Abstract -.The article discusses and researches how machine learning techniques can assist in defending zero-day cyber-attacks, which are of the greatest concern in cyber defense. The research targets various machine learning algorithms like gradient boosting classifiers, random forests, decision trees, and support vector machines (SVM). The research analyzes how effective these algorithms are in detecting and blocking zero-day attacks. For this, we preprocess a dataset with various network features for processing so that categorical variables are treated correctly. We test and train the chosen algorithms on this dataset. According to the data, random forest performs better among all the algorithms in detection rates and accuracy. This is because random forest's capability to detect complex patterns associated with zero-day attacks is promoted by its ongoing learning from poor models. The findings show how machine learning can help enhance cybersecurity defense against emerging threats such as zero-day attacks. The CSE-CIC-IDS2018 Dataset was employed in the study's implementation and evaluation. (E-Journal UPI1) Key Words: AI, Zero-Day Threats, Prediction, Mitigation, Anomaly Detection, Threat Intelligence, Machine Learning, Behavior Analysis, NLP, Cybersecurity, Automated Response, Reverse Engineering, Vulnerability Detection, Proactive Defense, Adversarial AI.
Jotangiyai et al. (Sat,) studied this question.