Datasets are not enough: Challenges in labeling network traffic

In contrast to previous surveys, the present work is not focused on reviewing datasets used in network security field. The fact that many of available public labeled represent behavior just for a particular time period. Given rate change malicious and serious challenge label, maintain these datasets, they become quickly obsolete. Therefore, this analysis current labeling methodologies applied network-based data. field security, process representative traffic dataset particularly challenging costly since very specialized knowledge required classify traces. Consequently, most methods are based automatic generation synthetic traces, which hides essential aspects necessary correct differentiation between normal behavior. Alternatively, few other incorporate non-experts users real with help visual statistical tools. However, after conducting an in-depth analysis, it seems all suffer from fundamental drawbacks regarding quality, volume, speed resulting dataset. This lack consistent continuously generating accurate validated methodology must be addressed by research community. Moreover, label condition helping acceptance novel detection approaches machine learning techniques.