KR20200022273A - Method for performing mining in parallel with machine learning and method for supproting the mining, in a distributed computing resource shring system based on block chain - Google Patents

Abstract

Provided is a mining method performed in nodes on a distributed computing resource sharing system based on a block chain, wherein a distributed computing resource sharing system includes a server and a plurality of nodes. The method of the present disclosure comprises the steps of: receiving an instruction to perform a predetermined operation; notifying a server of completion of the operation according to the received instruction; receiving a predetermined nonce value from the server in response to the notification; applying the nonce value to a predetermined hash function; determining whether a hash result obtained by applying the nonce value to the predetermined hash function satisfies a predetermined criterion; and notifying the nonce value and the hash result to a block chain when it is determined that the hash result satisfies the predetermined criterion.

Description

METHODO FOR PERFORMING MINING IN PARALLEL WITH MACHINE LEARNING AND METHOD FOR SUPPROTING THE MINING, IN A DISTRIBUTED COMPUTING RESOURCE SHRING SYSTEM BASED ON BLOCK CHAIN}

The present disclosure relates to a distributed computing resource sharing system, and more specifically, based on the blockchain, the blockchain on a distributed computing resource sharing system configured to share distributed computing resources that can be used for machine learning. The present invention relates to a method of performing mining and a method of supporting such a mining.

Recently, there has been a growing interest in artificial intelligence research based on machine learning. The expectation is that AI based on machine learning will develop the future of mankind to a different level than before. The AI study based on such machine learning basically consists of three elements: hardware resources that can support a large amount of computation, learning data that can be used for machine learning, and machine learning models (algorithms). By the way, some big companies, including Facebook, Amazon, Google, Microsoft, etc., monopolize each element of the above artificial intelligence research in their own way, and individual researchers or small companies in relatively poor environment They are deprived of the opportunity to participate in AI research, which raises concern that AI and the resulting developmental by-products will be private property of these giants.

On the other hand, since Satoshi Nakamoto introduced the blockchain technology to the world through cryptocurrency bitcoin in 2009, interest in various forms of cryptocurrencies and the blockchain technology that supports them is getting hotter. Blockchain technology is being investigated in various fields in that it enables transactions based on a decentralized distributed shared ledger scheme and provides high openness, reliability, transparency, and interactivity. In particular, new blockchain technologies such as Bitcoin and Ethereum support smart contracts, that is, digital contracts that specify contract terms on the blockchain and automatically execute the contract contents when the conditions are met. As a result, the application of such blockchain technology is being attempted in a wider variety of industries.

By the way, many existing cryptocurrencies, including Bitcoin, have proven that they have completed the task of storing transaction records in blocks, encrypting them, and adding them to the existing blockchain (shared ledger). It is operated / maintained through the method of paying coins (ie Proof of Work (PoW)). The process of adding a new block requires significant computation, which requires a large amount of power consumption to drive the hardware to perform this computation. As a large number of hardware resources compete competitively in mining, each time a mining failure occurs, a large amount of power is wasted. In recent years, the problem of power waste has been found to be a problem for the cryptocurrency system based on the existing PoW method. It is pointed out as one.

Patent Literature: Korean Patent Registration No. 10-1852079

Build resources for artificial intelligence research based on machine learning, such as hardware resources for computation and storage, high quality learning data to be used for machine learning, and efficient sharing network of various machine learning models (algorithms) and provide sharing of these resources. And to provide a platform that can support an objective and reasonable compensation system for use. In particular, with respect to mining for blockchain, it is to provide a mining technology that can solve the problems such as resources and electricity waste caused by the mining according to the existing PoW method and to use limited resources more efficiently.

According to one aspect of the present disclosure, there is provided a mining method performed at a node on a blockchain-based distributed computing resource sharing system, including a server and a plurality of nodes. The method of the present disclosure includes receiving an instruction to perform a predetermined operation; Notifying the server of completion of the operation according to the received instruction; In response to the notification, receiving a predetermined nonce value from the server; Applying a nonce value to a predetermined hash function; Determining whether a hash result obtained by applying a nonce value to a predetermined hash function satisfies a predetermined criterion; And if it is determined that the hash result satisfies a predetermined criterion, notifying the nonce value and the hash result on the blockchain.

According to one embodiment of the present disclosure, the operation may include an operation of a predetermined number of executions of a predetermined machine learning code.

According to an embodiment of the present disclosure, a node may include receiving a nonce value and a hash result from another node of a plurality of nodes; And verifying the nonce value and the hash result received from the other node by applying the hash function.

According to one embodiment of the present disclosure, the predetermined criterion may be determined so that new blocks may be generated at predetermined time intervals on the blockchain.

According to another aspect of the present disclosure, there is provided a mining support method performed by a server on a blockchain-based distributed computing resource sharing system including a server and a plurality of nodes. The method of the present disclosure includes receiving, from the outside, a request for one or more tasks, each of the one or more tasks including a predetermined number of executions of a predetermined operation; Allocating execution of operations according to one or more tasks on the plurality of nodes; Receiving, from any one of the plurality of nodes, a notification indicating completion of execution of assigned operations for one node; And in response to receiving the notification, sending a nonce value to one node, wherein the nonce value is one of a plurality of values generated on the server and used by one node as input to the hash function. It includes.

According to one embodiment of the present disclosure, allocating execution of operations on the plurality of nodes may include allocating such that the time taken to execute the operations on the plurality of nodes is equalized.

According to one embodiment of the present disclosure, each of the operations may include execution of predetermined machine learning code.

According to an embodiment of the present disclosure, when the nonce value is used as an input of a hash function on one node, a hash result obtained from the hash function corresponding thereto may satisfy a predetermined criterion.

According to one embodiment of the present disclosure, the predetermined criterion may be determined so that new blocks may be generated at predetermined time intervals on the blockchain.

According to yet another aspect of the present disclosure, a computer readable recording medium having recorded thereon instructions executable by a computer, wherein the instructions, when executed by the computer, cause the computer to perform any one of the methods described above. A computer readable recording medium is provided.

