In my first and second articles about neural networks, I was working with perceptrons, a single-layer neural network. And even though our AI was able to recognize simple patterns, it wasn’t possible to use it, for example, for object recognition on images. That’s why today we’ll talk about hidden layers and will try to upgrade perceptrons to the multilayer neural network.
Hidden Layers
Why do we need hidden layers? Perceptrons recognize simple patterns, and maybe if we add more learning iteration, they might learn how to recognize more complex patterns?