Neural Networks: Computer Recognition and Understanding
Computers do not 'think', they act upon basic instructions provided by programmers. Nor can computers cannot learn - their capabilities are again limited by their programming. Neural networks are a method of allowing computers to simulate thinking by harnessing the computer's ability to perform millions of calculations per second. By performing many comparisons in the blink of an eye, a computer can be made to seem to recognize complicated characters such as those found in the alphabet.

This program demonstrates an application of a simple neural network by recognizing capital letters. The neural network in this program has already been 'taught' with a preset idea of what the alphabet looks like, and it can make reasonable guesses as to what letters are drawn by the user. In the picture below, the program has correctly identified the shape the user drew as the letter 'B'.