Nowadays the Internet provides the infrastructure for offering a large number of services.
Despite its almost ubiquitous presence, the Internet is not completely understood, with network providers and customers alike having a hard time in breaching its complexity to extract useful information.
Given the limits of active network measurement, a viable alternative is an approach coupling passive measurement and Big Data techniques.
The seminar will initially provide an overview of the issues related to network measurement, then some examples of machine learning applications will provide a hint on the problems that can be faced with such methods.
This talk is based on the work by the TNG and DBDMG groups of Politecnico di Torino, presented during the Summer School "Data Gathering and Processing: From Cyber Space to Physical World", held in Como last September.