CMP 450 Networking and Machine Learning Mid Term

Study Notes

AI and Machine Learning in Network Intrusion Detection

Intelligence can be defined as the ability to make the correct decision from a list of potential decisions given a set of inputs.
The Turing test was designed to determine if a computer could be considered intelligent by comparing its responses to a human's through a terminal.
Games were used as early test-beds for AI methods because they provided bounded problems that could demonstrate the power of AI algorithms.
Arthur Samuel set the bar for learning programs in the 1950s by having two copies of his program play against each other to learn from increasingly complex strategies.
The LISP language was the first language developed specifically for AI, and Prolog was developed in the 1970s to support logic programming.
Strong AI aims to match or exceed human general intelligence, while weak AI focuses on providing bounded intelligence in specific areas.
The 1969 monograph "Perceptrons" by Marvin Minsky and Seymour Papert is often credited with leading to AI's Winter.
Scruffy and Neat approaches refer to Weak AI and Strong AI, respectively.
After AI's Winter, the re-emergence of AI focused on hidden applications rather than openly promoting AI solutions.
The systems approach explores AI from the perspective of its application within an environment, considering inputs, outputs, and interactions with the environment.
Intrusion Detection Systems (IDS) aim to protect networks from malware by monitoring traffic and network devices for anomalous activities.
Host-based IDS (HIDS) and network-based IDS (NIDS) are two types of IDS that offer different advantages based on network size and configuration.