JP2020144614A - Task decentralization processing system - Google Patents

Abstract

To reduce the processing load of a task with a large amount of information.SOLUTION: The present invention is a system in which task decentralization processing is performed between a management server and a plurality of processing devices. The management server divides an acquired task and sends divided tasks to the processing devices according to a predetermined condition. Each of the processing devices processes the acquired task. In the task division, at least the size thereof is determined according to a processing terminal in which processing is scheduled.SELECTED DRAWING: Figure 1

Description

The present invention relates to a task distributed processing system.

In recent years, a P2P type system has been used for trading cryptocurrencies such as Bitcoin (registered trademark) (see, for example, Patent Document 1).

In particular, recently, the pool manager of mining in the blockchain network is the recipient of the reward, the participating miners are made to mine the same block, and the reward paid to the pool manager when the mining is successful is the amount of mining work. A mechanism for distributing among participants has also been proposed.

Japanese Patent No. 6326173

The amount of calculation required to maintain the blockchain network as described above (for example, hash power by GPU) operates efficiently because it only performs mining despite its ability to perform enormous calculations. I can't say that.

Therefore, an object of the present invention is to efficiently operate the resources for maintaining the blockchain network.

Modern de-neural networks, such as those used in self-driving cars, require enormous amounts of computing power. In particular, considering the computing power required for recent neural networks and the dataset size required for training, the current situation is that even a so-called supercomputer is not sufficient.

For example, systems where safety is extremely important and detection accuracy requirements are much higher than in the Internet industry, such as self-driving cars, are required to operate without problems regardless of weather conditions, field of view, or road surface conditions. To achieve this level of recognition accuracy, it is necessary to train the neural network with a sample dataset consisting of examples of all possible driving operations, weather, and situational conditions. The present invention has been invented in view of such circumstances.

That is, according to the present invention, the management server and the plurality of processing devices are task distribution processing systems.
The management server divides the acquired task and sends it to the processing device according to a predetermined condition.
Each of the processing devices processes the acquired task.
A task distributed processing system is obtained.

According to the present invention, it is possible to efficiently operate the resources for maintaining the blockchain network.

It is a figure which shows the structural example of the system by embodiment of this invention. It is a block diagram which shows the hardware configuration of the management server of FIG. It is a block diagram which shows the hardware configuration of the user terminal of FIG. It is a figure which shows the flow of processing of the system by this embodiment. It is a data image diagram of a task block in this embodiment. It is a table which shows the structural example of the task block of FIG.

The contents of the embodiments of the present invention will be described in a list. The content platform system according to the embodiment of the present invention has the following configuration.
[Item 1]
The management server and multiple processing devices are task distribution processing systems.
The management server divides the acquired task and sends it to the processing device according to a predetermined condition.
Each of the processing devices processes the acquired task.
Task distributed processing system.
[Item 2]
The task distribution processing system according to item 1.
The task division is at least sized according to the processing terminal scheduled to be processed.
Task distributed processing system.

<Details of the embodiment>
Hereinafter, a distributed machine learning system (hereinafter referred to as “system”) according to an embodiment of the present invention will be described with reference to the drawings.

<Overview>
The system according to the embodiment of the present invention utilizes or combines the computing power generated by a plurality of computers constituting the mining pool for the calculation of algorithms such as machine learning (for example, the process of reading teacher data and optimizing the algorithm). It is done efficiently by doing.

<Composition>
As shown in FIG. 1, in this system 100, the management server 1, the user terminals 2 (2a to 2d), and the pool management server 3 (3a to 3c) are connected to a network (Internet or the like, and a blockchain network). Included) Connected via 50.

In the present embodiment, at least the management server 1 and the pool management server 3 are connected to the ledger (blockchain) jointly managed in the blockchain network so that they can be accessed, referenced, and added.

As will be described later, the user terminal 2 according to the present embodiment is any object that can be automatically controlled, such as an automobile 2a, a railway 2b, a robot 2c in a factory, a bus 2d, and the like. May be good.

