To express this categorically, the neuron is a placeholder function that takes in inputs, processes them by applying the function on the input, and produces the output. Any simple function can be put in the defined placeholder:

The function that's used in a neuron is generally called an activation function. In the human body, there are three types of neurons: sensory neurons, motor neurons, and interneurons. In the artificial world, the activation function would probably create the different capability and functionality of the neuron.

Here are a few commonly used activation functions:

• step
• sigmoid
• tanh
• ReLU-Rectified
• Linear Unit (used mostly in deep learning)

It is outside the scope of this book to delve into the details of each function. However, it will be good for you to understand these functions and their intricacies if you want to study neural networks further.