The thesis proposed a new dimension reduction technique called Array Independent Component Analysis (AICA) that reduces the volume and noise in non-normally distributed high dimensional data sets. It is a generalization of Independent Component Analysis to data sets with more than two modes. One application of the methodology developed is remote sensing, where observations are described by longitude, latitude and wavelength. We focus on analyzing remotely sensed multispectral images of the White Oval on Jupiter (captured by the Hubble Space Telescope and the Infrared Telescope Facility). AICA uncovers similarities in temperature gradients between the White Oval and the Great Red Spot.
University of California, Los Angeles
Ph.D., M.S. in Statistics
B.A. and S. in Statistics and Economics, Minor in Mathematics