Improving the accuracy of contextual action recognition to detect cheating during exam using local outlier factor in comparison with support vector machine

• Main Authors: Suresh, H., Latha, S. Puspha
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Activity recognition; Classification; Data analysis; Datasets; Monitoring; Sequences; Support vector machines
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0233236

With regard to the Support Vector Machine, the objective is to improve the accuracy of contextual action recognition by identifying instances of cheating in examination monitoring through the utilisation of Novel Local Outlier Factor methods. This will be accomplished in comparison to the Support Vector Machine. The Constituents and the Methods involved: The dataset, which includes a total of 1650 photos for each category and a total of 37 video sequences, is subjected to both the Local Outlier Factor (N=10) and the Support Vector Machine (N=10). Both of these models are applied to the dataset. Three rows and one hundred and forty-four columns were used to construct each image for the aim of implementing a categorization system for the examination monitoring. Using Clincalc, the size of the sample is decided by setting the G-power to 0.8 and the alpha to 0.05. This is done in order to optimise the results. The degree of accuracy with which the examination management system operates is one of the criteria that is used to measure its effectiveness. The results show that the mean accuracy of the categorization of test monitoring using Local Outlier Factor is strong, coming in at 91.62 percent. This is in comparison to the Support Vector Machine, which has a mean accuracy of 100 percent (90.50 percent ). A result of 0.664 is obtained from the independent sample T-test for accuracy, which indicates that there is a difference between the two approaches that is statistically insignificant. This is demonstrated by the fact that the p-value for accuracy is more than 0.05. When the accuracy of the Novel Local Outlier Factor is compared to the accuracy of the Support Vector Machine, the conclusion that can be drawn is that the former is superior to the latter.