According to embodiments of the present disclosure, resources for artificial intelligence based on machine learning, such as hardware resources for computation, high quality learning data to be used for machine learning, and efficient sharing of various machine learning models (algorithms) and The provision of objective and reasonable compensation for the provision of resources may be made.

Individual hardware resources (eg, processing devices such as GPUs and various storage space devices) are not always 100% utilized, and a large number of hardware resources are idle for many hours. According to embodiments of the present disclosure, an idle hardware resource held by an individual may be utilized in a useful form for the valuable purpose of contributing to the development of artificial intelligence technology. In addition, according to embodiments of the present disclosure, shared use of software resources developed and / or possessed by individuals, such as data for machine learning and various learning models, is actively performed, and each creator or holder is a resource. The right to profit from a shared share of the project will be available, and researchers will be able to use the resources for new tasks instead of consuming duplicates for the same task, ultimately accelerating AI development.

According to embodiments of the present disclosure, the provision and utilization of various resources on the platform and the payment thereof can be made based on blockchain technology, thereby enabling open, transparent and reliable transactions, and voluntary participation of members. And encourage active participation.

According to embodiments of the present disclosure, in connection with mining for a blockchain, each node has an opportunity to obtain a nonce value on the completion of a predetermined machine learning operation while correspondingly having a difficulty level of a hash function. By lowering the mining, a mining technique that can more efficiently use the limited hardware and / or energy resources without causing problems such as resources and electricity waste caused by the mining according to the existing PoW method can be provided.

1 is a diagram conceptually illustrating an overall configuration of a blockchain-based distributed computing resource sharing system 100 supporting smart contracts according to an embodiment of the present disclosure.
FIG. 2 is a functional block diagram schematically illustrating a functional configuration of the resource sharing node 130 of FIG. 1, according to an embodiment of the present disclosure.
3 is a blockchain and each block 300 included therein, which may be stored in the blockchain distributed ledger storage module 208 included in the resource sharing node 130 of FIG. 2, according to an embodiment of the present disclosure. It is a diagram conceptually showing the configuration of.
4 is an example of transaction information 304 that may be included in block 300 of FIG. 3, in accordance with an embodiment of the present disclosure, for providing shared resources on a distributed computing resource sharing system 100. Illustrates transaction information by way of example.
5 exemplarily illustrates a query for searching for a predetermined shared resource on the distributed computing resource sharing system 100, generated by the resource sharing node 130 of FIG. 1, according to one embodiment of the disclosure. The figure is shown.
FIG. 6 is an example of transaction information 304 included in block 300 of FIG. 3, in accordance with an embodiment of the present disclosure, for requesting the use of shared resources on distributed computing resource sharing system 100. Illustrates transaction information by way of example.
7 is a diagram schematically illustrating a functional configuration of the platform server 140 of FIG. 1, according to an embodiment of the present disclosure.
FIG. 8 schematically illustrates an example of a process of providing shared resources onto the distributed computing resource sharing system 100 through the resource sharing node 130 of FIG. 1, according to an embodiment of the present disclosure. This is a flow chart.
9 is an operational flow diagram schematically illustrating an example of a process of requesting, providing and using a data resource, performed on the distributed computing resource sharing system 100, according to an embodiment of the present disclosure.

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following, when it is determined that there is a risk of unnecessarily obscuring the subject matter of the present disclosure, a detailed description of already known functions and configurations will be omitted. In addition, it should be understood that what is described below is only related to an embodiment of the present disclosure, but the present disclosure is not limited thereto.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present disclosure. For example, a component expressed in the singular should be understood as a concept including a plurality of components unless the context clearly indicates the singular. It is to be understood that the term "and / or" as used in this disclosure encompasses all possible combinations by one or more of the listed items. The terms "comprise" or "having" as used in the present disclosure are merely intended to designate that there exists a feature, number, step, operation, component, part, or combination thereof described on the present disclosure. It is not intended to exclude the possibility of the presence or the addition of one or more other features or numbers, steps, actions, components, parts or combinations thereof by use.

In an embodiment of the present disclosure, the module or unit refers to a functional part that performs at least one function or operation, and may be implemented in hardware or software, or in a combination of hardware and software. Also, a plurality of 'modules' or 'units' may be integrated by at least one software module and implemented by at least one processor, except for 'modules' or 'units', which need to be implemented by specific hardware. have.

In addition, all terms used in this disclosure, including technical or scientific terms, unless otherwise defined, have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It is to be understood that the terms used in the dictionary, which are generally used, should be interpreted to have a meaning consistent with the contextual meaning of the related art, and shall not be construed as being excessively limited or extended unless clearly defined otherwise in the present disclosure. .

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

1 is a diagram conceptually illustrating an overall configuration of a blockchain-based distributed computing resource sharing system 100 supporting a smart contract according to an embodiment of the present disclosure. As shown, the distributed computing resource sharing system 100 includes a blockchain network 110, a plurality of shared computing resource regions 120, a plurality of resource sharing nodes 130, and a platform server 140. do.

According to an embodiment of the present disclosure, the blockchain network 110 may include each shared computing resource region 120, each resource sharing node 130, and a platform server 140 on the distributed computing resource sharing system 100. It may be any wired or wireless communication network that enables communication between them. According to an embodiment of the present disclosure, the blockchain network 110 may include, for example, a P2P distributed network implemented on a TCP / IP communication network. According to an embodiment of the present disclosure, the blockchain network 110 may be, for example, a pure P2P network or a hybrid P2P network implemented on a Wi-Fi network, a LAN network, a WAN network, an Internet network, and the like. It is not limited. According to one embodiment of the present disclosure, each transaction created and distributed by each resource sharing node 130 through the blockchain network 110 is included in the blockchain for the distributed computing resource sharing system 100 and is thus different. It may be shared with the resource sharing nodes 130 and the platform server 140.

