NIDS are critical component in protecting the networks of an organization since they can detect various invasions. The constant emergence of complex threats and the traffic load in computer networks are growing exponentially, and traditional security solutions are not enough in this case. This paper focuses on the differential use of machine learning to improve the performance of NIDSs in particular. Namely, it focuses on the analysis of Support Vector Machines (SVM) and K-means clustering algorithms. SVM is a supervised learning techniques that is very efficient in classification of high dimensions and hence plays a very big role in differentiating normal and malicious traffic. K-means which is an unsupervised learning algorithm sorts behaviors similar to the network and defines sophisticated actions as valuable by singling out odd cases as defects. This work also highlights some of the issues that are currently facing NIDS such as high traffic rate, dynamically changing threats, false positives and false negatives and encrypted traffic. Furthermore, interaction between NIDS and other security layers like firewalls, IPS, and SIEM is considered in order to describe the efficient security tactic.