockchain technology according to still another embodiment of the present application;

[0047] Figure 5 A flowchart of a deep learning data sharing method based on blockchain technology according to still another embodiment of the present application;

[0048] Figure 6 A structural diagram of a deep learning data sharing device based on blockchain technology according to an embodiment of the present application;

[0049] Figure 7 A structural diagram of a blockchain network according to an embodiment of the present application. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments only represent some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0051] In order to improve the data security during deep learning data sharing, the present application provides a deep learning data sharing method, device, blockchain network and machine readable storage medium based on blockchain technology.

[0052] Next, a deep learning data sharing method based on blockchain technology provided by the present application will be introduced.

[0053] The execution subject of the deep learning data sharing method based on the blockchain technology provided in the embodiments of the present application is a blockchain network, the blockchain network is an open platform, the open platform refers to that a software system enables external programs to increase the functions of the software system or use the resources of the software system by publicly providing application programming interfaces (APIs) or functions, therefore, the blockchain network can be a virtual service platform, or can be an entity server or device. The deep learning data sharing method based on the blockchain technology provided in the embodiments of the present application can be implemented in at least one of the following manners: software, hardware circuit and logic circuit.

[0054] As shown in Figure 2 The deep learning data sharing method based on the blockchain technology provided in the embodiments of the present application can include the following steps.

[0055] S201, receiving an acquisition request sent by a demand end device, wherein the acquisition request includes identification information of requested deep learning data and a target token number.

[0056] S202, determining a reference token number corresponding to the identification information based on the received identification information and a pre-set smart contract, wherein the smart contract is set by a publishing end device when publishing deep learning data to the blockchain network, and the smart contract sets a corresponding relationship between the identification information of the deep learning data and the reference token number.

[0057] S203, if the target token number is greater than or equal to the reference token number, sending pre-stored deep learning data with the identification information to the demand end device, so that the demand end device uses the deep learning data for model training.

[0058] With the embodiments of the present application, when the publishing end device publishes deep learning data, a smart contract is set on the blockchain network, and the correspondence between the identification information of the deep learning data and the reference token number is set in the smart contract. Then, after receiving the acquisition request, the reference token number corresponding to the identification information of the requested deep learning data is determined according to the smart contract, and the target token number and the reference token number provided by the demand end device are compared. If the target token number is greater than or equal to the reference token number, it means that the token number provided by the demand end device when requesting the deep learning data meets the token number requirement set by the publishing end device when publishing the deep learning data, and then the blockchain network sends the deep learning data to the demand end device. That is, the blockchain network does not send the deep learning data requested by the demand end device to the demand end device as soon as it receives the acquisition request, but sends the deep learning data requested by the demand end device to the demand end device only after determining that the target token number is greater than or equal to the reference token number, that is, when the token number provided by the demand end device when requesting the deep learning data meets the token number requirement set by the publishing end device when publishing the deep learning data, the deep learning data requested by the demand end device is sent to the demand end device, thereby improving the data security when sharing deep learning data. And no matter it is public data or private data, through certain incentive measures (agreed reference token number), the maximum sharing of resources is realized. Due to the non-modifiability of the blockchain, the correctness of the deep learning data stored on the blockchain is ensured. Due to the sharing of deep learning data, each device can obtain deep learning data on demand, thereby meeting the local model training requirements, having stronger applicability, and saving the storage resources of the device.

[0059] A plurality of devices are connected together according to the blockchain technology, each device can obtain certain sample data, deep learning model and other deep learning data by itself, and can upload the local deep learning data to the blockchain network. Each device can know which deep learning data is stored on the blockchain network. When a device has a model training requirement, it can send an acquisition request to the blockchain network based on the actual requirement, and the acquisition request includes the identification information of the requested deep learning data and the target token number. The identification information can be the storage location, name, number, etc. of the deep learning data in the blockchain network, which uniquely represents the identification of a deep learning data. Token refers to digital currency representing data value, and the higher the token number, the higher the value of the corresponding deep learning data. The target token number refers to the number of tokens provided by the demand end device for the requested deep learning data.

