KR102197989B1 - Election System Using Identification of Artificial Intelligence Based on Block Chain - Google Patents

Abstract

The present invention uses artificial intelligence to identify a voter's picture online so that the voter's vote can be made in the case of a picture of a real person, and artificial intelligence identity authentication that enables the voter's voting information to be reliably secured with a blockchain. It relates to a blockchain-based election system using
The block chain-based election system according to the present invention is an election system in which voters are authenticated and voted online, and whether a photo registered by a voter through their client terminal 100 is a real photo or a fake photo created by a computer. In the case of an actual photo, it is determined through machine learning, and if it is an actual photo, it performs member registration of the voter, provides voting information to the client terminal 100 of the registered voter, and blocks voting information that the voter proceeds through the client terminal 100 Consisting of including the election management server 200 secured through a chain, it is provided to prevent the voting of an unjustified person and increase the security of the voting.

Description

Election System Using Identification of Artificial Intelligence Based on Block Chain

The present invention relates to a block-chain-based election system, and more particularly, by using artificial intelligence to determine a voter's picture online, in the case of a real person picture, the voter's vote can be made, and the voter's voting information is blocked. It relates to a blockchain-based election system using artificial intelligence self-authentication that enables stable security with a chain.

Election is the act of freely electing a representative or officer of a group or group by a person with certain qualifications among its members. These elections are decided by voting, which is a voter expressing his or her intention on the ballot and giving it to a certain place, indicating the voter's intention to appoint which candidate to be elected.

In these elections, after the voters go to the polling place to vote and verify their identity, they are provided with a ballot and vote with a stamp under the name of the candidate who wishes to vote, and the ballot box after the vote is collected by constituency and counting is performed. The election proceeds.

However, in such a conventional election, both the voting and counting work by the voters are carried out manually, so that processing time and processing manpower are required, resulting in severe resource consumption. In particular, since the voters have to go to the polling place and proceed to vote, it is cumbersome for the voters, and such manual voting also causes the turnout to decrease.

In response to these problems, in recent years, an electronic voting system has been proposed. In such an electronic voting system, the voter moves to the electronic voting machine at an electronic voting place and authenticates itself through an accredited certificate. Voting takes place through the selection process. Since the voter's voting results are collected and analyzed in real time through an electronic voting system and the counting results are produced, there is an advantage that it can be proceeded quickly and accurately without consuming time and manpower compared to manual operations.

However, in such an electronic voting system, only voters with a public certificate can vote electronically, so voters without a public certificate have a cumbersome problem of having to obtain a separate public certificate.If the public certificate is stolen or abused, a person who is not a legitimate voter There was a concern that an illegal voting could be made. In addition, there still remains a cumbersome problem in that voters have to move to an electronic polling place where an electronic voting machine is installed.

Korean Registered Patent Publication No. 10-0643182 (registered on October 31, 2006)

The present invention has been proposed in order to solve the problems of the conventional election system, and an object of the present invention is that the voter can easily and safely receive their own authentication using their own mobile phone without moving to a separate polling place to stably vote. It is to provide an election system that allows it to be conducted.

The election system according to the present invention for achieving the above object is an election system in which a voter is authenticated and voting is performed online, and whether a photo registered by a voter through their client terminal is a real photo or a fake photo made by a computer. Election in which the voter registers as a member in the case of an actual photo by machine learning, provides voting information to the registered voter's client terminal, and secures the voting information conducted by the voter through the client terminal through a blockchain. It includes a management server.

Here, a member management unit for registering and managing a member by assigning a member code after authenticating a voter who accesses the election management server through a client terminal; A voting management unit that provides a voting page to a client terminal so that a registered voter can vote, and collects and manages voting results conducted by the voters; An artificial intelligence deep learning model for receiving a voter picture from a client terminal of a voter requesting membership registration through the member management unit and determining whether the voter picture is a real person or a virtual person through artificial intelligence deep learning, which is machine learning; Through the member management unit, a block chain account is created and provided to the registered voter, and the voter is verified through the generated block chain account, and voting is made online through the voting management unit, and the information on the votes conducted by the voter is blocked. Blockchain model that is stored in and secured; is provided.

