Single Neuron

In the context of neural networks, a "neuron" is the basic unit of information processing. Each neuron receives some data as input (for example, image pixels, values from a table, etc.), processes this data, and passes the result on. To process the input data, a certain weight is assigned to each input of the neuron.

Note

All of these numbers are floating point numbers and mostly range from -1 to 1. Even if the original data is not represented as floating point numbers, we need to pre-process it so that we can use it in calculations.

w1, w2, w3 – weights of the Neuron.

x1, x2, x3 – input values of the Neuron.

The input the neuron receives is multiplied by the weights. Weights are special coefficients that determine how important each input is to the neuron. At the beginning of neural network training, these weights are set randomly, but then they are adjusted during training.

After each input has been multiplied by its corresponding weight, the results of all these multiplications are added together. This sum is then passed through a special function called the activation function. This function converts the sum to the neuron's output value. The activation function can be different, and the choice of a specific activation function depends on the tasks that are set for the neural network.

Neuron as Part of a Neural Network

The output value of the neuron is then used as input for the next layer of neurons, and so on, until the signal reaches the final layer of the neural network and the answer to the problem that the neural network solves is formed.

During the learning process, the weights of the neuron are adjusted in such a way as to minimize the error between the values predicted by the neural network and the real values.

This is done using the backpropagation algorithm which will be discussed later in the section. If the neural network makes a mistake, the weights of the neuron are changed in order to make the prediction more accurate in the future.