Although not specifically illustrated, according to an exemplary embodiment of the present disclosure, the distributed computing resource sharing system 100 may include a plurality of user accounts and resource management accounts (eg, management of use and compensation of shared computing resources). , A smart contract account of the blockchain of the distributed computing resource sharing system 100). According to one embodiment of the present disclosure, each user of the distributed resource sharing system 100 can access the blockchain network 110 through their user account (and client), respectively, and the resource sharing node (described below) By creating and / or distributing a transaction through 130, it is possible to provide its computing resources on the distributed computing resource sharing system 100 or to use the computing resources provided on the distributed computing resource sharing system 100. According to one embodiment of the present disclosure, each user account of the distributed resource sharing system 100 may share and use hardware and / or software resources owned by other users on the distributed resource sharing system 100. It may be an account for providing a resource provider. According to one embodiment of the present disclosure, each user account of the distributed resource sharing system 100 is a developer who wants to perform machine learning according to a predetermined algorithm, and is provided for shared use on the distributed resource sharing system 100. It may be an account for a developer who wants to use hardware and / or software resources at a reasonable cost. According to one embodiment of the present disclosure, each user account of the distributed resource sharing system 100, as a predetermined service developer, is provided with shared GPU resources, data sets, and learning provided on the distributed computing resource sharing system 100. It may be an account for a service developer who wants to use the result of machine learning performed according to the model for the above service. According to one embodiment of the present disclosure, each user account of the distributed resource sharing system 100 may be an account for a user who is a resource provider and a developer.

According to one embodiment of the present disclosure, each of the plurality of shared computing resource regions 120 includes one or a group of computing resources on which distributed usage may be made among a plurality of users on the distributed computing resource sharing system 100. can do. According to an embodiment of the present disclosure, the shared computing resource region 120 may be a computing resource that may be used for machine learning, for example, a hardware resource such as a processing unit unit or a storage unit such as a GPU and / or learning for machine learning. It can include software resources such as data sets or learning models. According to one embodiment of the present disclosure, each computing resource on the shared computing resource area 120 may be associated with a corresponding predetermined resource sharing node 130. According to one embodiment of the present disclosure, each computing resource of the shared computing resource area 120 is provided by one user (ie, a provider) on the distributed computing resource sharing system 100 and may be used by other users. The provision and use of each resource in the shared computing resource area 120 may include a blockchain for the distributed computing resource sharing system 100 (specifically, a smart contract and corresponding smart contract account by the blockchain). Can be operated and managed.

According to an embodiment of the present disclosure, a processing unit (eg, a GPU) among computing resources of the shared computing resource area 120 may perform a predetermined operation (eg, a predetermined operation) through a resource sharing node 130 associated with the corresponding resource. Instructions for performing a predetermined number of execution operations, etc. of the machine learning code) may be received. According to one embodiment of the present disclosure, the processing unit may perform an operation (eg, a machine learning operation) according to the received instructions above and then inform the resource sharing node 130 associated with the resource of completion of performing such an operation. Can be.

According to an embodiment of the present disclosure, a data set resource among the computing resources of the shared computing resource region 120 is divided into a plurality of portions of the data set resource and distributed to the plurality of shared computing resource regions 120. Can be stored. In this case, when the movement of data set resources is required to perform machine learning, each part of the necessary data set resources can be transmitted in parallel, thereby solving the problem of a transmission time region or a lack of bandwidth of a large amount of data movement. Can be. In addition, as described above, when a set of data set resources is divided into a plurality of parts and distributed and stored in the plurality of shared computing resource areas 120, all the shared computing resource areas 120 of the system 100 are stored. ), Each partitioned portion of the above data set resources may be stored redundantly (ie, at least two pieces of data having the same contents over the entire area). Thus, even when the shared computing resource area 120 in which a portion of the data set resource is stored is blocked, the integrity of the entire data set resource in the entire system 100 may not be impaired.

According to another embodiment of the present disclosure, a data set resource among the computing resources of the shared computing resource area 120 may be stored in one shared computing resource area 120 in its entirety. Data set resources may be stored in the corresponding shared computing resource region 120, along with a set of preprocessed data, obtained by encoding the raw data to conform to a predetermined training code for machine learning. Thus, when the movement of data set resources is required to perform machine learning, instead of data set resources as large raw data, a corresponding preprocessing data set is transmitted so that there is a problem of delay of transmission time or lack of bandwidth in data movement. Can be solved.

According to one embodiment of the present disclosure, each of the plurality of resource sharing nodes 130 may be each computing device connected to the blockchain network 110 of the distributed computing resource sharing system 100. According to one embodiment of the present disclosure, the resource sharing node 130 generates a predetermined transaction according to the blockchain protocol of ERC 20 or the like, but is not limited thereto, and generates the generated transaction and transfers the generated transaction to the blockchain network. Via 110 may be used to distribute on the blockchain for the distributed computing resource sharing system 100. According to one embodiment of the present disclosure, a transaction that may be created and distributed from each resource sharing node 130 may be provided such that certain hardware or software resources may be shared and used on distributed computing resource sharing system 100, for example. (Eg, register for shared use). According to one embodiment of the present disclosure, a transaction that may be created and distributed from each resource sharing node 130 may include, for example, any hardware provided for shared use among users on a distributed computing resource sharing system 100 or It may include a request to use software resources for machine learning. According to one embodiment of the present disclosure, each user may create a predetermined transaction (eg, a transaction for providing or using a resource) through the resource sharing node 130 using his or her user account (and client). To be distributed on the blockchain network 110.

According to one embodiment of the present disclosure, each resource sharing node 130 may receive and process each transaction generated from another resource sharing node 130 and distributed on the blockchain. According to one embodiment of the present disclosure, each resource sharing node 130 is directed to shared use of program code (eg, on hardware and / or software resources on distributed computing resource sharing system 100) on a received transaction. Program code for smart contracts). According to one embodiment of the present disclosure, each resource sharing node 130 executes program code for a smart contract on the received transaction upon receiving a transaction relating to the sharing provision of predetermined hardware and / or software resources. And store information related to the provided resource (e.g., various information including resource type, location, attributes, etc.) in association with a smart contract account (e.g., resource management account) for management of shared computing resources. According to one embodiment of the present disclosure, each resource sharing node 130, upon receiving a transaction that includes a request to use a given hardware and / or software resource, receives program code for a smart contract on the received transaction. Smart contract accounts for the management of shared computing resources, including information about usage of the resources, such as various information, including user accounts, types, locations, attributes, and usage of the resources. (E.g., resource management account).