<Hardware configuration>
As shown in FIGS. 2 and 3, the management server 1 and the user terminal 2 according to the present embodiment have the following hardware configurations. The following configuration is an example, and may have other configurations. Further, each functional block may exist in each terminal, or may be logically configured such as being available through a cloud service or the like.

<Management server 1>
As shown in FIG. 2, the management server 1 constitutes a part of the system by executing information processing via communication between the user terminal 2 and the pool management terminal 3. The management server 1 may be a general-purpose computer such as a workstation or a personal computer, or may be logically realized by cloud computing.

The management server 1 includes at least a processor 10, a memory 11, a storage 12, a transmission / reception unit 13, an input / output unit 14, and the like, and these are electrically connected to each other through a bus 15.

The processor 10 is an arithmetic unit that controls the operation of the entire management server 1, controls the transmission and reception of data between each element, and performs information processing and the like necessary for executing an application. For example, the processor 10 is a CPU (Central Processing Unit), and executes each information processing by executing a program or the like stored in the storage 12 and expanded in the memory 11.

The memory 11 includes a main storage composed of a volatile storage device such as a DRAM (Dynamic Random Access Memory) and an auxiliary storage composed of a non-volatile storage device such as a flash memory or an HDD (Hard Disk Drive). .. The memory 11 is used as a work area or the like of the processor 10, and also stores a BIOS (Basic Input / Output System) executed when the management server 1 is started, various setting information, and the like.

The storage 12 stores various programs such as application programs, authentication programs for blockchain networks, and algorithms used for machine learning. A database storing data used for each process may be built in the storage 22.

The transmission / reception unit 13 connects the management server 1 to the network 3. The transmission / reception unit 13 may be provided with a short-range communication interface of Bluetooth (registered trademark) and BLE (Bluetooth Low Energy).

The input / output unit 14 is an information input device such as a keyboard and a mouse, and an output device such as a display, which are used as needed.

The bus 15 is commonly connected to each of the above elements and transmits, for example, an address signal, a data signal, and various control signals.

<User terminals 2a and 2b>
As shown in FIG. 3, the user terminal 2 is a terminal installed or integrally configured in the above-mentioned automobile 2a, railway 2b, robot 2c in a factory, and bus 2d. The user terminal 2 constitutes a part of the system by executing information processing via communication with the management server 1. The user terminal 2 may be configured such that, for example, a terminal that controls a control system of an automobile exerts its function.

As shown, the user terminal 2 includes at least a processor 20, a memory 21, a storage 22, a transmission / reception unit 23, an input / output unit 24, and the like, which are electrically connected to each other through a bus 25. These functions may be shared with a control system such as an automobile.

Similar to the management server 1 described above, the processor 20 is an arithmetic unit that controls the operation of the entire user terminal 2, controls the transmission and reception of data between each element, and performs information processing and the like necessary for executing an application. For example, the processor 20 is a CPU (Central Processing Unit), and executes each information processing by executing a program or the like stored in the storage 22 and expanded in the memory 21.

The memory 21 includes a main storage composed of a volatile storage device such as a DRAM (Dynamic Random Access Memory) and an auxiliary storage composed of a non-volatile storage device such as a flash memory or an HDD (Hard Disk Drive). .. The memory 21 is used as a work area or the like of the processor 20, and also stores a BIOS (Basic Input / Output System) executed when the user terminal 2 is started, various setting information, and the like.

The storage 22 stores various programs such as an application program, an authentication program for a blockchain network, and an algorithm for controlling an automobile or the like. A database storing data used for each process may be built in the storage 22.

The transmission / reception unit 23 connects the user terminal 2 to the network 3. The transmission / reception unit 23 may be provided with a short-range communication interface of Bluetooth (registered trademark) and BLE (Bluetooth Low Energy).

The input / output unit 24 is an information input device such as a keyboard and a mouse, and an output device such as a display, which are used as needed.

The bus 25 is commonly connected to each of the above elements and transmits, for example, an address signal, a data signal, and various control signals.

