Nearest-Instance-Centroid-Estimation Linear Discriminant Analysis (NICE LDA)

We propose a novel ensemble classification technique called the Nearest Instance Centroid Estimation (NICE) LDA algorithm. Our algorithm (inspired from NICE KLMS) performs a combination of two weak classifiers threshold based clustering and linear discriminant classification to achieve stateof-the-art results on various high dimensional UCI datasets. We discuss the important ways in which our method of clustering followed by LDA implementation is more robust towards skewed datasets and computationally faster than previous ensemble methods. We also develop an efficient aggregation method based on instance based learning that implements this classification technique using significantly less computational power. We demonstrate that our method of data clustering and LDA implementation, while introducing only one free parameter, leads to results that are similar and often better than those achieved by the state-of-the-art kernel RBF SVMs.