We have already discussed the fact that a pattern vector x appears as a point in the pattern space, and that a pattern space can be partitioned into subregions for patterns belonging to different categories. The decision surfaces that partition the space may be linear, piecewise linear, or nonlinear, and can be generalized as () d ( x ) = f ( w , x ) https://s3-euw1-ap-pe-df-pch-content-public-u.s3.eu-west-1.amazonaws.com/9780429207549/260361e8-b194-4d99-bbbf-50cb8f5fb184/content/eq112.tif"/>