The artificial intelligence deep learning model includes a learning data generation module that generates learning data that is a fake photo through Generative Adversarial Networks (GAN) so that learning to distinguish between a real photo and a fake photo can be performed, and the learning material generation module. After learning the provided fake photos and real photos through DenseNet, a photo learning and identification module that determines whether or not the voter photo is authentic is provided.

In addition, the blockchain model creates an Ethereum account with an ID and password that is a member code assigned according to a member registration of a voter, records the generated Ethereum account information in a smart contract, and the voter registers and Ether When logging in with the private key, ID and password created according to the Leeum account, the voter is compared with the public key contained in the block, and when the voter votes, the value is stored in the block.

Meanwhile, the client terminal is preferably a personal terminal including any one of a smart phone, a personal digital assistant (PDA), a notebook, and a personal computer (PC) used by voters.

According to the present invention, by allowing voters to vote online using their mobile phone, it is possible to solve the troublesome problem of having to move to a conventional polling place to vote, and thereby increase the turnout rate. In addition, by using artificial intelligence deep learning technology to determine the authenticity of a voter's photo and perform identity verification, it is possible to prevent voting by an unjustified person, and by managing the voter's personal information and voting information using a blockchain There is an effect that can increase security.

1 is an overall network connection diagram of a block chain-based election system according to the present invention,
2 is a block diagram of an election management server according to the present invention;
3 is a flow chart showing a process for a voter to vote online according to the present invention.

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

1 shows an overall network connection diagram of a blockchain-based election system according to an embodiment of the present invention.

As shown in FIG. 1, the blockchain-based election system provides voting information to the client terminal 100, which voters use to perform voting, and the client terminal 100, and proceeds by the client terminal 100. It comprises an election management server 200 for collecting and managing the voting results.

The client terminal 100 is a personal terminal such as a smart phone, a personal digital assistant (PDA), a notebook, and a personal computer (PC) used by voters, and the client terminal 100 is a remote election By connecting communication with the management server 200, voting-related information is transmitted and received. Voters access the election management server 200 through the client terminal 100 in their possession, register their photo to determine the authenticity of the photo, and then perform membership registration, and after membership registration, the election management server ( 200), you can vote online through the voting page. Since the client terminal 100 corresponds to each voter, a plurality of client terminals 100 are connected to the election management server 200 through a network according to the number of voters.

The election management server 200 authenticates a voter who accesses through the client terminal 100 in the hopes of voting and assigns a member code, and provides a voting page to the client terminal 100 for which the member is authenticated to vote for the voter online. It is a server computer that collects and manages voter voting information. In an embodiment of the present invention, the election management server 200 uses artificial intelligence to check whether the picture registered by the voter is a virtual person created by a computer or an actual person, and if it is found to be a real person, It is provided for voting by performing member registration by authenticating. In addition, it is possible to increase the security of voting by securing the voter's information and voting results using a blockchain.

2 is a block diagram of an election management server according to an embodiment of the present invention.

As shown in Fig. 2, the election management server 200 according to the present invention includes an input unit 220 and an output unit 230 for inputting and displaying data, and an external client terminal 100 through a network. A communication unit 240 performing communication with a computer system, an interface unit 250 for transmitting and receiving data with an external peripheral device, and an election management engine 260 that authenticates voters to vote and manage the voter's voting results. And, it comprises a central control unit 210 for controlling the operation of each of the components.

The central control unit 210 is a device that controls and manages each component of the election management server 200, and the central control unit 210 includes a general central processing unit (CPU), RAM, and ROM. A hardware device such as, etc., and software that recognizes and drives the hardware device are provided to control the overall operation of the election management server 200.

The election management engine 260 is a program that authenticates the identity of a voter accessing through the client terminal 100, provides voting information to the voter, and collects and manages the voting results conducted by the voters. ) Is provided with a member management unit 261, a voting management unit 262, an artificial intelligence deep learning model 263, and a block chain model 264.

