Material Detail

Linear classifiers, that is, classifiers based on linear discriminant functions, are formally introduced first. Then, a well-known learning technique, the so-called Perceptron algorithm, is described for general multiclass classification learning. A simple yet very instructive working example is provided with detailed calculations. The example is followed by a brief discussion on what one can expect from Perceptron's convergence and quality of the solution. The presentation ends by citing two basic redferences on the Perceptron algorithm. The training ojectives of the learning object are: 1) To apply the Perceptron algorithm to a classification task; and 2) To describe the Perceptron algorithm's behaviour as a function of its parameters.

Disciplines:

• Science and Technology  / Computer Science

More...