%Aigaion2 BibTeX export from Idiap Publications
%Sunday 21 April 2024 12:28:39 AM

         author = {Tommasi, Tatiana and Orabona, Francesco and Caputo, Barbara},
       projects = {Idiap, EMMA},
          month = {12},
          title = {CLEF2008 Image Annotation Task: an SVM Confidence-Based Approach},
           type = {Idiap-RR},
         number = {Idiap-RR-77-2008},
           year = {2008},
    institution = {Idiap},
           note = {CLEF 2008 Working Notes},
       abstract = {This paper presents the algorithms and results of our participation to the medi-
cal image annotation task of ImageCLEFmed 2008. Our previous experience in the
same task in 2007 suggests that combining multiple cues with di{\"{\i}}¬€erent SVM-based
approaches is very e{\"{\i}}¬€ective in this domain. Moreover it points out that local features
are the most discriminative cues for the problem at hand. On these basis we decided
to integrate two di{\"{\i}}¬€erent local structural and textural descriptors. Cues are combined
through simple concatenation of the feature vectors and through the Multi-Cue Ker-
nel. The trickiest part of the challenge this year was annotating images coming mainly
from classes with only few examples in the training set. We tackled the problem on
two fronts: (1) we introduced a further integration strategy using SVM as an opinion
maker. It consists in combining the {\"{\i}}¬rst two opinions on the basis of a technique
to evaluate the con{\"{\i}}¬dence of the classi{\"{\i}}¬er{\^{a}}€™s decisions. This approach produces class
labels with {\^{a}}€{\oe}don{\^{a}}€™t know{\^{a}}€ wildcards opportunely placed; (2) we enriched the poorly
populated training classes adding virtual examples generated slightly modifying the
original images. We submitted several runs considering di{\"{\i}}¬€erent combination of the
proposed techniques. Our team was called {\^{a}}€{\oe}idiap{\^{a}}€. The run using jointly the low cue-
integration technique, the con{\"{\i}}¬dence-based opinion fusion and the virtual examples,
scored 74.92 ranking {\"{\i}}¬rst among all submissions.},
            pdf = {https://publications.idiap.ch/attachments/reports/2008/Tommasi_Idiap-RR-77-2008.pdf}