The member management unit 261 is a program module that authenticates a voter who accesses through the client terminal 100 to register and manage a member, and the member management unit 261 is linked with an artificial intelligence deep learning model 263 to provide a picture of the voter. The member code (ID, Password) is given by authenticating the voter through

The voting management unit 262 is a program module that provides a voting page through the member management unit 261 through which a member is authenticated to vote, and collects and manages the voting results conducted by the voters, and the voting management unit 262 ) Maintains the security of voting using blockchain technology in conjunction with the blockchain model 264.

The artificial intelligence deep learning model 263 is a machine learning program that determines whether a voter's picture is a real person or a virtual person through artificial intelligence's deep learning technology, and the artificial intelligence deep learning model 263 includes Learning material generation module (263a) that generates learning materials that are fake photos so that learning to distinguish between real photos and fake photos can be performed, and learning and discrimination of photos that determine the authenticity of photos by learning real photos and fake photos. A module 263b is provided. The deep learning technology of artificial intelligence uses DenseNet, which adds a new concept called Dense connectivity from the sequential combination of sequence convolution, activation, and pooling of CNN, which is a conventional deep learning technology. As the network gets deeper, the information initially held may disappear or be washed out. In DenseNet, the concept of solving this problem is that the information accumulated for the first time is continuously accumulated even if the network deepens. On the other hand, GAN (Generative Adversarial Networks) has modules with different purposes, Generator and Discriminator, which learns actual data and creates similar fake data, with the purpose of creating actual data. do. In addition, the discriminator aims to distinguish whether the input data is real or fake data, and an image is created with the confrontation structure of these two modules. In an embodiment of the present invention, the learning material generation module 263a learns a picture through a GAN, generates a fake picture learning material, and provides it to the picture learning and identification module 263b, and the picture learning and identification module 263b Learns the learning materials provided by the learning material generation module 263a through DenseNet, and then determines whether the voter's picture is a real picture.

The blockchain model 264 is a program responsible for security of voting conducted through the voting management unit 262, and in the embodiment of the present invention, the blockchain module is implemented through Ethereum with a lot of information. There is a smart contract (264a) in Ethereum that composes the blockchain module, and the smart contract (264a) possessed by Ethereum is one of the blockchain technology, and it is digital as a real person performs according to a contract. This is a function that allows you to create a contract with commands and automatically execute the contract contents according to the conditions. When using this technology, when the actual conditions are satisfied, the contract is automatically executed, so that the user can use it more conveniently. In the present invention, an Ethereum account is created with an ID and password according to a member registration of a voter, and the generated record is recorded in the smart contract 264a. If you log in with the private key, ID and password created by creating an Ethereum account and signing up as a member, you can verify that you are the voter by comparing the public key in the block, and when the voter votes, the value is saved in the block. Becomes. On the other hand, this block chain model 264 can use not only Ethereum but also bitcoin that is implemented with Ethereum.

Hereinafter, a process of voting through a block chain-based election system configured as described above will be described.

3 is a flowchart showing a process in which a voter votes online according to an embodiment of the present invention, and a voter joins a member by accessing a website operated by the election management server 200 using a client terminal 100 It indicates that all the voter's actions are recorded due to the smart contract 264a of the blockchain model 264 when the voter logs in and the voter logs in.

First, for voting, the voter connects to the website of the election management server 200 through a client terminal 100 such as a smartphone that they use, and signs up for membership through the member management unit 261. Through the voter's picture, whether the picture is a picture of a real person or a fake picture made by a computer is determined through the artificial intelligence deep learning model 263, which is machine learning, and the membership registration is successful only in the case of the actual picture.

In an embodiment of the present invention, GAN was used as the learning material generation module 263a for generating photographic learning data, and DenseNet was used as the photographic learning and identification module 263b for learning and discriminating photographs. Here, GAN learns actual data and generates similar fake data and provides it to DenseNet, and DenseNet performs learning through fake data and actual data provided by GAN, and then, based on the learning result, whether the voter's picture is a real picture. It determines if it is a fake photo.

