What is a Perceptron?

A ‘Perceptron’ is the building block, or single node, of a Neural Network inspired from the neurons that are found in the brain. The Perceptron takes in a set of inputs, performs some computation on the inputs (activation), and then produces an output. In order to ensure that a zero value for one of the features does not cause the entire output to become zero, bias terms are added in addition to the weights on each input feature. The basic Perceptron Model can only be used as a classification problem on classes that are linearly separable. 

The operation is thus: 

  1. A vector of 1…n inputs is passed to the algorithm
  2. Weightings are applied to each element of the input vector and a bias is passed along with
  3. A summation is performed – Weighted Sum
  4. The result is passed to an activation function. 
  5. The activation function then returns a classification decision.