Online fractal dimensionality reduction in time decaying stream environment

Dimensionality reduction, the process of reduce the number of dimension of the original feature set, plays an important role in a wide variety of contexts such as classification, Prediction and clustering. It is common to introduce the dimensionality reduction prior to the subsequent data mining tasks in the classical static data. However, this aforehand option can not capture the essence of datum because of the inherent time variety and one-pass constraint of data stream. In this case, dimensionality reduction should interact with the evolution of stream data with time elapsed. We introduced the interaction between dimensionality reduction in the time decaying high dimensional stream environment and propose the on-line fractal dimensionality reduction technique. Our performance experiments over a number of real and synthetic data sets illustrate the effectiveness and efficiency provided by our approach.