Projects
RANDOM DECISION FOREST APPROACH FOR MITIGATING SQL INJECTION ATTACKS
Preventing SQL injection attacks is possible by anticipating and neutralizing potentially harmful SQL statements before they are executed. Our method of choice is the use of Random Decision Forest, which has demonstrated outstanding results in mitigating SQLi attacks, with a 95% accuracy and 97% precision rate.
View Project
PREDICTION OF OCCURRENCE OF VENTRICULAR ARRHYTHMIA USING QRS FEATURES
The purpose of this research is to explore the possibility of predicting Ventricular Tachycardia Arrhythmia (VTA) early through the use of features extracted from QRS complexes. We have analyzed two characteristics, the QRS signed area and R-peak amplitude, and evaluated the VTA prediction accuracy using various machine learning techniques.
View Project
WEDDING IMAGE CLASSIFICATION
This project aims to classify wedding images based on different regions of the world using a new approach. It utilizes UNet and a modified VGG network for segmentation and classification, with the VGG network being modified by adding Inception blocks and Residual networks. Transfer learning was used on the Cifar 10 dataset before fine-tuning for the wedding image dataset and various classifiers were added to the model, with the artificial neural network being the best choice for most classifications. However, for some classifications where the dataset was smaller, traditional machine learning models like SVM and Random Forest were used. The testing data showed an accuracy of 85.96% for Indian vs International images and 83.64% for North India vs South India classification with Random Forest as the best classifier. This project showcases the potential to use image information for classification and can be extended to other fields of research.