[0060] The publishing end device sets a corresponding smart contract when publishing the deep learning data to the blockchain network. The smart contract is based on the computer protocol provided by the blockchain system, is a set of promises defined in digital form, and includes protocols that contract participants can use to execute these promises, and implements protocols related to token management. The smart contract has the characteristics of being unchangeable, automated, and intelligent. The smart contract sets a corresponding relationship between the identification information of the deep learning data and the benchmark token number, so based on the received identification information and the pre-set smart contract, the benchmark token number corresponding to the identification information can be determined. Generally, one deep learning data corresponds to one benchmark token number, but of course, multiple deep learning data can also correspond to one benchmark token number.

[0061] After determining the benchmark token number of the deep learning data requested by the demand end device, the received target token number and the benchmark token number can be compared. If the target token number is greater than or equal to the benchmark token number, it means that the token number provided by the demand end device when requesting the deep learning data meets the token number requirement set by the publishing end device when publishing the deep learning data. At this time, the blockchain network will send the deep learning data to the demand end device. After receiving the deep learning data, the demand end device can use the deep learning data to train the model and complete the model training requirement. Correspondingly, if the target token number is less than the benchmark token number, it means that the token number provided by the demand end device when requesting the deep learning data does not meet the token number requirement set by the publishing end device when publishing the deep learning data. Therefore, the blockchain network will not send the deep learning data to the demand end device.

[0062] Optionally, the deep learning data includes sample data and / or a deep learning model.

[0063] Correspondingly, the step of sending the pre-stored deep learning data with the identification information to the demand end device in S203 can be specifically:

[0064] The pre-stored sample data with the identification information is sent to the demand end device, so that the demand end device trains the local deep learning model using the received sample data.

[0065] And / or,

[0066] The pre-stored deep learning model with the identification information is sent to the demand end device, so that the demand end device trains the received deep learning model using the local sample data.

[0067] The deep learning data in the embodiments of the present application mainly refers to sample data and / or a deep learning model. According to the actual needs of the demand-side device, if the demand-side device has a deep learning model but no sample data or only a small amount of sample data, the demand-side device can request the sample data from the blockchain network to obtain a more accurate deep learning model. If the demand-side device only has a small amount of sample data, the demand-side device can request the deep learning model of other devices from the blockchain network, and train the received deep learning model using the local sample data. We generally call this training method as incremental learning, which refers to a training method of using a small amount of sample data or specific sample data to obtain an optimal model based on the original model.

[0068] Optionally, the method can further include: if the target token number is greater than or equal to the reference token number, sending the tokens of the target token number to the publishing-side device.

[0069] Since the token is a digital currency representing the value of data, after the blockchain network obtains the tokens of the target token number, if it is determined that the target token number is greater than or equal to the reference token number, in addition to sending the deep learning data to the demand-side device, the tokens of the target token number also need to be sent to the publishing-side device, so that the publishing-side device can obtain the digital currency matching the value of the deep learning data, and achieve the purpose of data transaction.

[0070] Based on Figure 2 As shown in the embodiments, the present application also provides a deep learning data sharing method based on blockchain technology, which can include the following steps. Figure 3 As shown in the embodiments, the present application also provides a deep learning data sharing method based on blockchain technology, which can include the following steps.

[0071] S301, receiving an acquisition request sent by a demand-side device, wherein the acquisition request includes identification information of requested deep learning data and a target token number.

[0072] S302, determining a reference token number and a data key corresponding to the identification information based on the received identification information and a pre-set smart contract, wherein the smart contract is set by a publishing-side device when publishing deep learning data to the blockchain network, and the smart contract sets a corresponding relationship between the identification information of the deep learning data and the reference token number and the data key.

