Machine Learning (ML) has become a central focus in cyber security research, providing intelligent methods to detect, classify, and mitigate sophisticated and evolving threats more effectively than traditional rule-based systems. Research papers in this domain explore supervised, unsupervised, and reinforcement learning techniques for applications such as intrusion detection, anomaly detection, malware classification, phishing detection, and spam filtering. Studies highlight how ML models can analyze vast amounts of network traffic, system logs, and user behavior data to identify hidden attack patterns and zero-day threats with high accuracy. Recent works emphasize deep learning for advanced feature extraction, adversarial machine learning to understand attacker evasion tactics, and explainable ML to improve transparency and trust in cyber defense systems. Additionally, hybrid models combining ML with blockchain, federated learning, and edge intelligence are being investigated to enhance security in distributed environments like IoT, cloud computing, and critical infrastructure. Despite promising results, challenges such as imbalanced datasets, adversarial attacks, and scalability remain active research areas, making ML for cyber security a rapidly evolving and impactful field.