Paper Details

Name: Sarala M, Muralidhara B L
Country: India

Year: 2025
Volume: Volume-12, Issue-2 (July-December)
Page Number: 67-74
DOI: https://doi.org/10.5281/zenodo.17097310

ABSTRACT
Forgery detection in educational credentials is a challenging task. Machine learning(ML)
techniques are highly used in fraud detection and spam detection. Our primary objective is to
detect counterfeit educational credentials using ML algorithms. In this research work, we
conducted an empirical study as follows: (i) extracted the features of those credentials using
the Gray Level Histogram Analysis (GLHA) metrics such as standard deviation, mean,
skewness, kurtosis, and entropy, (ii) Trained the ML models with extracted features using
different classification algorithms including Support Vector Machine(SVM), Logistic
Regression, Decision Tree, Naïve Bayes, Random Forest, and K-Nearest Neighbours(KNN).
(iii) Assessed the effectiveness of classification models using hyper-parameters. Random
Forest got 99.38% accuracy, and outperformed well than other algorithms. SVM, Decision
Tree, Logistic Regression, KNN, and Naïve Bayes got accuracies of 98.75%, 98.13%,
95.00%, 92.50%, and 90.00% respectively.
Keywords: machine learning, GLHA, counterfeit detection, supervised algorithms.