[0073] S303, if the target token number is greater than or equal to the reference token number, sending the pre-stored deep learning data with the identification information and the data key corresponding to the identification information to the demand-side device, so that the demand-side device decrypts the deep learning data using the data key, and trains the model based on the decrypted deep learning data.

[0074] The smart contract can also set the correspondence between the identification information of the deep learning data and the data key. Therefore, based on the identification information and the smart contract, the data key corresponding to the received identification information can also be determined. Generally, the deep learning data of each device is private data of the device. In order to ensure the privacy of the private data, the deep learning data is encrypted when uploaded. The encryption process can be performed by the publishing end device before publishing the deep learning data, or the deep learning data can be sent to the blockchain network for processing by the blockchain network. Each deep learning data corresponds to a data key. The correspondence between the identification information of the deep learning data and the data key is set in the smart contract.

[0075] If the target token number is greater than or equal to the reference token number, the requested deep learning data and the corresponding data key are sent to the demand end device. After receiving the deep learning data and the data key, the demand end device can decrypt the deep learning data using the data key, and can use the decrypted deep learning data to perform model training to complete the model training requirement. If the demand end device does not receive the data key, it cannot decrypt the deep learning data, and thus cannot perform subsequent model training process, further ensuring the privacy and security of the data, and avoiding illegal devices from obtaining private deep learning data and causing data leakage.

[0076] Based on Figure 2 As shown in the embodiment, the application also provides a deep learning data sharing method based on blockchain technology, which can include the following steps. Figure 4 As shown in the embodiment, the application also provides a deep learning data sharing method based on blockchain technology, which can include the following steps.

[0077] S401, receiving an acquisition request sent by a demand end device, wherein the acquisition request includes identification information of requested deep learning data and a target token number.

[0078] S402, determining a reference token number corresponding to the identification information based on the received identification information and a pre-set smart contract, wherein the smart contract is set by a publishing end device when publishing deep learning data to a blockchain network, and the smart contract sets the correspondence between the identification information of the deep learning data and the reference token number.

[0079] S403, if the target token number is greater than or equal to the reference token number, sending the pre-stored deep learning data with the identification information to the demand end device, so that the demand end device uses the deep learning data to perform model training.

[0080] S404, obtaining the trained deep learning data sent by the demand-side device, the identification information of the trained deep learning data, and the reference token number, wherein the trained deep learning data comprises a deep learning model that has completed training, and / or sample data used for training the deep learning model.

[0081] S405, deploying the smart contract, and setting the correspondence between the identification information and the reference token number in the smart contract based on the obtained identification information and reference token number.

[0082] After receiving the deep learning data, the demand-side device trains the model using the deep learning data, completes the model training, and uploads the trained deep learning data to the blockchain network after completing the model training. The trained deep learning data mentioned herein refers to a deep learning model that has completed training and / or sample data used for training the deep learning model.

[0083] The demand-side device can also deploy a smart contract when uploading the trained deep learning data to the blockchain network, and the smart contract needs to set the correspondence between the identification information of the trained deep learning data and the reference token number. The reference token number of the trained deep learning data can be set by the demand-side device itself, or can be set by the blockchain network according to the mathematical characteristics of the deep learning data.

[0084] Of course, the blockchain network can also obtain the data key of the trained deep learning data, and accordingly, set the correspondence between the identification information and the data key in the smart contract.

[0085] In order to ensure the privacy and security of private data, the trained deep learning data uploaded by the demand-side device can be encrypted data, and therefore, the demand-side device uploads the data key when uploading the trained deep learning data, so that the correspondence between the identification information and the data key can be set in the smart contract.

[0086] Based on Figures 2-4 As shown in the embodiment, the overall process of the deep learning data sharing method provided by the embodiment of the present application is as shown in the figure. Figure 5 As shown in the embodiment, the overall process of the deep learning data sharing method provided by the embodiment of the present application is as shown in the figure. As shown in the embodiment, the overall process of the deep learning data sharing method provided by the embodiment of the present application is as shown in the figure.

