ANDROID MALWARE DETECTION THROUGH INTELLIGENT PATTERN RECOGNITION USING DEEP LEARNING AND EQUILIBRIUM OPTIMIZATION

By precisely identifying malicious software using cutting-edge machine learning techniques, the project "Malware Analysis and Detection Using Machine Learning Algorithm" seeks to improve cyber-security measures. The project, which was created in Python, uses HTML, CSS, and JavaScript to create a responsive and interactive frontend interface and the Flask web framework for backend functions. The Extra Tree Classifier and Logistic Regression are two machine learning models that are essential to this project. With 97.42% training accuracy and 97.23% testing accuracy, the Extra Tree Classifier model performs quite well. with contrast, 94.84% training accuracy and 93.67% testing accuracy are attained with the Logistic Regression model. The TUNADROMD dataset, which includes 242 attributes and 4465 instances, is used to train and evaluate both models. The target classification attribute is used to differentiate between malware and goodware. Based on their significance and effect on the categorisation task, a subset of 23 attributes was chosen for the analysis. The goal of this calculated choice is to minimise computational complexity and maximise model performance. According to the project's findings, the Extra Tree Classifier provides a dependable solution for malware detection in practical applications by effectively differentiating between harmful and benign software. All things considered, this study shows how effective machine learning algorithms are in cyber-security, offering a reliable malware detection solution that can be included into many digital security infrastructures.