Skip to content

Page393

Artificial Intelligence

Computers compute: they do exactly what they are told. The term “computer” was first used in 1613 to describe a person who added numbers. Artificial Intelligence is the science of programming electronic computers to “think” more intelligently, sometimes mimicking the ability of mammal brains.

Expert Systems

Expert systems consist of two main components. The first is a knowledge base that consists of “if/then” statements. These statements contain rules that the expert system uses to make decisions. The second component is an inference engine that follows the tree formed by the knowledge base, and fires a rule when there is a match.

Here is a sample “the Internet is down” Expert System, which may be used by a help desk when a user calls to complain that they cannot reach the Internet:

  1. If your computer is turned on
    a. Else: turn your computer on
  2. Then if your monitor is turned on
    a. Else: turn your monitor on
  3. Then if your OS is booted and you can open a cmd.exe prompt
    a. Else: repair OS
  4. Then if you can ping 127.0.0.1
    a. Else: check network interface configuration
  5. Then if you can ping the local gateway
    a. Else: check local network connection
  6. Then if you can ping Internet address 192.0.2.187
    a. Else: check gateway connectivity
  7. Then if you can ping syngress.com
    a. Else: check DNS

Forward chaining begins with no premise (“Is the computer turned on” in our previous example), and works forward to determine a solution. Backward chaining begins with a premise (“Maybe DNS is broken”), and works backwards.

The integrity of the knowledge base is critical. The entire knowledge base should form a logical tree, beginning with a trunk (“Is the computer turned on” in our previous example). The knowledge base should then branch out. The inference engine follows the tree, branching or firing as if/then statements are answered.

There should be no circular rules; an example of a circular rule using our previous example: “If your computer is turned on, then if your monitor is turned on, then if your OS is booted and you can open a cmd.exe prompt, then if your computer is turned on … .” There should also be no unreferenced rules (branches that do not connect to the knowledge base tree).

Artificial Neural Networks

Artificial Neural Networks (ANNs) simulate neural networks found in humans and animals. The human brain’s neural network has 100 billion neurons, interconnected by thousands or more synapses each. Each neuron may fire based on synaptic input. This multilayer neural network is capable of making a single decision based on thousands or more inputs.