Widrow-Hoff LMS Adaline Demonstrator for Schools and Colleges

The Widrow-Hoff LMS (or ‘Adaline’) algorithm developed originally in 1960 is fundamental to the operation of countless signal processing machine learning systems in use even today. Bernard Widrow and Ted Hoff famously developed an Adaline machine demonstrator using basic analog off the shelf components to show how a ‘perceptron’ could be trained manually [1]. This paper details the design and development of a fully digital Adaline Least-Mean-Square algorithm demonstrator. The simplistic design presented here is completely open-source with all code, bill of materials and 3D casing models available at this Github Repository for download and reproduction. The demonstrator enables quick visualisation of the training and testing of a simple perceptron algorithm running on the inexpensive Arduino platform. The total costs of the device is estimated to be less than $60 and could be used in classrooms and colleges the world-over to demonstrate the seminal work of Widrow and Hoff [2] to a wide audience.