Learning Image to Symbol Conversion

Malini Bhandaru, Bruce Draper and Victor Lesser

A common paradigm in object recognition is to extract symbolic and/or numeric features from an image as a preprocessing step for classification. The machine learning and pattern recognition communities have produced many techniques for classifying instances given such features. In contrast, learning to extract a distinguishing set of features that will lead to unambiguous instance classification has received comparatively little attention. We propose a learning paradigm that integrates feature extraction and classifier induction, exploiting their close interrelationship to give improved classification performance.

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.