[0087] Corresponding to the above method embodiments, the embodiments of the present application provide a deep learning data sharing device based on blockchain technology, applied to a blockchain network, as shown in the figure, the device can include: Figure 6

[0088] The receiving module 610 is configured to receive an acquisition request sent by a demand-side device, wherein the acquisition request includes identification information of requested deep learning data and a target token number;

[0089] The searching module 620 is configured to determine a reference token number corresponding to the identification information based on the identification information and a pre-set smart contract, wherein the smart contract is set by a publishing-side device when publishing deep learning data to the blockchain network, and the smart contract sets a corresponding relationship between the identification information of the deep learning data and the reference token number;

[0090] The sending module 630 is configured to send the pre-stored deep learning data with the identification information to the demand-side device if the target token number is greater than or equal to the reference token number, so that the demand-side device performs model training using the deep learning data.

[0091] Optionally, the smart contract also sets a corresponding relationship between the identification information of the deep learning data and a data key;

[0092] The searching module 620 can also be configured to determine a data key corresponding to the identification information based on the identification information and the smart contract;

[0093] The sending module 630 can also be configured to send the data key to the demand-side device if the target token number is greater than or equal to the reference token number, so that the demand-side device decrypts the deep learning data using the data key and performs model training based on the decrypted deep learning data.

[0094] Optionally, the deep learning data includes sample data and / or a deep learning model;

[0095] The sending module 630 can be specifically configured to send pre-stored sample data with the identification information to the demand-side device, so that the demand-side device trains a local deep learning model using the sample data; and / or send pre-stored deep learning models with the identification information to the demand-side device, so that the demand-side device trains the deep learning model using local sample data.

[0096] Optionally, the device can also include:

[0097] ​The acquisition module is configured to acquire the trained deep learning data sent by the demand-side device, identification information of the trained deep learning data, and a reference token number, wherein the trained deep learning data comprises a deep learning model that has completed training, and / or sample data used for training the deep learning model;

[0098] The deployment module is configured to deploy the smart contract, and set a correspondence between the identification information and the reference token number in the smart contract based on the identification information and the reference token number.

[0099] Optionally, the acquisition module is further configured to acquire a data key of the trained deep learning data.

[0100] The deployment module is further configured to set a correspondence between the identification information and the data key in the smart contract.

[0101] Optionally, the sending module 630 is further configured to send the target token number of tokens to the publishing-side device if the target token number is greater than or equal to the reference token number.

[0102] According to the embodiments of the present application, when the publishing-side device publishes deep learning data, a smart contract is set on the blockchain network, and a correspondence between identification information of the deep learning data and a reference token number is set in the smart contract. After receiving an acquisition request, the reference token number corresponding to the identification information of the requested deep learning data is determined according to the smart contract, and the target token number sent by the demand-side device is compared with the reference token number. If the target token number is greater than or equal to the reference token number, it indicates that the token number provided by the demand-side device when requesting the deep learning data meets the token number requirement set by the publishing-side device when publishing the deep learning data. Therefore, the blockchain network sends the deep learning data to the demand-side device. That is, the blockchain network does not send the deep learning data requested by the demand-side device to the demand-side device as soon as the acquisition request is received, but sends the deep learning data requested by the demand-side device to the demand-side device only after determining that the target token number is greater than or equal to the reference token number, i.e., when the token number provided by the demand-side device when requesting the deep learning data meets the token number requirement set by the publishing-side device when publishing the deep learning data. Therefore, the data security during sharing of the deep learning data is improved.

[0103] The embodiments of the present application further provide a blockchain network, as shown in Figure 7 The processor 701 and the machine readable storage medium 702 are configured to perform the following steps:

[0104] The machine readable storage medium 702 is configured to store machine executable instructions that can be executed by the processor 701.

[0105] The processor 701 is configured to execute all the steps of the method for sharing deep learning data based on the blockchain technology provided by the embodiments of the present application by machine-executable instructions stored in the machine-readable storage medium 702.