According to one embodiment of the present disclosure, each of the plurality of resource sharing nodes 130, under the control of the platform server 130 or in accordance with a predetermined algorithm on the distributed computing resource sharing system 100, distributed computing resource sharing Blocks for recording each transaction on system 100 may be generated (or mined) and / or verified. According to one embodiment of the present disclosure, as described above, each computing resource (eg, processing unit) on the shared computing resource area 120 may be associated with a corresponding predetermined resource sharing node 130. According to an embodiment of the present disclosure, as described above, the processing unit of the computing resources of the shared computing resource area 120 may be a predetermined operation (eg, a predetermined number of execution operations of a predetermined machine learning code, etc.) from the outside. ) May receive an instruction for performing the operation, and after performing an operation (eg, a machine learning operation) according to the received instruction, inform the resource sharing node 130 associated with the resource of completion of the operation. According to an embodiment of the present disclosure, the resource sharing node 130 that has received the notification of the completion of the calculation operation described above is the platform server 140 (or nonce for block generation) which describes the completion of the received calculation operation. (nonce) can be informed of any server that created or retained the value. According to an embodiment of the present disclosure, the shared resource node 130 may receive a predetermined nonce value from the platform server 140 and apply the received nonce value to a predetermined hash function. According to an embodiment of the present disclosure, the resource sharing node 130 applies the hash value obtained by applying the nonce value to the hash function to satisfy a predetermined criterion (eg, predetermined difficulty condition predetermined for block generation). It can be determined. According to one embodiment of the present disclosure, if it is determined that the hash result satisfies the predetermined criteria, the resource sharing node 130 may notify the nonce value and the hash result on the blockchain. According to one embodiment of the present disclosure, the predetermined criteria described above are significantly lower than the difficulty of the hash function in the conventional PoW method used in the conventional Bitcoin blockchain, etc. (eg, included in the hash result). Such as greatly reducing the number of zeros). According to one embodiment of the present disclosure, the predetermined criterion described above may be determined such that new blocks may be generated at predetermined time intervals on the corresponding blockchain, for example. According to an embodiment of the present disclosure, each resource sharing node 130 may also receive a nonce value and a hash result that another resource sharing node on the blockchain notifies on the blockchain, and receives the nonce value. You can verify this by applying the hash result to the hash function. According to one embodiment of the present disclosure, when verification by the other resource sharing nodes 130 for the nonce value and the hash result provided on the blockchain by any resource sharing node 130 is completed, the new block is generated. Can be generated. According to one embodiment of the present disclosure, transactions generated by each shared resource node 130 during a predetermined time interval and distributed on the blockchain network 110 for the distributed computing resource sharing system 100 are newly generated. The generated blocks can be stored together.

According to one embodiment of the present disclosure, each of the resource sharing nodes 130 is a distributed computing resource sharing system (under the control of the platform server 140 or in accordance with a predetermined algorithm on the distributed computing resource sharing system 100). At least a portion of the blockchain (ie, blockchain distributed ledger) for 100), although the present disclosure is not so limited. According to one embodiment of the present disclosure, each of the resource sharing nodes 130 is a distributed computing resource sharing system (under the control of the platform server 140 or in accordance with a predetermined algorithm on the distributed computing resource sharing system 100). At least information about smart contracts maintained and managed by the blockchain for 100 (eg, smart contract accounts, etc. that manage the shared use of hardware and / or software resources on distributed computing resource sharing system 100). Some may be stored (though the present disclosure is not so limited).

According to one embodiment of the present disclosure, each user uses his or her user account and resource sharing node 130 to make a query according to a predetermined protocol (eg, a query for a resource management account for searching for a predetermined resource). Can be created and processed. According to an embodiment of the present disclosure, the resource sharing node 130 may process a query related to a predetermined user account by itself or transmit it through the blockchain network 110. According to one embodiment of the present disclosure, a query generated through each user account and resource sharing node 130 may, for example, state of hardware and / or software resources provided for shared use on the distributed computing resource sharing system 100. (E.g., is there a data set resource having a predetermined specification among the shared computing resource regions 120 of the distributed computing resource sharing system 100, etc., but is not limited thereto) , By itself, may not be creating or modifying a new smart contract. Therefore, according to one embodiment of the present disclosure, a query generated through each user account and resource sharing node 130 may not be stored on the blockchain of the distributed computing resource sharing system 100. According to an embodiment of the present disclosure, the resource sharing node 130 generates a response to the query by itself or receives a response to the query from the blockchain network 110, and corresponds to a user (query requester) corresponding to the query. ) You can provide the above response on your account.

According to one embodiment of the present disclosure, the resource sharing node 130 is also configured to distribute the distributed computing resource sharing system in accordance with the payment and receipt of a reward for each user according to the resource sharing on the distributed computing resource sharing system 100. Maintain and manage a digital wallet of user account (s) associated with 100 (although this disclosure is not so limited). According to one embodiment of the present disclosure, a reward for each user (eg, a resource provider) due to resource sharing on the distributed computing resource sharing system 100 may be a predetermined value associated with the distributed computing resource sharing system 100. It can be done by digital money. According to one embodiment of the present disclosure, compensation for resource sharing on the distributed computing resource sharing system 100 may be made in proportion to the amount of usage for each resource. According to one embodiment of the present disclosure, the reward for sharing resources on the distributed computing resource sharing system 100 is evaluated according to a predetermined criterion for each resource used (eg, distributed computing resource sharing of each resource). Assessment of contributions to system 100, etc.).

According to one embodiment of the present disclosure, the platform server 140 may receive a transaction from each resource sharing node 130, including a predetermined resource usage request, provided on the distributed computing resource sharing system 100. Can be. According to an embodiment of the present disclosure, the platform server 140 receives a transaction including a use request for a predetermined resource (eg, GPU resource), processes the received transaction, and obtains information obtained from the transaction. According to an algorithm, an appropriate resource may be allocated among the resources available on each shared computing resource area 120 of the distributed computing resource sharing system 100. According to one embodiment of the present disclosure, the platform server 140, in connection with assigning a predetermined operation (eg, a machine learning operation) to each processing unit unit resource of the shared computing resource area 120, each processing device. It can be allocated so that the time taken to execute the operation allocated to the unit resource is equalized. Although primarily described herein as platform server 140 of distributed computing resource sharing system 100 controls the allocation of each resource on shared computing resource region 120, the present disclosure is not so limited. . According to another embodiment of the present disclosure, on the distributed computing resource sharing system 100, the allocation of the use of each resource on the shared computing resource area 120, etc., is performed by the block server network without intervention by the platform server 140. It should be appreciated that this can be done automatically according to a predetermined algorithm on 110.

