Axes-oblique partitioning strategies for local model networks

Local model networks, also known as Takagi-Sugeno neuro-fuzzy systems, have become an increasingly popular nonlinear model architecture. Usually the local models are linearly parameterized and those parameters are typically estimated by some least squares approach. However, widely different strategies have been pursued for the partitioning of the input space which determines the validity regions of the local models. The model properties crucially depend on the chosen strategy. This paper proposes an axes-oblique partitioning strategy and an efficient construction algorithm for its realization. Many advantages over the existing approaches are demonstrated