A0553
Title: Locally-adaptive boosting for modeling non-stationary processes
Authors: Daisuke Murakami - The Institute of Statistical Mathematics (Japan) [presenting]
Abstract: A boosting algorithm is proposed for modeling non-stationary spatial processes in regression coefficients. Following the idea of progressive learning, the model is trained in a staged manner, progressively learning from coarser patterns, followed by finer processes. In each learning step, a local model, which may explain anisotropic pattern, is estimated and ensembled. The performance of the developed method is examined by simulation experiments and application to residential land price data.
