Machine learning of the well known things

TL;DR

This paper explores whether well-known facts can be effectively represented using the specific function forms of current machine learning models, proposing a systematic reformulation approach to understand the limits and capabilities of ML.

Contribution

It introduces elementary examples demonstrating the potential to represent known knowledge with ML-compatible functions and advocates for a systematic reformulation of existing knowledge in ML terms.

Findings

Elementary examples show known facts can be represented in ML-compatible forms

Suggests reformulating existing knowledge systematically for ML integration

Highlights implications for scientific and epistemological understanding

Abstract

Machine learning (ML) in its current form implies that an answer to any problem can be well approximated by a function of a very peculiar form: a specially adjusted iteration of Heavyside theta-functions. It is natural to ask if the answers to the questions, which we already know, can be naturally represented in this form. We provide elementary, still non-evident examples that this is indeed possible, and suggest to look for a systematic reformulation of existing knowledge in a ML-consistent way. Success or a failure of these attempts can shed light on a variety of problems, both scientific and epistemological.