Missing Values in Nonlinear Factor Analysis

摘要：

The properties of the nonlinear factor analysis (NFA) model are studied by measuring how well it reconstructs missing values in observations. The NFA model uses a multi-layer perceptron (MLP) network for approximating the nonlinear mapping from factors to observations. The NFA model is compared with linear factor analysis (FA) and with the self-organising map (SOM). The number of parameters in the NFA model is closer to FA than the SOM, but unlike FA, NFA is able to model nonlinear manifolds. Based on experiments with real world speech data and Boston housing data, we conclude that the performance of the NFA model is closer to FA.

作者: Tapani Raiko Harri Valpola

作者单位: Neural Networks Research Centre Helsinki University of Technology P.O.Box 5400, FIN-02015 HUT, Espoo, Finland

会议类型: 国际会议

会议名称: 8th International Conference on Neural Information Processing(ICONIP 2001)(第八届国际神经信息处理大会)

会议地点: 上海

会议语种:英文

页码: 860-865

在线出版日期: 2001-11-14（万方平台首次上网日期，不代表论文的发表时间）