Classification of land cover based on hyperspectral data is very
challenging because typically tens of classes with uneven priors are
involved, the inputs are high dimensional, and there is often scarcity
of labeled data.  Several researchers have observed that it is often
preferable to decompose a multi-class problem into multiple two-class
problems, solve each such sub-problem using a suitable binary
classifier, and then combine the outputs of this collection of
classifiers in a suitable manner to get the answer to the original
multiclass problem. This approach is taken by the popular error
correcting output codes (ECOC) technique, as well by the binary
hierarchical classifier (BHC).  Classical techniques for dealing with
small sample sizes include regularization of covariance matrices and
feature reduction.  In this paper we address the twin problems of
small sample sizes and multi-class settings by proposing a feature
reduction scheme that adaptively adjusts to the amount of labeled data
avail able.  While exploiting the highly correlated nature of certain
adjacent hyperspectral bands.  This scheme can be used in conjunction
with ECOC and the BHC, as well as other approaches such as round-robin
classification that decompose a multi-class problem into a number of
two (meta)-class problems.  In particular, we develop the best-basis
binary hierarchical classifier (BB-BHC) and best basis ECOC (BB-ECOC)
families of models that are adapted to ``small sample size''
situations.  Currently, there are few studies the compare the efficacy
of different approaches to multi-class problems in general sett ings
as well as in the specific context of small sample sizes.  Our
experiments on two sets of remote sensing data show that both BB-BHC
and BB-ECOC methods are superior to their non-ad aptive versions when
faced with limited data, with the BB-BHC showing s slight edge in
terms of classification accuracy as well as interpretability.