Abstract

The work presents a fundamental theory of artificial learning that aims to explain existing optimization algorithms and support the development of new ones. It is based on a thermodynamic framework, using geometric and Hamiltonian formalisms to derive the evolution equations of learning models. These equations provide a physical grounding for the learning process and show that it can be understood as a thermodynamic process in which models evolve according to the information they perceive.