What Is Perceptron?

The Building Block Of Neural Networks

A perceptron is a type of artificial neuron that mimics the behavior of a biological neuron.

In machine learning, Perceptron is an algorithm for supervised learning used primarily for binary classification tasks.
It is also known as one of the simplest forms of neural networks and the foundation for more complex neural network architectures.

Design of Perceptron:

The perception consists of 4 parts.

1. Input Nodes :
   This is the primary component of perception, It accepts the features of the data for further processing. Each input is represented as x1,x2,…,xn, where n is the number of features.
2. Weights and Bias:
   - Weight is associated with Each input feature, here weight represents the strength of the input units, so the higher the weight the more it will affect the output.
   - Bias is the line of intercept in a linear equation and it allows the model to make adjustments that are independent from the input.
3. Weighted Sum or Summation Function:
   This function basically performs the summation (using the dot product) of the inputs with their respective weights and produces a weighted sum.
4. Activation Function:
   This function is used to put the value of the summation received from an earlier stage in a certain range or required values like (0, 1) or (-1, 1).
   There are many activation functions like Step Function, Sigmoid, Tanh, Softmax, ReLU, Leaky ReLU, etc. We have to choose the function that is suitable to our problem.
   For Example: Unit Step Activation Function.

Figure: Unit Step Activation Function

How Does a Perceptron Work?

The perceptron first receives the values of features, weights, and bias then it computes a weighted sum by performing a dot product and finally applying the activation function to this sum to get output.

Mathematically, this can be represented as:

Here, f is the activation function, w is the wait, x is the input feature and b is the bias.

Perceptron is usually used to classify the data into two parts. Therefore, it is also known as a Binary Classifier.

The perceptron may seem simple, but it is a fundamental concept in the field of deep learning. Understanding how it works provides a strong foundation for exploring more advanced topics in AI. As you delve deeper into neural networks, you may notice that even the most sophisticated models are built upon the perceptron.