[0106] The machine-readable storage medium 702 and the processor 701 can transmit data through wired connection or wireless connection, and the electronic device can communicate with other devices through wired communication interface or wireless communication interface.

[0107] The machine-readable storage medium can include a RAM (Random Access Memory) and a NVM (Non-volatile Memory), for example, at least one disk memory. Optionally, the machine-readable storage medium can also be at least one storage device located away from the aforementioned processor.

[0108] The processor can be a general-purpose processor, including a CPU (Central Processing Unit), an NP (Network Processor), etc. The processor can also be a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0109] In this embodiment, the processor of the electronic device can realize the following by reading the machine-executable instructions stored in the machine-readable storage medium and running the machine-executable instructions: when the publishing end device publishes the deep learning data, the smart contract is set on the blockchain network, and the corresponding relationship between the identification information of the deep learning data and the reference token number is set in the smart contract; after receiving the acquisition request, the corresponding reference token number of the identification information of the requested deep learning data is determined according to the smart contract, the target token number and the reference token number sent by the demand end device are compared, and if the target token number is greater than or equal to the reference token number, it indicates that the token number provided by the demand end device when requesting the deep learning data meets the token number requirement set by the publishing end device when publishing the deep learning data, and then the blockchain network sends the deep learning data to the demand end device. That is, the blockchain network will not send the deep learning data requested by the demand end device to the demand end device as soon as the acquisition request is received, but will send the deep learning data requested by the demand end device to the demand end device after determining that the target token number is greater than or equal to the reference token number, that is, when the token number provided by the demand end device when requesting the deep learning data meets the token number requirement set by the publishing end device when publishing the deep learning data, the deep learning data requested by the demand end device will be sent to the demand end device, thereby improving the data security when sharing the deep learning data.

[0110] In addition, corresponding to the deep learning data sharing method based on the blockchain technology provided in the above embodiment, the embodiment of the present application provides a machine-readable storage medium for machine-executable instructions, which cause the processor to perform all steps of the deep learning data sharing method based on the blockchain technology provided in the embodiment of the present application.

[0111] In this embodiment, the machine readable storage medium stores machine executable instructions for performing the deep learning data sharing based on the blockchain technology at runtime, so as to achieve that when the publishing end device publishes the deep learning data, the smart contract is set on the blockchain network, and the correspondence between the identification information of the deep learning data and the reference token number is set in the smart contract, then after receiving the acquisition request, the reference token number corresponding to the identification information of the requested deep learning data is determined according to the smart contract, the target token number and the reference token number sent by the demand end device are compared, if the target token number is greater than or equal to the reference token number, it means that the token number provided by the demand end device when requesting the deep learning data meets the token number requirement set by the publishing end device when publishing the deep learning data, then the blockchain network will send the deep learning data to the demand end device. That is, the blockchain network will not send the deep learning data requested by the demand end device to the demand end device as soon as it receives the acquisition request, but will send the deep learning data requested by the demand end device to the demand end device after determining that the target token number is greater than or equal to the reference token number, that is, when the token number provided by the demand end device when requesting the deep learning data meets the token number requirement set by the publishing end device when publishing the deep learning data, the deep learning data requested by the demand end device will be sent to the demand end device, thereby improving the data security when sharing the deep learning data.

[0112] For the blockchain network and the machine readable storage medium embodiment, the method content involved is basically similar to the foregoing method embodiment, so the description is relatively simple, and the related parts are described in the method embodiment.

[0113] It should be noted that in this paper, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment. Without more limitations, the element defined by the statement "including a…" does not exclude the presence of other identical elements in the process, method, article or equipment including the element.

[0114] Each of the embodiments in the specification is described in a relevant manner, and the same or similar parts between the embodiments can be referred to each other. Each of the embodiments focuses on the difference from other embodiments. In particular, for the device, blockchain network, and machine readable storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.

