Yeah... all the machine learning domains have problems that get a lot easier or harder depending on the number of items to be identified and how much different examples of each item tend to vary. A 1993-era OCR could probably recognize letters and numbers in a narrow range of typical fonts. The Nielsen face-recognition box only had to distinguish between a few different family members and could spend minutes analyzing any given image.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
no subject
Date: 2010-06-09 04:38 am (UTC)