KDD lattest dataset

The NSL-KDD (Network Security Laboratory–Knowledge Discovery in Databases) dataset is a widely used benchmark dataset for evaluating network intrusion detection systems (IDS) and machine learning–based cybersecurity models. It is an improved version of the original KDD Cup 1999 dataset, which was developed for the Knowledge Discovery and Data Mining competition aimed at building predictive models capable of distinguishing between normal network traffic and malicious attacks. The NSL-KDD dataset was introduced to address several limitations of the KDD’99 dataset, such as redundant records and data imbalance that could bias learning algorithms. In the original dataset, many duplicate samples caused classifiers to overfit frequent patterns, leading to inaccurate evaluation results. NSL-KDD eliminates these redundant records and creates a more balanced and manageable dataset, enabling fairer comparisons between different machine learning and deep learning techniques used for intrusion detection. The dataset contains records of simulated network traffic collected from an intrusion detection environment. Each record represents a network connection and is described using 41 features, which include basic features of individual TCP connections, content features within packets, and traffic features derived over time windows. These features capture various characteristics of network behavior such as protocol type, service, duration, number of failed logins, and traffic statistics. In addition to these features, each record is labeled as either normal traffic or a specific type of attack. The attacks are grouped into four main categories: Denial of Service (DoS), Probe, Remote to Local (R2L), and User to Root (U2R). These categories represent different strategies used by attackers to disrupt services, gather information, or gain unauthorized access to systems. Overall, the NSL-KDD dataset remains one of the most commonly used datasets in cybersecurity research for evaluating machine learning algorithms and developing intelligent intrusion detection models due to its structured features, labeled attack categories, and improved data quality.