Paper Contents
Abstract
The widespread use of Android smartphones has resulted in a proportional rise in harmful software (malware) specifically designed to exploit this platform. Conventional strategies for detecting malware, which depend on signatures, face difficulties in keeping up with the quickly changing environment of mobile threats. This thesis investigates the utilization of artificial intelligence (AI) and dimensionality reduction approaches to improve the detection of harmful software on Android devices. Artificial Intelligence (AI), namely machine learning (ML) and deep learning (DL) algorithms, presents a hopeful answer because of its capacity to acquire knowledge and adjust based on extensive datasets. These algorithms have the ability to detect patterns and abnormalities that indicate harmful behaviour, even in the presence of novel and previously unknown malware variations. Nevertheless, the efficacy of these models is frequently hindered by the elevated dimensionality of the feature space, which can result in over fitting and escalated computing expenses.To address these challenges, this study integrates dimensionality reduction techniques with artificial intelligence systems to improve the effectiveness of identifying malware. Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and t-Distributed Stochastic Neighbour Embedding (t-SNE) are employed to reduce the dimensionality of the feature space while retaining the crucial information. By using this strategy, the models become more efficient and less prone to overfitting. The project entails gathering and preprocessing an extensive dataset of both benign and malicious Android applications. Feature extraction is performed to obtain significant characteristics from these applications, which are subsequently submitted to dimensionality reduction. Afterwards, different AI models such as Support Vector Machines (SVM), Random Forests (RF), and Convolutional Neural Networks (CNN) are trained and assessed using the reduced feature set.The results indicate that combining AI with decreasing dimensionality significantly enhances the precision and effectiveness of identifying malware. The models exhibit enhanced generalization abilities and accelerated processing speeds, rendering them suitable for real-time implementation in resource-limited settings like mobile devices. This thesis enhances the field of cyber security by introducing a strong framework for detecting Android malware. The system utilizes the advantages of artificial intelligence and dimensionality reduction. The findings highlight the capacity of these technologies to offer more flexible and effective security solutions, facilitating their wider implementation in protecting mobile ecosystems.
Copyright
Copyright © 2025 SURESH KUMAR. This is an open access article distributed under the Creative Commons Attribution License.
