In this post, we will take a look at the difference between the artificial neuron and the real neurons present in the human mind. The neurons in the human brain look something like this: It contains some kinds of soma, dendrites, axon kind of things. Consider an example of an eye firing neurons in the brain. Whenever human eye is exposed to light in the environment, it triggers neurons in the retina to fire, which sends electrical signals into the brain vision centres by which our brain recognizes the object through which the light has been reflected our eyes.
On the other hand
An artificial neuron is a mathematical formula which gets an input x and multiplies it by weight(numerical value) set in the connection from one neuron to the other. This x input can be from external sensors or can be a pixel value from the camera, and the weights are adjusted in such a way during the training of a neural network so it can predict the output accurately when exposed to the actual data.
The single artificial neuron will do a dot product between w and x and then add a bias. The result is passed to an activation function that will add some non-linearity to it to get a discrete output.
So The mathematical function of our neuron (complete with an activation) is:
Where it takes x as an input, multiplies it with weight w, and adds a bias b.
If we want to think upon neurons in our computers, then we can say it’s just numbers stored in the variable or more precisely we can say the numbers stored in the computer memory to which we call weights and biases. They change their values according to the mathematical calculation performed on the input values while training them with the external data. Once trained, whenever a new input is given to the neuron, it performs a calculation based on the weight and biases it has stored previously and gives out the output result.
These artificial neurons collectively will form the neural network.
Check out the course below to get started with Embedded AI