To do this, first, real data is input to the GAN, and the GAN's discriminator learns the distribution of the real data to distinguish between real data and fake data. The generator of GAN composes a fake image with a random value, so that the discriminator can judge that it has the actual data. In other words, the purpose of the Generator is to make the discriminator's judgment about fake data real data, so continued learning forms a fake image that is closer to the actual data. The closer the fake data is to the real data, the closer to 1 will be, and if you check the value of the input data distribution using this, you can check the fake and real data. CNN's DenseNet learns fake data and real data generated by GAN, and plays a role of classifying real and fake data with the learned distribution. That is, when data comes in and analyzes it with DenseNet to check the distribution and the distribution of the values is similar to the real data, DenseNet classifies it as real data.

When it is determined as an actual photo of a voter through the photo learning and identification module 263b of the artificial intelligence deep learning model 263, the member management unit 261 proceeds to sign up for the voter and assigns a member ID and password. On the other hand, in the present invention, when an ID and a password are given by signing up as a member, a private key according to the corresponding ID and password is generated by the block chain model 264, and the record and public key for this are generated through block mining. ). The basic principle of the blockchain is a system in which users' data is grouped in the form of blocks, and these blocks are shared with each other, where the smart contract 264a has a role of creating and storing blocks. When a voter logs in with an ID and password, it is verified that he is a voter by comparing the generated private key with the public key stored in the block. When a voter verified as a voter votes through a web page provided through the voting management unit 262, the value is stored in a block through block mining, thereby maintaining security.

As described above, in the present invention, the voter is authenticated by determining whether the voter's photo is a real photo or a fake photo through the artificial intelligence deep learning model 263 online, and the voter's voting information is converted into a block through the block chain model 264. By storing and securing, the security of voting can be increased.

The present invention is not limited to the above-described embodiments, and various modifications and variations within the scope of the technical idea of the present invention and the scope of the claims to be described below by those of ordinary skill in the technical field to which the present invention belongs. Of course this can be done.

100: client terminal 200: election management server
210: central control unit 220: input unit
230: output unit 240: communication unit
250: interface unit 260: election management engine
261: Member Management Department 262: Voting Management Department
263: artificial intelligence deep learning model 263a: learning data generation module
263b: Photo learning and identification module 264: Blockchain model
264a: Smart Contract

Claims (5)

It performs membership registration of voters who access through the client terminal 100, provides voting information to the client terminal 100 of the registered voter, and manages the voting information that the voters proceed through the client terminal 100. An election system including an election management server 200,
The election management server 200 includes a member management unit 261 for registering and managing a member by granting a member code after authenticating a voter accessing through the client terminal 100; A voting management unit 262 for providing a voting page to the client terminal 100 so that the registered voters can vote, and for collecting and managing the voting results conducted by the voters; Artificial intelligence that receives voter pictures from the client terminal 100 of the voter requesting membership registration through the member management unit 261, and determines whether the voter picture is a real person or a virtual person through artificial intelligence deep learning, which is machine learning. A deep learning model 263; Through the member management unit 261, a block chain account is created and provided to the registered voters, and the voter is verified through the generated block chain account, and the vote is made online through the voting management unit 262. Blockchain model 264 to secure by storing the proceeded voting information in a block; is provided,
The artificial intelligence deep learning model 263 includes a learning data generation module 263a that generates learning data that is a fake photo through Generative Adversarial Networks (GAN) so that learning to distinguish between a real photo and a fake photo can be performed, and the Blockchain-based, characterized in that a photo learning and identification module (263b) that determines whether or not the voter's photo is authentic is provided after learning fake photos and real photos provided through the learning material generation module (263a) through DenseNet. Election system.
delete delete The method of claim 1,
The blockchain model 264 is
Create an Ethereum account with the ID and password, which are member codes given according to the voter's membership registration, and record the generated Ethereum account information in the smart contract 264a,
When a voter logs in with a private key, ID and password created according to membership registration and Ethereum account, it is compared with the public key contained in the block to confirm that the voter is a voter, and when the voter votes, the value is saved in the block. Blockchain-based election system, characterized in that.
The method of claim 1 or 4,
The client terminal 100 is a block chain, characterized in that it is a personal terminal including any one of a smart phone, a personal digital assistant (PDA), a notebook, and a personal computer (PC) used by voters. Based election system.

Images