All You Need To Know About Adaptive Linear Neuron

Original article was published on Deep Learning on Medium

All You Need To Know About Adaptive Linear Neuron

Adaptive linear element or Adaptive linear Neuron (ADALINE) is a single layer artificial neural network, developed by Widrow and Hoff in 1960. Multiple layers of ADALINE is known as multi- ADALINE (MADALINE). ADALINE is made of memristors (memory plus resistors) which are nanoelectronics circuits capable of performing logical operations.

ADALINE is an improvement of the perceptron model and utilized for binary classification problems. Each node of the ADALINE performs an affine transform of the input set to the neuron. ADALINE uses threshold functions, and weights are learned using a stochastic gradient descent algorithm. The significant difference between ADALINE and perceptron is that ADALINE is learned based on the error computed at the output without applying any activation function. It differentiates ADALINE from perceptron.