An adaptive linear element or Adaline, proposed by Widrow (1959, 1960), is a simple perceptron-like system that accomplishes classification by modifying weights in such a way as to diminish the mean squared error (MSE) at every iteration. The architecture of the Adaline is the simplest of all neural networks. It is a simple processing element capable of sorting a set of input patterns into two categories. It has an ability to learn through a supervised learning process. Although the Adaline works quite well for many applications, it is restrictd to a linear problem space. The input patterns in the Adaline's training set must be linearly separable; otherwise, the Adaline will never categorize all of the training patterns correctly even when it reaches the low point of the error surface paraboloid. However, the Adaline is guaranteed to reach its minimum error state since there are no obstacles along the error surface (like local minima) to interfere with the training process.Training occurs by repeatedly presenting sets of data composed of input patterns and their desired outputs. Learning occurs as the Adaline minimized the number of errors it makes when sorting the patterns into their correct categories. Once trained, the Adaline can categorize new inputs according to the experience it gained.