Attribute-based classification with label-embedding
Zeynep Akata, Florent Perronnin, Zaid Harchaoui, Schmid Cordelia
Attributes are an intermediate representation whose purpose is to enable parameter sharing between classes, a must when training data is scarce. We propose
to view attribute-based image classification as a label-embedding problem: each
class is embedded in the space of attribute vectors. We introduce a function which
measures the compatibility between an image and a label embedding. The parameters of this function are learned on a training set of labeled samples to ensure that,
given an image. the correct class has a higher compatibility than the incorrect ones.
Experimental results on two standard image classification datasets arc presented,
resp. on the Animals With Attributes and on Caltech—UCSD—Birds datasets.
Neural Information Processing Systems Conference and Workshops, Lake Tahoe, Nevada, USA, December 5-10, 2013.