<![CDATA[As digital technology evolves, cyberattacks are becoming more diverse and difficult to detect. Conventional detection methods are often incapable of recognizing new and sophisticated attack patterns. Therefore, machine learning techniques are starting to be widely used because of their ability to study data patterns and detect unusual or anomalous activities. This study aims to systematically examine the effectiveness of various machine learning techniques in detecting anomalies as an effort to prevent cyberattacks. The research was conducted using the Systematic Literature Review (SLR) method on 20 scientific articles from reputable journals published between 2020 and 2025. The articles were selected through a search, selection, and analysis process following PRISMA guidelines. The results of the study show that algorithms such as Random Forest and Decision Tree consistently provide accurate detection results, especially in network systems and the Internet of Things (IoT). Meanwhile, deep learning techniques such as CNN and LSTM show high performance in handling large and complex data. However, challenges are still found in terms of data imbalances, high computing requirements, and lack of model interpretability. The conclusions of this study show that machine learning techniques are very promising for anomaly detection in cybersecurity, but an adaptive and easy-to-explain approach is needed. Researchers are further advised to develop models that are more efficient, transparent, and able to adapt to evolving cyber threats.]]>
