Introduction - If you have any usage issues, please Google them yourself
The WH class is an implementation of the Widrow-Hoff.
The Widrow-Hoff (WH) algorithm, often called Least Mean Square (LMS), is an online-algorithm. The WH can be interpreted as a gradient descent procedure on the error space. In other words, the WH tries to move in the direction in which the loss is descreasing the fastest.