We develop algorithms for use with a recently developed VLSI architecture
for pattern classification. This architecture is based on the evaluation
of class discriminant functions without crossterms. We refer to
classifiers that use such discriminant functions as Sigma classifiers.
Tradeoffs in architecture design have been required to allow high
throughput and consequently the decision regions in feature space which
can be generated using this architecture form a restricted subset of
possible decision regions. We investigate the properties of such decision
regions and associated classifier training algorithms.