Neural ANOVA Information Leak Shunt (NAILS)

The proliferation and sophistication of malicious code today is staggering. Within hours of the discovery of a new vulnerability, related exploits appear. Furthering the problem, virus protection methods are becoming less effective in detecting or responding to even known exploits due to the polymorphic and metaphoric nature of advanced malicious code. In many cases these exploits already exist and go undetected in already infective hosts, are opportunistically leaking information without detection, or they lay dormant awaiting activation. Once activated, these technologies look for opportunities to steal and leak vital information, and/or disrupt operations at the most critical time. Current solutions fall short in detecting malware information leakage behavior post infection. An effective solution needs to be scoped correctly to address the information leakage problem. It needs to work with both currently available as well as future distributed systems and that it need to be adaptable as well as extendable to new threats as they emerge. We shall produce a SELF verses OTHER approach to information leakage detection by leveraging and applying proven aspects of previous work including ontology creation, ANOVA analysis, neural network generation and automated network control, toward the specific goal of detecting and shunting information leakage.