On Dimensionality Reduction for Indexing and Retrieval of Large-Scale Solar Image Data

摘要: This work investigates the applicability of several dimensionality reduction techniques for large-scale solar data analysis. Using a benchmark dataset that contains images multiple types phenomena, we investigate linear and nonlinear methods in order to reduce our storage processing costs maintain good representation new vector space. We present comparative analysis different numbers target dimensions by utilizing classifiers determine degree can be achieved with these methods, discover method is most effective images. After determining optimal number dimensions, then preliminary results on indexing retrieval dimensionally reduced data.