Copyright © 2025 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{182076,
        author = {P. Usha Manikyam and G.V.N. Kishore and Sheik. Faridha Akather},
        title = {ENHANCING CYBER DETECTION WITH MACHINE LEARNING: A STACKING-BASED APPROACH FOR LARGE AND IMBALANCED DATASETS},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {12},
        number = {2},
        pages = {892-895},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=182076},
        abstract = {The present world has become dependent on cyberspace based on every aspect of our daily life. The usage of cyberspace is rising with each passing day. The world is spending more time on the internet than before. As a result, the risk of cyber threats and cyberattacks is increased. The term cyber threat refers to the illegal activity performed using the internet. Cyber criminals are changing their techniques over time to pass through the wall of protection. Conventional techniques are not capable of detecting zero-day attacks and sophisticated attacks. So that this is to investigate the use of Machine Learning techniques to improve cybersecurity measures, with a particular emphasis on threat detection, prevention, and response. To begin, an examination of the principles of Machine Learning and the importance of this bill to cybersecurity is presented. When it comes to recognising and mitigating cyber threats, a number of different Machine Learning methodologies, including deep learning, signature-based detection, and anomaly detection, are evaluated in terms of how effective they are. Machine Learning based behaviour analysis within the IDS has considerable potential for detecting dynamic cyber threats. I didn't find abnormalities, good identities malicious conduct within the network. However, as the number of data points grows, dimension reduction becomes an increasingly difficult task when training Machine Learning models. At present, we are going to introduce an ML-based network intrusion detection model that uses Random Oversampling (RO) to address data imbalance and stacking feature embedding based on clustering results, as well as Principal Component Analysis(PCA) for dimension reduction, and is specifically designed for large and imbalanced datasets. 
This model preference is carefully evaluated   using three cutting mark datasets: UNSW-NB15, CIC-IDS-2017, and CIC-IDS-2018. On the UNSW-NB15 dataset, the trial shows the RF and ET model your accuracy rates of 99.59% and 99.95%, respectively. Furthermore, using the CIC-IDS-2017 dataset, DT, RF, and ET obtained 99.99% accuracy, while DT and RF models obtained 99.94% accuracy on CIC-IDS-2018. This performance result continuously outperforms the state of our indicating significant process in the field of network impression detection. This achievement demonstrates the efficiency of the suggested methodology, which can be used particularly to accurately monitor and identify network traffic intrusions, thereby blocking possible threats.},
        keywords = {Machine Learning, Random Oversampling, Cyber Threats, Principal Component Analysis, Dynamic Cyber Threats, Recognising, Mitigating, Network, Dimension, Reduction, Recognising.},
        month = {July},
        }