CNN-FSPM-Based Fingerprint Indexing and Matching for Detecting, Predicting, and Preventing Cheating in Online Examinations

Abstract

This paper presents a comprehensive approach to the detection and prevention of cheating in online exams using AI. The authors employ various technical solutions to monitor proctors throughout all stages of the exam: before, during, and after. To address the formulations and ensure the continuous expansion of the database, the authors rely on a fast convolutional neural network (CNN) that utilizes a full-scope pattern matching algorithm (FSPM) to enhance the ability to match fingerprint formats using descriptive network cryptography. The authors anticipate reliable matching across the complete fingerprint image set through the utilization of deep-learning (DL) symbols. Furthermore, the authors demonstrate that solving image-matching problems does not necessitate tool training data, which is typically required for such problems. Thanks to the highly parallelizable nature of these tasks, the authors provide an efficient method with minimal computational cost during test time to detect cheating during some exams at School of Computer Science at Danang University of Technology (SCS-DTU) University, Vietnam.