<Processing flow>
Subsequently, with reference to FIG. 4, the information exchanged with the operation of each element shown in FIG. 1 will be described.

As shown in FIG. 4, the user terminal 2 (automobile 2a) transmits its own driving data as teacher data to the management server 1 (SQ101). The driving data includes, but is not limited to, for example, behavior / control data of the automobile itself and data acquired from a sensor attached to the automobile (such as an image outside the automobile).

The automobile 2a performs three processes of "cognition", "judgment", and "operation" in conjunction with each other by a predetermined algorithm. For example, there are a collision prevention system, a system that adjusts the distance to the preceding vehicle, a system that prevents deviation from the lane, etc., but it makes an appropriate judgment according to the surrounding vehicles and surrounding conditions and moves to the destination. Also includes advanced ones. The algorithm is provided by the management server 1 and is kept up to date by being constantly updated.

The management server 1 updates and optimizes an algorithm (for example, neural network parameters, etc.) based on the teacher data obtained from the user terminal 2. However, as described above, the process from the extraction of the feature amount of the teacher data to the optimization of the algorithm often has a huge amount of calculation, and as the number of user terminals 2 increases, the calculation impossible becomes exponential. Ascend to.

Therefore, the management server 1 divides the task for optimizing the algorithm from the provided teacher data (SQ102). The task may be simply the teacher data alone, the teacher data may be subjected to primary processing, or may be batched so that the calculation can be started immediately after being read. Various methods can be adopted as the method of division.

The management server 1 broadcasts the task to the pool management server 3 to process the task (SQ103). In the present embodiment, the tasks are exhausted, and the size is optimized according to the specifications and the state of the processing terminal. The task according to the present embodiment causes the mining pool to perform the same calculation as the search for the hash value (nonce) in the blockchain network. As a result, a predetermined reward is given to the person who updates (optimizes) the algorithm earliest. That is, the distributed machine learning system according to the present invention is configured by replacing the mining work in the blockchain with the update work of the machine learning algorithm.

FIG. 5 shows block information when the present invention is mounted on a blockchain network. Task information is posted on the blockchain, and information of the processing terminal that processed the task is also recorded. Here, each task is associated with a user terminal from which it is provided.

Further, as shown in FIG. 6, the structure of each task includes, for example, the size of the entire data (block size), header information (Header), the number of tasks (task counter), and information on the tasks to be processed (task). ) Is included.

The blow management server 3 assigns the processing terminal (“miner” in the blockchain network) connected to the pool to execute the task and causes the calculation to be performed (SQ104). The allocation may be random, or may be allocated in order from the end of the connected processing terminal.

If any of the processing terminals succeeds in updating the algorithm, the result information is transmitted to the management server 1 via the pool management server 3 (SQ105). The management server 1 updates the algorithm based on the result information (SQ106) and provides it to the user terminal 2 (SQ107).

According to the system described above, it is possible to efficiently perform machine learning based on teacher data obtained from various user terminals.

The processing target of the system described above is the teacher data, but any other task may be used as long as it is processing using the device.

In addition, a plurality of main tasks may be combined to customize the size and breakdown according to the device that performs the actual processing. In this case, the management server side may acquire information on the processing capacity of each connected device in advance (or in response to a request).

The above-described embodiment is merely an example for facilitating the understanding of the present invention, and is not intended to limit the interpretation of the present invention. It goes without saying that the present invention can be modified and improved without deviating from the gist thereof, and the present invention includes an equivalent thereof.

1 Management server (learning management terminal)
2,2a-2d user terminal (teacher data supply terminal)
3, 3a to 3c pool management terminal (processing pool management terminal)
30 processing pool 50 network 100 system (distributed machine learning system)

Claims (2)

The management server and multiple processing devices are task distribution processing systems.
The management server divides the acquired task and sends it to the processing device according to a predetermined condition.
Each of the processing devices processes the acquired task.
Task distributed processing system. The task distributed processing system according to claim 1.
The task division is at least sized according to the processing terminal scheduled to be processed.
Task distributed processing system.