The increasing complexity of cyber threats has made traditional security approaches insufficient, driving the need for intelligent and adaptable solutions. With the adoption of machine learning, network security can be made stronger, since it learns from data and finds patterns. This paper considers new and enhanced approaches to supervised and unsupervised learning methods for identifying different types of network attacks such as DDoS, unauthorized access, malware, and other new attacks. The strengths and limitations Different algorithms are discussed and also explored hybrid models that combine the best aspects of both supervised and unsupervised methods. This survey aims to provide a comprehensive literature review for researchers and practitioners interested in the state of the art and future trends in machine learning for network attack detection.