Machine Learning
Evolutionary Machine Learning
Perhaps, machine learning is simpler than it seems ... if we follow nature of course.
Lets imagine that the behavior of any organism (in a given state space) can be represented by the product of 3 matrices: L,D and M. By applying this product to the current state, the organism would transition to a next position P in the space.
P(s) = L*D*M (s)
The next position migh be more or less stable, but this will come later.
M (Mendel operator) is the first one applied, followd by D (Darwin operator). Finally L (Lamarck operator) is applied. Unlike the two others, L is functional, meaning that the elements in the matrix are functions.
The structure of M is fairly simple, it is indeed a block diagonal matrix, where every block represents a chromosome
The goal is to make changes to those blocks to gradually approach a matrix with “good“ properties, like a Jordan matrix for instance.
But first things first, visit my openlambda, look at the code under ml(mendel.js/mendel.py) and run some tests.