According to one embodiment of the present disclosure, the platform server 140, the resource sharing node 130, generates (or mining) and / or block for the recording of each transaction on the distributed computing resource sharing system 100 Can be managed to verify. According to one embodiment of the present disclosure, platform server 140 receives a notification of completion of a predetermined operation by a predetermined processing unit associated with the resource sharing node 130 from any resource sharing node 130. can do. According to an embodiment of the present disclosure, the platform server 140 receiving the aforementioned notification may transmit a predetermined nonce value to the corresponding resource sharing node 130. According to one embodiment of the present disclosure, as described above, the nonce value may be used as an input for a predetermined hash function on the corresponding resource sharing node 130.

According to one embodiment of the present disclosure, the platform server 140 also assists in providing a fair reward to the provider of each resource with respect to the use of each resource provided for sharing on the distributed computing resource sharing system 100. Can be managed. According to an embodiment of the present disclosure, the platform server 140, in accordance with a predetermined criterion, in relation to the use of each resource provided for sharing on the distributed computing resource sharing system 100, for each resource (E.g., evaluating the contribution of each resource to the distributed resource specific system 100, etc.), and according to the evaluation result, the provider of each resource can be managed to make a reasonable compensation accordingly. In the present specification, the platform server 140 of the distributed computing resource sharing system 100 is mainly described as performing evaluation and corresponding compensation for each resource, but the present disclosure is not limited thereto. According to another embodiment of the present disclosure, on the distributed computing resource sharing system 100, the evaluation of each resource and the performance of the rewards thereof may be performed by the blockchain network 110 without intervention by the platform server 140. It should be understood that this can be done automatically according to a predetermined algorithm on the top.

In the figure, the distributed computing resource sharing system 100 is shown to include a platform server 140, but the present disclosure is not limited thereto. According to another embodiment of the present disclosure, it should be understood that the distributed computing resource sharing system 100 may not include the platform server 140.

FIG. 2 is a functional block diagram schematically illustrating a functional configuration of the resource sharing node 130 of FIG. 1, according to an embodiment of the present disclosure. As shown, the resource sharing node 130 includes a transaction processing / generation module 202, a communication module 204, a block generation / verification module 206, a blockchain distributed ledger storage module 208, and a digital wallet. Management module 210.

According to an embodiment of the present disclosure, each user (client) of the distributed computing resource sharing system 100 may provide a predetermined resource for sharing use with other users on the distributed computing resource sharing system 100. If you want to register it, through the transaction processing / generation module 202 of the corresponding resource sharing node (resource sharing node associated with the user or client) 130, it is possible to create a transaction for the resource registration. According to an embodiment of the present disclosure, when a user (client) of the distributed computing resource sharing system 100 wants to use a predetermined resource provided on the distributed computing resource sharing system 100, a corresponding resource sharing node Through the transaction processing / generation module 202 (a resource sharing node associated with the user or client above) 130, a transaction regarding the resource usage may be generated. According to one embodiment of the present disclosure, the transaction processing / generation module 202 of the resource sharing node 130, in addition to transactions relating to the registration and use of resources, various other transactions to be performed through the blockchain network 110 (Such as, but not limited to, trading of digital currency, etc.).

According to one embodiment of the present disclosure, the transaction processing / generation module 202 of each resource sharing node 130 may receive and process a transaction generated by another resource sharing node 130 and distributed on the blockchain. Can be. According to one embodiment of the present disclosure, transaction processing / generation module 202 may execute program code related to a smart contract (eg, a smart contract regarding the provision and use of shared resources, etc.) on a received transaction. According to one embodiment of the present disclosure, transaction processing / generation module 202 executes program code relating to a smart contract on a received transaction, thereby executing a smart contract account (eg, resource management account) based on the information on the received transaction. Information can be updated.

According to an embodiment of the present disclosure, the communication module 204 may operate so that the resource sharing node 130 may communicate with the blockchain network 110 according to a predetermined protocol. According to one embodiment of the present disclosure, the communication module 204 may distribute the blockchain transaction generated by the transaction processing / generation module 202 on the blockchain network 110 according to a predetermined protocol, In addition, a variety of information received through the blockchain network 110 (for example, notification of completion of computation from the processing unit unit on the shared computing region 120 associated with the corresponding resource sharing node 130, other resource sharing node 130) Transaction generated and distributed by the server, a nonce value by the platform server 140, various control signals, etc.) may be received.

According to one embodiment of the present disclosure, the block generation / verification module 206 generates a block for the blockchain network 110 and collects transactions occurring on the distributed computing resource sharing system 100 for a predetermined time. The appropriate block information can be written to the generated block. According to one embodiment of the present disclosure, the block generation / verification module 206 may announce the generated block on the blockchain network 110 through the communication module 204. According to an embodiment of the present disclosure, the block generation / verification module 206 may perform verification on a block generated by another resource sharing node 130, which is known on the blockchain network 110. .

According to one embodiment of the present disclosure, block generation / verification module 206 receives a notification of completion of a given machine learning operation from an associated shared computing resource (eg, a processing unit), and communicates with the communication module 204 and the block. This may be notified to the platform server 140 or the like through the chain network 110 or the like. According to an embodiment of the present disclosure, the block generation / verification module 206 may receive a predetermined nonce value from the platform server 140 through the blockchain network 110, the communication module 204, or the like. The received nonce value can be applied to a predetermined hash function. According to an embodiment of the present disclosure, the block generation / verification module 206 may apply a nonce value to a hash function to obtain a predetermined result (for example, a predetermined difficulty condition predetermined for block generation). Can be determined. According to an embodiment of the present disclosure, the block generation / verification module 206 may notify the nonce value and the hash result on the blockchain when the hash result is determined to satisfy the predetermined criteria. According to one embodiment of the present disclosure, the block generation / verification module 206 may also receive a nonce value and a hash result that another resource sharing node on the blockchain notifies on the blockchain, and receives the received nonce value. We can verify this by applying the hash result to the hash function.

