Designs and applies neural and conventional classification networks.
Systems: Windows 2000 and
options and utilities:
Sizing, feature selection, and KL transformation.
included: MLP, functional
link net, piecewise linear network (PLN), nearest neighbor classifier.
Self-organizing map, K-means clustering.
features: Many options
include real-time graphics. Training speed is considerably faster and error is
smaller than that available elsewhere. Feature selection (subsetting) is
Freeware limitations: MLP has 10 or fewer hidden units, PLN has 10 or fewer clusters, nearest neighbor classifier has 50 or fewer clusters.