Our attack type recognition based on machine learning which can at first produce lexems and, secondly. syntax constructions (patterns) by existing attacks. For

Our attack type recognition based on machine learning which can at first produce lexems and, secondly. syntax constructions (patterns) by existing attacks. For example, in the case of memcached injections (more details: https://www.blackhat.com/docs/us-14/materials/us-14-Novikov-The-New-Page-Of-Injections-Book-Memcached-Injections-WP.pdf https://www.blackhat.com/docs/us-14/materials/us-14-Novikov-...) we can train system to detect these attacks without regexps or new heuristic rules.