Discussion of "Selecting Relevant Information and Delaying Irrelevant Data for Objects Recognition"

Eric Neufield

I argue that absolute relevance is really a comparative relevance that measures the increased probability the simple token is meaningful given the measurements. Each token is then multiplied by a proximity measure, which measures its closeness to a hypothesized region of attention, capturing the notion that a token is more relevant if it is near some object.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.