Development of an Electronic Voting System using Blockchain Technology and Deep Hybrid Learning

Abstract

Democratic people cannot function properly in today's sophisticated societies (where voting is a prominent issue) without electronic voting technologies. This study explores the use of hybrid learning algorithms for biometric authentication of voters, and blockchain technology for secure electronic voting. The thorough analysis includes a collection of more than 50,000 fingerprint samples using custom Convolutional Neural Network (CNN), VGG16, VGG19, Xception, and Inception. The algorithms are evaluated using F1-score, recall, accuracy, and precision. By combining Random Forest with a specially designed CNN, a novel hybrid learning algorithm is developed for authentication purposes. This blended model provides the best outcome in terms of accuracy (99.32%) and precision (99.32%). In addition, a web application was developed. This application integrates blockchain technology for electronic voting using Flask, HTML, and Solidity. By using blockchain, tampering and unauthorized access are prevented. It also ensures impartial voting and secure storage. The tabular presentation of the results provides a clear summary of each candidate's total votes.