POLYNOMIAL SOFTMAX FUNCTIONS FOR PATTERN CLASSIFICATION
A. Tuerk and S.J. Young
February 2001
This report discusses softmax functions with polynomial and more general exponents and investigates the problem of optimising their parameters with regard to the cross-entropy error function. It is shown that, in this situation, the error surface is always convex and that in many cases there exists exactly one optimal set of parameters. In addition to these theoretical results, the practical estimation of the optimal parameter set is implemented using the Newton algorithm with line search and backtracking. In a number of test cases this algorithm is shown to converge reliably to the correct results.
If you have difficulty viewing files that end '.gz',
which are gzip compressed, then you may be able to find
tools to uncompress them at the gzip
web site.
If you have difficulty viewing files that are in PostScript, (ending
'.ps' or '.ps.gz'), then you may be able to
find tools to view them at
the gsview
web site.
We have attempted to provide automatically generated PDF copies of documents for which only PostScript versions have previously been available. These are clearly marked in the database - due to the nature of the automatic conversion process, they are likely to be badly aliased when viewed at default resolution on screen by acroread.