Basically Neural Network is a technique that allows you to achieve many different goals, facial recognition, predictions, music composition, etc.
An NN contains virtual "neurons" that are organized into layers connected to each other. Neurons transmit the information and thus perform the necessary calculations to achieve the result.
The connections between neurons also have values associated with them, called weights.
The weights tell us how much information from one layer matters to the next layer. Neuron values and connection weights are essentially the free parameters of the network.
When training the network, we want to find those parameter values that minimize a certain function, called a "loss function". It's an optimization problem, we use a method called backpayment to solve it.
Backpayment means that if the NN gives us a result that is not what we want, we go back and change the weights of neurons and their connections.
In this way, the network learns from the identified failure.
Many wonder if NN is Profitable?
I say YES! ! !
Many wonder if NN is Accessible?
I say YES! ! !
