Anomaly Detection in Document Verification System Using Deeplearning in Hyperledger

Abstract

The educational industry is being integrated with technology and it has raised various challenges in maintaining the documents of academic details for each candidate in a longer period. The individual seeks higher education or recruiting for any industry then the verification of documents holding academic details is essential as well as to provide a reliable solution to avoid any academic fraud. Thus verification of certificates is done by blockchain in a completely decentralized transparent manner. Hyperledger Indy provides a digital identity that is rooted in blockchain for all records involved in the verification process. Records used for verification must not hold any anomalies to provide accurate data. Detecting anomalies is essential and increases the efficiency of the system. In this paper, an anomaly detection model is proposed for academic records data using deep learning by Keras on the top of the TensorFlow. Experimentation involves deep learning algorithms using the facial recognition model specifying some features like image size, facial weight, and width processing time all that are considered. The test results show that the proposed system model provides high performance, with high efficiency and low cost includes the minimum amount of processing time. By detecting, the anomalies using ML algorithms assure the trustworthiness of the documents involved with more transparent transactions.