wiki:ML/SupervisedLearning

Supervised Learning

Supervised learning process is a process of continuous developing (and using for pattern recognition and classification) of some internal structures which reflects/represents some aspects of reality (of what is perceived). Reality is the supervisor.

The butterfly principle

Butterflies evolved eye-like dots on its wings because it this particular shared environment creatures has eyes. Butterflies know nothing about this, but creatures do have eyes and this is the cause of why the dots appeared on the wings of butterflies.

Small children believe that cars have eyes too. This is because of the common pattern which they have hard-wired in brain pattern recognition cortex because in this particular planet creatures have eyes.

So, the supervised learning is essentially the process of capturing such actual (not imaginary or mind-made) patterns of what is. An algorithm would do incomparably better if it would know that creatures have eye or cars have wheels.

Most of pattern-recognition algorithms does not aware of such high-level features, while all the animals do.

The principle of understanding Nature

Mother Nature does not count or compute. It does only pattern-matching on physical structures and electrical signals . Mother Nature does not use abstractions. It is very concrete. As the saying goes - God does not produce straight lines. Nature is blind. It does not observe.

Last modified 3 years ago Last modified on Nov 8, 2017, 10:41:36 AM
Note: See TracWiki for help on using the wiki.