Learning Local Invariant Mahalanobis Distances
Published on Sep 27, 20151724 Views
For many tasks and data types, there are natural transformations to which the data should be invariant or insensitive. For instance, in visual recognition, natural images should be insensitive to rota