The widespread use of Internet and computer networks
experienced in the past years has brought, with all its benefits,
another kind of threat: that of people using illicit
means to access, invade and attack computers. To understand
that the threat is real it is enough to look at the
statistics. Ten major government agencies, accounting for
98 % of the Federal budget had been compromised in the
past [9]. Recently, a massive, coordinated attack directed
at the major e-commerce sites was staged [22]. What is
worse, it is estimated than less than 4 % of these attacks
will ever be detected or reported. The issue is so pressing
that has prompted the administration to propose a new
Federal Intrusion Detection Network and a plan to put resources
into what is called Defensive Information Warfare.
Lately, systems have been trying to use data mining techniques
to meet the intrusion detection challenge. Data mining
can be defined as a set of tasks that enable users to look
for patterns in the data (good introductions to the topic
can be found [8], [10])