抄録
In this paper, we show the experimental study on adaptive robust neural network Principal Component Analysis (PCA) based on a reconstruction error model. Firstly we explain the traditional batch PCA method which is based on eigenvalue decomposition and discuss its problems of computational complexity and poor robustness. To overcome such problems, the adaptive robust neural network Principal Component Analysis will be introduced. This adaptive robust approach is based on the structure of single-layer neural network with modification of the reconstruction error model. From the experiments, it can be seen that this method can reduce the effect of outliers existing in the training sample set.
