Elementary Features of Local Learning Algorithms
Locally Weighted Projection Regression (LWPR) and the learning classifier system XCSF share a common divide and conquer approach to approximate nonlinear function surfaces by means of three elementary features:
Clustering A complex non-linear problem is broken down into several smaller problems via kernels. The kernel structures are further optimized for accurate approximations.
Unable to display preview. Download preview PDF.
© Springer Fachmedien Wiesbaden 2014