A neural network can learn to organize the world it sees into concepts—just like we do

Karen Hao, MIT Technology Review, Jan 11, 2019
Commentary by Stephen Downes
files/images/gabriel-santiago-1598-unsplash2.jpg

I would use the word 'reflect' rather than 'represent' throughout, but this article nonetheless paints a fascinating picture of how a neural network organizes itself into concepts. "Asking the GAN to paint what it thought, the researchers found distinct neuron clusters that had learned to (paint) a tree, for example... In other words, it had managed to group tree pixels with tree pixels and door pixels with door pixels regardless of how these objects changed color from photo to photo in the training set." This is important because conceptualization is the basis for abstraction, which is the doorway to higher-level cognitive capacity.

Views: 2 today, 2 total (since January 1, 2017).[Direct Link] [Edit] [Spam]
Creative Commons License. gRSShopper

Copyright 2015 Stephen Downes ~ Contact: stephen@downes.ca
This page generated by gRSShopper.
Last Updated: Jan 11, 2019 07:50 a.m.