Skip to content

Welcome!

This website is dedicated to sharing my ongoing theoretical research in machine learning. It is intended as a living repository of ideas, evolving over time as I expand on my work.

Here, you will find a series of individual notes, preprints and publications that explore various aspects of modern machine learning theory under a physics perspective. Each article is meant to stand on its own but collectively, they aim to grow into a cohesive, evolving “living book” of machine learning insights.

Topics

  • Learning Dynamics

    The work presents a fundamental theory of artificial learning as a thermodynamic process in which models evolve according to the information they perceive.

Contribute

Contributions to this repository are warmly welcome, whether in the form of suggestions, corrections, or extensions to the research, or improvements of the website itself. All contributions will be acknowledged here to recognize your support in advancing this evolving body of work.

If you have questions or suggestions, please feel free to email me at eric.hermosis@gmail.com. Your feedback is highly valued and can help improve the clarity and depth of the work.

Thank you for visiting, and I hope you find these ideas thought-provoking and useful in understanding the theoretical aspects of deep learning.