According to an embodiment of the present disclosure, the blockchain distributed ledger storage module 208 may share distributed computing resources under the control of the platform server 130 or according to a predetermined algorithm on the distributed computing resource sharing system 100. Store at least a portion of a blockchain database for the system 100 (ie, a distributed ledger containing all of the history of transactions). In accordance with one embodiment of the present disclosure, the blockchain distributed ledger storage module 208 is also a blockchain network, under the control of the platform server 130 or in accordance with a predetermined algorithm on the distributed computing resource sharing system 100. Information about smart contracts maintained and managed by the blockchain on 110 (eg, information about smart contract accounts regarding the shared use of hardware and / or software resources on distributed computing resource sharing system 100), such as At least a portion of each resource, including various information such as location, format, provider, user, usage history, and usage frequency, etc., but the present disclosure is not limited thereto.

According to one embodiment of the present disclosure, the digital wallet management module 210 may store and manage wallet information for each user account associated with the resource sharing node 130. According to an embodiment of the present disclosure, the digital wallet management module 210 may perform transactions related to digital currency supported on the distributed computing resource sharing system 100, such as resource sharing on the distributed computing resource sharing system 100. The wallet information for each user account on the resource sharing node 130 may be updated by reflecting the payment and receipt of the reward according to the transaction information and other transaction information related to the digital currency.

3 is a blockchain and each block 300 included therein, which may be stored in the blockchain distributed ledger storage module 208 included in the resource sharing node 130 of FIG. 2, according to an embodiment of the present disclosure. It is a diagram conceptually showing the configuration of. As shown, the blockchain may include a plurality of blocks, and each block 300 may include a block header 302 and a plurality of transaction information 304a-304n.

According to an embodiment of the present disclosure, the block header 302 may include a hash value of a previous block header, a nonce value, and a merge of transaction information 304a-304n to be included in the corresponding block 300. , A timestamp indicating the time when the block 300 is generated, and a mining difficulty level of the corresponding block 300. According to one embodiment of the present disclosure, a plurality of transaction information 304a-304n included in block 300 is distributed computing resource sharing system from each resource sharing node 130 of FIG. 1 for a predetermined time interval. It may be a plurality of transaction information distributed on the (100). According to one embodiment of the present disclosure, the transaction information 304a-304n of block 300 is shared resources on the distributed computing resource sharing system 100, generated by the resource sharing node 130 of FIG. 1. Transaction information for providing (ie, registering). According to one embodiment of the present disclosure, transaction information 304a-304n may be transaction information for using shared resources on distributed computing resource sharing system 100, generated by resource sharing node 130. .

4 is an example of transaction information 304 that may be included in block 300 of FIG. 3, in accordance with an embodiment of the present disclosure, for providing shared resources on a distributed computing resource sharing system 100. Illustrates transaction information by way of example. As shown, transaction information 304 may include a resource management account 402, a resource provider account 404, a resource registration execution command 406, and resource information 408.

According to one embodiment of the present disclosure, the resource management account 402 included in the transaction information 304 may be, for example, an address of a smart contract account for managing shared resources on the distributed computing resource sharing system 100. . According to one embodiment of the present disclosure, the resource provider account 404 included in the transaction information 304 may be an account of the user (client) that caused the creation of a transaction for provision of the corresponding shared resource. According to an embodiment of the present disclosure, the resource registration execution command included in the transaction information 304 for providing a shared resource may be, for example, a program code for a smart contract according to ERC20, and may be transmitted to the resource sharing node 130. When executed, the resource information 408 of resources shared by the resource provider account 404 may be stored in association with the resource management account 402. According to one embodiment of the present disclosure, the resource information 408 may include various information including, for example, the type, attribute, location, etc. of the corresponding resource, and the present disclosure is not limited thereto.

5 illustrates a query 500 for retrieving a given shared resource on the distributed computing resource sharing system 100, generated by the resource sharing node 130 of FIG. 1, in accordance with an embodiment of the present disclosure. Illustrated as an example. As shown, the query 500 may include a resource management account 502, a resource search execution command 504, and a requester account 506.

According to one embodiment of the present disclosure, the resource management account 502 included in the query 500 is, as described above, the address of the smart contract account for management of shared resources on the distributed computing resource sharing system 100. Can be. According to an embodiment of the present disclosure, the resource search execution instruction 504 included in the query 500 may be, for example, program code according to ERC20, and when executed by the resource sharing node 130, the predetermined specification may be determined. Command to perform a search for the available resources and provide a response to the search to requestor account 506. According to one embodiment of the present disclosure, the requestor account 506 included in the query 500 may be a user (client) account for generating the corresponding query. As noted above, the query 500 is, according to one embodiment of the present disclosure, for example only for querying the status of hardware and / or software resources provided for shared use on the distributed computing resource sharing system 100. , By itself, may not be creating or modifying a new smart contract. Therefore, according to one embodiment of the present disclosure, the query 500 may not be stored on the blockchain of the distributed computing resource sharing system 100, unlike each transaction information 304.

FIG. 6 is an example of transaction information 304 included in block 300 of FIG. 3, in accordance with an embodiment of the present disclosure, for requesting use of shared resources on distributed computing resource sharing system 100. Illustrates transaction information by way of example. As shown, the transaction information 304 may include a resource management account 602, a resource user account 604, a resource usage execution command 606, and a utilization resource information 608.

