Dr. Manuel Masías Vergara and Dr. Daniel Gavilanes Aray, researchers at the Universidad Internacional Iberoamericana (International Iberoamerican University, UNIB), are participating in a study that has developed a stacked machine learning model that improves the identification of cyberattacks with high precision and robustness.
The Internet of Things (IoT) has transformed the way people interact with technology, allowing everyday devices to be integrated into a vast interconnected global network. However, this advance also carries significant risks, including botnets, which are networks of infected devices used to execute large-scale cyberattacks.
Botnets are made up of compromised devices—ranging from computers to home security cameras—and are remotely controlled by attackers to carry out illegal activities such as denial-of-service (DDoS) attacks, data theft, or mass spam mailings.
The diversity of devices that make up the IoT ecosystem, with differences in protocols and capabilities, makes protecting them an increasing challenge for cybersecurity. Traditional techniques are insufficient, especially in a scenario where cybercriminals are incorporating artificial intelligence (AI) to refine their methods.
In this context, an innovative approach based on machine learning (ML) is proposed to anticipate and neutralize these threats, marking an important step in the digital protection of smart networks.
The study, in which the researchers are participating, proposes a model called KSDRM, which combines different machine learning algorithms—including k-Nearest Neighbors (KNN), support vector machines (SVM), decision trees (DT), random forests (RF), and multilayer perceptrons (MLP). To integrate the results of these classifiers and improve accuracy, the model uses logistic regression as a meta-classifier.
This approach not only seeks to optimize detection capabilities, but also to reduce overfitting and increase interpretability, compared to approaches based exclusively on deep neural networks.
Results that reinforce digital security
The results obtained confirm the effectiveness of the proposed model. The KSDRM achieved an accuracy of 97.94% in tests, surpassing the best individual classifier (RF), which obtained 97.34%. In addition to its accuracy, the system was shown to reduce processing time, making it a viable alternative for real-time applications where immediacy is critical.
This advance not only represents a step forward in the defense against botnets, but also lays the foundation for future research in cybersecurity applied to the IoT. In a hyperconnected world, where every device can become a vulnerable point, AI emerges as the most effective strategic ally to ensure digital protection.
If you want to learn more about this study, click here.
To read more research, check out the UNIB repository.
The International Ibero-American University (UNIB) offers a Master's Degree in Strategic Management with a Specialization in Information Technology. This master's degree is designed for professionals who seek to lead the integration of information technology (IT) into business strategies. It provides the necessary tools to align ICT with organizational objectives, fostering innovation and competitiveness in an ever-evolving business environment.