Student Cheating Detection of Academic Integrity Violations Based on Lightweight Deep Convolutional Neural Network and Discrete Fourier Transform

From past few years, cheating in exam is increasing world-wide, As the academic institutions and universities are discussing for detecting the cheating in exam hall and need to maintain discipline by providing good atmosphere. As such exam monitoring increases more attention. Academic cheating undermines prompting institutions to adopt detection methods, such as automated plagiarism detection, proctoring software of student behavior, However, False positives and misidentification of legitimate behavior as increasing the cheating that leads to inaccurate accusations, causing stress and unfair consequences for innocent students, undermining trust in the education system. To overcome these drawbacks proposed a classifier of soft voting based on the Lightweight Deep Convolutional Neural Network (LDCNN) and Discrete Fourier Transform (DFT) by collecting input from the 7wiseup behavior dataset. Then the preprocessing is done by normalizing data, handling the data, determining missing values and duplicating the input data. At last classification is performed by the LDCNN-DFT. From the results, it clearly shows that the proposed LDCNN-DFT obtained higher values in the context of precision, recall and accuracy when compared to existing Support vector machine (SVM).