Finding needles in a haystack

Cyber attacks are becoming increasingly problematic as more sensitive information is being stored digitally. Clearly, cyber attackers are improving the sophistication of their attacks at a faster rate than software can be developed to protect sensitive information and to detect attacks. Neurala has developed a first generation intrusion detection system that is able to detect suspicious network intrusions by looking at sparse attack patterns that may be occuring across many access points and across various periods of time. In addition, our system incorporates online learning so that new attacks can be detected and used to train the system on that particular attack pattern. This helps the system to efficently and automatically adapt to new attack schemes.

Benefits: enables real time, automatic and efficient detection of network attacks so that faster action can be taken to remedy the situation, allowing network administrators to focus on the relevant subset of network traffic.

Demos: Neurala Network Defender is a prototype of Neurala Technology Platform applied to intrusion detection. The system autonomously learns to identify a specialized and widespread type of network intrusion, port scan, which consists of brief attacks distributed in time. After a brief training session, Neurala Network Defender is able to correctly identify the early instances of even very diluted attacks, while maintaining the number of false alarms at a minimum. Neurala Network Defender allows the network administrator to train several systems to recognize various types of custom threats, reducing the time spent in manually analyzing irrelevant log files, and increasing the overall safety of the network. Network warnings are generated by the system, and a tunable threshold can be learned by the network administrator to decide the "sensitivity" of the system to threats.

For more information, please click here.