Nonlinear equalization was previously shown to be beneficial for recor
ding channels with nonlinearities and signal-dependent noise. In this
work, we study two types of simplified nonlinear filter architectures
that are suitable for VLSI implementation. The performance loss due to
the simplifications are not significant in most cases, compared to th
e signal-to-noise ratio gain achieved by nonlinear equalization. In on
e form of simplification the need for full multiplications is eliminat
ed and in the other the number of multiplications in the original nonl
inear filter architecture is reduced significantly. Work on experiment
al and simulated data are presented.