Family of machine learning (see the diagram under machine learning), whose design was originally schematically inspired by the functioning of biological neurons and which, subsequently, came close to statistical methods.
The so-called formal neuron is designed as an automaton with a transfer function that transforms its inputs into outputs according to precise logical, arithmetic and symbolic rules. Assembled in a network, these formal neurons are able to quickly operate classifications and gradually learn to improve them.
This type of learning has been tested by tests on games (Go, video games). It is used for robotics, automated translation, etc.