Kris,
Much thanks for the link. This is something that computer scientists have been searching for for a long time. For example, recognition of biological images. Imagine a tissue sample that is suspected of being malignant. However, a room full of histoligists, pathologists and cytologists sometimes cannot agree if 1) it is malignant and 2) what type of malignancy, an 3) what stage the malignancy is in (I have witnessed such problems when I was working on automated cell recognition in cervical cancer samples).
Many of have pondered how to achieve fusion of human and machine recognition. Humans are designed to be the the best at recognizing patterns but we cannot tell how we do it!
This is a major breakthrough in machine vision and has applications to such areas as target recognition (e.g. Where's Waldo?) and criminal justice.
Thanks again for the link - it certainly excited me to see this work.
MamaMoose