基于EKF的神经网络自适应在线学习算法，包含例子和文档。-We show that a hierarchical Bayesian modeling approach allows us to perform regularization in sequential learning. We identify three inference levels within this hierarchy: model selection, parameter estimation, and