Max Karpenkov
2015-01-30 09:21:24 UTC
Hello,
after the recent hacking of one of my clients, I have implemented
anti-Froud system, which blocks the user in real time, which is too much to
call the same direction for a certain period of time, such as module
hashlimit in iptables.
In addition to audit my server and my clients inside and outside on the
vulnerability, which I will do later, I started thinking about artificial
intelligence, which is great for pattern recognition.
The problem now is that the neural network input should be given some
value, then output it must determine that this hack or not.
You must define what data or parameters can be determined, it is likely to
hacking and then train the network on such examples.
So I have a question for you: what data to collect or monitor me to
determine on their basis, it is hacking ???
after the recent hacking of one of my clients, I have implemented
anti-Froud system, which blocks the user in real time, which is too much to
call the same direction for a certain period of time, such as module
hashlimit in iptables.
In addition to audit my server and my clients inside and outside on the
vulnerability, which I will do later, I started thinking about artificial
intelligence, which is great for pattern recognition.
The problem now is that the neural network input should be given some
value, then output it must determine that this hack or not.
You must define what data or parameters can be determined, it is likely to
hacking and then train the network on such examples.
So I have a question for you: what data to collect or monitor me to
determine on their basis, it is hacking ???
--
with respect, Max
with respect, Max