Multiple-model multiscale data fusion regulated by a mixture-of-experts network

Multiscale Kalman smoothers (MKS) have been traditionally employed for data fusion applications and estimation of topography. The standard MKS algorithm embedded with a single stochastic model has been found to give suboptimal performance in estimating nonstationary topographic variations, particularly when there are sudden changes in the terrain. In this work, multiple models are regulated by a mixture-of-experts (MOE) network to adaptively fuse the estimates. Though MOE has been widely applied to one-dimensional data, its extension to multiscale estimation is new