Probabilistic topic models have been applied to image classification and permit to obtain good results. However, these methods assumed that all topics have an equal contribution to classification. We propose a weight learning approach for identifying the discriminative power of each topic. The weights are employed to define the similarity distance for the subsequent classifier, e.g. KNN or SVM. Experiments show that the proposed method performs effectively for image classification. © 2011 Springer-Verlag.
CITATION STYLE
Liu, Y., & Caselles, V. (2011). Image classification based on weighted topics. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7063 LNCS, pp. 268–275). https://doi.org/10.1007/978-3-642-24958-7_31
Mendeley helps you to discover research relevant for your work.