Neural networks have become an increasingly popular solution for network detection systems (NIDS). Their capability of learning complex and behaviors make them a suitable solution for differentiating normal traffic and network attacks. However, a drawback of neural is the amount of resources needed to train them. Many network gateways routers devices, which could potentially host an NIDS, simply do not have memory or processing power to train and sometimes even execute such models. importantly, the existing neural network solutions are trained in a manner. Meaning that an expert must label the network traffic and the model manually from time to time. In this paper, we present Kitsune: a plug and play NIDS which can learn to attacks on the local network, without supervision, and in an efficient manner. Kitsune's core algorithm (KitNET) uses an ensemble of neural called autoencoders to collectively differentiate between normal and traffic patterns. KitNET is supported by a feature extraction which efficiently tracks the patterns of every network channel. Our show that Kitsune can detect various attacks with a performance to offline anomaly detectors, even on a Raspberry PI. This that Kitsune can be a practical and economic NIDS.
No takes yet. Share an insight, caveat, or question.
Mirsky et al. (2018) studied this question.