Efficient intrusion detection method based on Conditional Random Fields

With the rapid advancement in the network and communication technologies, intrusion detection becomes a very troublesome problem. An important method for intrusion detection is to model the dynamic behavior of network user based on Hidden Markov Models (HMM), but the HMM model requires strong independence assumptions between the observation sequences of the behaviors of network user, in practice the behaviors of users in networks is dependent. So, this motivates us to use Conditional Random Fields (CRFs) to model the behaviors of users because this model has no assumptions on the dependencies among observation sequences.