A major problem for networked systems is hostile trespassers. These trespassers may be users or software used by users to trespass a network. Trespasses may be in the form of authorized logon to a machine or acquisition of privileges and performance of actions beyond that have been authorized when in the case of an authorized user. To protect these resources, we need to detect the intrusions or intended intrusions, as absolute security may not be possible. Automatic detection of attacks  requires some kind of machine learning approach for the detection purposes. For machine learning to be successful, the machines have to be trained for known types of attacks. © 2018 IEEE.