According to one embodiment of the present disclosure, the resource management account 602 included in the transaction information 304 may be, for example, an address of a smart contract account for managing shared resources on the distributed computing resource sharing system 100. . According to one embodiment of the present disclosure, the resource user account 604 included in the transaction information 304 may be an account of the user (client) that caused the creation of a transaction for the use request of the shared resource. According to an embodiment of the present disclosure, the resource use execution command included in the transaction information 304 for requesting the use of the shared resource may be, for example, program code for a smart contract according to ERC20, and the resource sharing node 130 When executed by, the resource determined in accordance with the request may be associated with the resource user account 404 to be used (and information about usage is stored). According to one embodiment of the present disclosure, the resource information 408 may include various information including a type, an attribute, a location, a usage amount, and the like of the corresponding resource, and the present disclosure is not limited thereto.

7 is a functional block diagram schematically illustrating the functional configuration of the platform server 140 of FIG. 1, according to an embodiment of the present disclosure. As shown, the platform server 140 includes a communication module 702, a shared resource distribution module 704, a block generation / validation management module 706, and an evaluation / reward module 708.

According to an embodiment of the present disclosure, the communication module 702 may operate so that the platform server 140 may communicate with the blockchain network 110 according to a predetermined protocol. According to an embodiment of the present disclosure, the communication module 702 may receive various transactions and the like distributed from each resource sharing node 130 through the blockchain network 110.

According to one embodiment of the present disclosure, when a transaction including a predetermined resource use request is received, the shared resource distribution module 704 may, based on a predetermined algorithm, according to information obtained from the received transaction, Appropriate resources among the resources available on the distributed computing resource sharing system 100 may be allocated to the request. According to one embodiment of the present disclosure, the shared resource distribution module 704 is further configured to assign a predetermined operation (eg, a machine learning operation) to each processing unit unit resource of the shared computing resource area 120. It can be allocated so that the time taken for the execution of the operation allocated to the processing unit unit resource is equalized.

According to one embodiment of the present disclosure, the shared resource distribution module 704 reflects information on resource usage requests on the received transactions and the allocation of resources made corresponding thereto, so as to provide information on the corresponding resources associated with the resource management account. Can be updated. According to one embodiment of the present disclosure, the shared resource distribution module 704 also stores each resource sharing node 130 to store at least a portion of a blockchain database (ie, a distributed ledger) for the distributed computing resource sharing system 100. ) May be controlled / managed, but the present disclosure is not limited thereto.

According to one embodiment of the present disclosure, the block generation / verification management module 706 may generate a series of nonce values according to a predetermined criterion. According to one embodiment of the present disclosure, the block generation / verification management module 706 completes the execution of a predetermined operation by any processing unit associated with the resource sharing node 130 from any resource sharing node 130. Receive notification of. According to an embodiment of the present disclosure, the block generation / verification management module 706 may sequentially allocate and transmit the generated nonce value to each resource sharing node 130. According to an embodiment of the present disclosure, the nonce value may be used as an input to a predetermined hash function on the corresponding resource sharing node 130 as described above.

According to one embodiment of the present disclosure, the evaluation / compensation module 708 is configured to evaluate each resource according to a predetermined criterion in relation to each resource provided for sharing on the distributed computing resource sharing system 100. For example, evaluation of each resource's contribution to the distributed computing resource sharing system 100 may be performed. According to an embodiment of the present disclosure, the evaluation / compensation module 708 may manage the provider of each resource to make a reasonable compensation according to the evaluation result. In the present specification, the distributed computing resource sharing system 100 includes a platform server 140, and has been described as performing management of allocation (distribution) and block generation / verification of shared resources through the platform server 140. However, the present disclosure is not limited thereto. According to another exemplary embodiment of the present disclosure, the distributed computing resource sharing system 100 may not include the platform server 140, and management of resource allocation or block generation and verification according to a request for using a shared resource may be performed. It should be noted that the distributed computing resource sharing system 100 may be automatically performed according to a predetermined algorithm.

FIG. 8 schematically illustrates an example of a process of providing shared resources on the distributed computing resource sharing system 100 through the resource sharing node 130 of FIG. 1, according to an embodiment of the present disclosure. This is a flow chart.

First, in step 802, a resource provider (eg, a predetermined GPU resource, a storage unit, a data set, or a learning model) held by the resource provider on the resource sharing node 130 is distributed on the distributed computing resource sharing system 100. A user who wants to provide a shared use may generate a transaction (ie, a resource provision registration transaction) for sharing a resource through a corresponding resource sharing node 130. Then, in step 804, the generated transaction can be distributed on the blockchain network 110. According to one embodiment of the present disclosure, a transaction distributed on blockchain network 110 may be received by both platform server 140 and other resource sharing nodes 130. According to one embodiment of the present disclosure, a blockchain (eg, resource sharing nodes 130 on a blockchain) that receives a resource provision registration transaction processes the received transaction and provides a resource provision registration transaction to the newly created block. Can be recorded. According to one embodiment of the present disclosure, in step 806, a transaction is processed by the blockchain (eg, each resource sharing node 130 on the blockchain), thereby updating the resource management account (ie, Registration of the corresponding resource) may be performed.

9 is an operational flow diagram schematically illustrating an example of a process of requesting, providing, and using data resources, performed on a distributed computing resource sharing system 100, according to an embodiment of the present disclosure.

As shown, first at step 902, a resource requestor on a resource sharing node 130 (e.g., a user as a developer who wants to utilize a given data set for machine learning) is passed through the resource sharing node 130; It may create a use request transaction (eg, a data use request transaction) for a resource (eg, a given data set) that it desires. According to one embodiment of the present disclosure, the use request transaction for a given data set may include a specific specification (eg, data content, format, etc.) of the requested data set. Then, in step 904, the generated transaction can be distributed on the blockchain network 110. According to one embodiment of the present disclosure, a transaction distributed on blockchain network 110 may be received by both platform server 140 and other resource sharing nodes 130.

In step 906, one of the resource sharing nodes 130 that has received the aforementioned data use request transaction may generate a resource, such as a data set, that meets the needs of the transaction. Then, in step 908, the resource sharing node may create a share offer transaction for the generated data set and distribute it through the blockchain network 110. In step 910, the data sharing provision transaction distributed in step 908 may be processed on the blockchain and the corresponding data resource may be registered and stored in association with the resource management account. The process of steps 906 to 910 may correspond to and refer to the disclosure regarding the operational flow in FIG. 8 described above.

