M.Harshavardhan, N.Madhu, M.Chanikya, M. Sai Hrushikesh, Mrs. Sandhya Rani,

Abstract

Using the Deep Fake dataset, this research will evaluate Convolutional Neural Networks (CNNs) with Decision Tree classifiers for detecting bogus photos. Tools and Procedures: In this study, the Deep Fake dataset is used. There are 20,000 training photos, 4,000 photographs for development, and 4,500 images for testing. The dataset is then divided in half. To increase accuracy, half of the dataset is trained using decision tree and convolutional neural network techniques for 10 iterations. By adjusting the following values in GPower 3.1: 80% power, 95% confidence interval, alpha = 0.05, and beta = 0.2, we may estimate that there are 40 participants in each of the training set classes's groups. Regarding accuracy, Using a significance threshold of 0.0293' (p<0.05), the data shows that the Decision Tree achieves an accuracy of 80.13% & the Convolutional Neural Network 82.22%. Because the two datasets are so different, it's clear that the Convolutional Neural Network is better at