Perceptron algorithm as equation

• If we call the input node i and the output node j we have:

?wji = ? (tj - aj) ai = ? ?jai

• ?wji is the weight change of the connection from node i to node j

• ai is the activation of node i, aj of node j

• tj is the target value for node j

• ?j is the error for node j

• The learning constant ? is typically chosen small (e.g., 0.1).