Then, in step 912, in response to the previously received data use request transaction 904, the requested resource (e.g., a given data set resource) is allocated for the requestor and is assigned to the allocation and use of that resource. May be stored in association with the resource management account. Subsequently, in step 914, the use of the resource may be made for the resource requester.

According to an embodiment of the present disclosure, a resource corresponding to a resource use request transaction is a predetermined data set, and the data set is a storage unit of multiple locations on the distributed computing resource sharing system 100 (eg, a plurality of shared computings). When distributed and stored in the resource areas 120, each part of the corresponding data set may be transmitted in parallel from the above storage units (e.g., the plurality of shared computing resource areas 120) for the resource requester. This can solve the problem of the transmission speed due to the transmission of large data. According to one embodiment of the present disclosure, the details of which part of the data set is to be transmitted from which of the plurality of storage units, for example, by the shared resource distribution module 704 of the platform server 140, The distance between the resources (e.g., the distance between the hardware resources involved in the operation and the storage unit of the data set to be used in the operation) or the communication time, etc. It should be understood, however, that the present disclosure is not limited thereto. According to an embodiment of the present disclosure, a resource corresponding to a resource use request transaction is a predetermined data set, and the data set is one storage unit (eg, one shared computing resource on the distributed computing resource sharing system 100). When stored in region 120, machine learning operations for the resource requester may be performed by the GPU at the same location as the storage unit (ie, shared computing resource region 120 to which the storage unit belongs), thereby The problem of the transmission speed due to the transmission of a large amount of data can be solved.

In step 914, after the use of the resource is made by the resource requester, in step 916 the settlement of the use of the resource is made from the resource requester (of the digital wallet) and the right reward for the provider of the resource. This can be done. According to one embodiment of the present disclosure, the settlement for resource usage performed in step 916 may be automatically made through a smart contract included in the resource usage request transaction. According to one embodiment of the present disclosure, settlement through a smart contract on a resource use request transaction may be made by a predetermined fee being provided to the platform server 140 from the digital wallet of the resource requester (user) (but the present disclosure Not limited thereto). According to one embodiment of the present disclosure, some of the fee received from each resource requester (user) is provided by the platform server 140 to the resource provider of each resource provided to the distributed computing resource sharing system 100. Some settlements may be made in the form of distribution depending on the amount of resources (although this disclosure is not so limited). According to one embodiment of the present disclosure, some of the royalties received from each resource requester (user) are evaluated by the platform server 140 for the provided resources (eg, the entire distributed computing resource sharing system of the corresponding resources). Based on the contribution evaluation to 100), etc., but may be partially settled in the form of differential distribution to the resource providers of each resource (though the present disclosure is not limited thereto).

In the present specification, mainly, the distribution and allocation of shared resources, generation and storage of blocks, settlement and compensation for resource use, etc. have been described under the intervention of the platform server 140, but the present disclosure is not limited thereto. According to another embodiment of the present disclosure, it should be appreciated that such an operation may be performed without the intervention of the platform server 140 according to a predefined algorithm on the distributed computing resource sharing system 100.

As will be appreciated by those skilled in the art, the present disclosure is not limited to the examples described herein but may be variously modified, reconfigured and replaced without departing from the scope of the present disclosure. It should be appreciated that the various techniques described herein may be implemented by hardware or software, or a combination of hardware and software.

A computer program according to an embodiment of the present disclosure may include a storage medium readable by a computer processor or the like, such as an EPROM, an EEPROM, a nonvolatile memory such as a flash memory device, a magnetic disk such as an internal hard disk and a removable disk, a magneto-optical disk, and It may be implemented in a form stored in various types of storage media, including a CDROM disk. In addition, the program code (s) may be implemented in assembly or machine language. All modifications and changes that fall within the true spirit and scope of this disclosure are intended to be covered by the following claims.

Claims (10)

A mining method performed at a node on a blockchain-based distributed computing resource sharing system including a server and a plurality of nodes,
Receiving an instruction to execute a predetermined operation;
Notifying the server of completion of execution of the operation according to the received instruction;
In response to the notification, receiving a predetermined nonce value from the server;
Applying the nonce value to a predetermined hash function;
Determining whether a hash result obtained by applying the nonce value to the predetermined hash function satisfies a predetermined criterion; And
If it is determined that the hash result satisfies the predetermined criteria, notifying the blockchain of the nonce value and the hash result.
Including, mining method. The method of claim 1,
And said operation includes an operation of a predetermined number of executions of a predetermined machine learning code. The method of claim 1,
The node,
Receiving a nonce value and a hash result from another node of the plurality of nodes; And
Verifying the nonce value and the hash result received from the other node by applying the hash function.
Including, mining method. The method of claim 1,
The predetermined criterion is a mining method that is determined such that new blocks can be generated at predetermined time intervals on the blockchain. A mining support method performed by the server on a blockchain-based distributed computing resource sharing system including a server and a plurality of nodes,
Receiving, from the outside, a request for one or more tasks, each of the one or more tasks comprising a predetermined number of executions of a predetermined operation;
Allocating execution of operations according to the one or more tasks on the plurality of nodes;
Receiving, from any one of the plurality of nodes, a notification indicating completion of execution of the assigned operations for the one node; And
In response to receiving the notification, transmitting, to the one node, a predetermined nonce value, wherein the nonce value is one of a plurality of values generated on the server, the hash function of the one being of a hash function. Used as input-
Mining support method comprising a. The method of claim 5,
Allocating execution of the operations on the plurality of nodes includes allocating such that the time taken to execute the operations on the plurality of nodes is equalized. The method of claim 5,
Wherein each of the operations includes execution of predetermined machine learning code. The method of claim 5,
And the nonce value, when used as an input of the hash function on the one node, causes a hash result obtained from the hash function correspondingly to satisfy a predetermined criterion. The method of claim 8,
The predetermined criterion is that the new block is determined so that new blocks can be generated at predetermined time intervals on the blockchain. A computer readable recording medium having recorded thereon instructions executable by a computer, wherein the instructions, when executed by the computer, cause the computer to perform the method according to any one of claims 1 to 9. Computer-readable recording medium.

Images