[0115] The above only describes the preferred embodiments of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A deep learning data sharing method based on blockchain technology, characterized in that, The deep learning data applied to the blockchain network includes sample data and / or a deep learning model, and the deep learning data is private data, and the method comprises: receiving an acquisition request sent by a demand-side device, wherein the acquisition request comprises identification information of requested deep learning data and a target token number; determining a reference token number corresponding to the identification information based on the identification information and a pre-set smart contract, wherein the smart contract is set by a publishing-side device when publishing deep learning data to the blockchain network, and the smart contract sets a corresponding relationship between identification information of deep learning data and a reference token number; if the target token number is greater than or equal to the reference token number, sending pre-stored sample data with the identification information to the demand-side device to enable the demand-side device to train a local deep learning model using the sample data, and / or sending pre-stored deep learning model with the identification information to the demand-side device to enable the demand-side device to train the deep learning model using local sample data, and sending a token of the target token number to the publishing-side device to achieve the purpose of data transaction; acquiring trained deep learning data sent by the demand-side device, identification information of the trained deep learning data, and a reference token number, wherein the trained deep learning data comprises a deep learning model that has completed training and / or sample data used for training the deep learning model; deploying a smart contract and setting a corresponding relationship between the identification information and the reference token number in the smart contract based on the acquired identification information and reference token number.

2. The method of claim 1, wherein, The smart contract also sets a corresponding relationship between identification information of deep learning data and a data key; The method further comprises: determining a data key corresponding to the identification information based on the identification information and the smart contract; if the target token number is greater than or equal to the reference token number, sending the data key to the demand-side device to enable the demand-side device to decrypt the deep learning data using the data key and perform model training based on the decrypted deep learning data.

3. The method of claim 1, wherein, The method further comprises: acquiring a data key of the trained deep learning data; setting a corresponding relationship between the identification information and the data key in the smart contract. 4.A deep learning data sharing device based on a blockchain technology, characterized by comprising: The deep learning data applied to the blockchain network includes sample data and / or a deep learning model, and the deep learning data is private data, and the device comprises: a receiving module configured to receive an acquisition request sent by a demand-side device, wherein the acquisition request comprises identification information of requested deep learning data and a target token number; a searching module configured to determine a reference token number corresponding to the identification information based on the identification information and a pre-set smart contract, wherein the smart contract is set by a publishing-side device when publishing deep learning data to the blockchain network, and the smart contract sets a corresponding relationship between identification information of deep learning data and a reference token number; The sending module is configured to, if the target token number is greater than or equal to the reference token number, send pre-stored sample data with the identification information to the demand-side device, so that the demand-side device trains a local deep learning model by using the sample data, and / or send a pre-stored deep learning model with the identification information to the demand-side device, so that the demand-side device trains the deep learning model by using local sample data, and send tokens of the target token number to the publishing-side device, so as to achieve the purpose of data transaction. The obtaining module is configured to obtain trained deep learning data sent by the demand-side device, identification information of the trained deep learning data, and a reference token number, wherein the trained deep learning data includes a deep learning model that has completed training and / or sample data used for training the deep learning model. The deploying module is configured to deploy a smart contract, and set a corresponding relationship between the identification information and the reference token number in the smart contract based on the obtained identification information and reference token number.

5. The apparatus of claim 4, wherein, The smart contract further sets a corresponding relationship between identification information of deep learning data and a data key. The finding module is further configured to determine the data key corresponding to the identification information based on the identification information and the smart contract. The sending module is further configured to, if the target token number is greater than or equal to the reference token number, send the data key to the demand-side device, so that the demand-side device decrypts the deep learning data by using the data key, and trains a model based on the decrypted deep learning data.

6. The apparatus of claim 4, wherein, The obtaining module is further configured to obtain a data key of the trained deep learning data. The deploying module is further configured to set a corresponding relationship between the identification information and the data key in the smart contract.

Images