A Writer's Notebook, Day Nine-Hundred-And-Seventy-Eight

In exploring deep learning and neural network programming, certain interesting things begin to become apparent.  As I have studied the ways these systems work, it becomes clear that what the computer is being programmed to do is to remove information, to filter out data.  That is, the computer is learning by deleting irrelevant features.  All of the information is present from the start, but the noise, the irrelevancies, must be extracted to leave behind what is meaningful.  

This seems, to me, a very deep realization about knowledge, but I am not certain how to explain what it seems to be suggesting.  It points towards the idea that learning is not acquisition, but is, in some sense, the opposite.  I can't quite piece it together yet, but I am sure there is more here.  I do not know if what I have said here will sound like anything more than repetitious nonsense, but it is that way, sometimes, when a new understanding is beginning.

Comments

Popular posts from this blog

A Writer's Notebook, Day Two-Hundred-And-Fifty

Le Guin, Steering The Craft, Chapter Five: Adjectives and Adverbs (Exercise Five, Chastity)

A Writer's Notebook, Two-Thousand-And-Fifty-Nine