Artificial neurons are the basic units of an Artificial Neural Network (ANN). They are inspired by the biological neurons in the human brain and function similarly by receiving inputs, processing them, and generating an output.
- Each artificial neuron (also called a node or unit) receives one or more inputs.
- These inputs are connected with weights, just like biological synapses.
- The neuron calculates a weighted sum of the inputs and applies an activation function to determine the output.
- The structure of a single artificial neuron is shown in Figure 10.2 (refer textbook).
Simple Model of an Artificial Neuron:
The first mathematical model of a neuron was proposed by McCulloch and Pitts in 1943.
This model includes two main steps:
- The neuron receives weighted inputs from other neurons.
- It then processes the input through an activation (threshold) function to produce the output.
Summation Function:
- The inputs are multiplied with their respective weights and summed together.
- This is known as the Net input or Net sum, and it is calculated using:
Activation Function:
- An activation function is applied to the net sum.
- If the net sum is greater than or equal to a threshold (θ), the neuron fires (output = 1), otherwise does not fire (output = 0).
Thus, the output of a neuron is binary (either 0 or 1) depending on the threshold condition.
McCulloch and Pitts Neuron Model:
- This model represents only Boolean functions.
- Both inputs and outputs are binary (0 or 1).
- Examples:
- AND function: fires only when all inputs are 1.
- OR function: fires even if one input is 1.
Note: In this model, weights and threshold values are fixed and predefined.