Global ETD Search

Novel Fast Algorithms For Low Rank Matrix Approximation

Svadlenka, John TThe Graduate School and University Center of The City University of New York2021

… The Subsampled Randomized Hadamard Transform (SRHT) is the most popular among transforms. An m x n matrix can be multiplied by an n x l SRHT matrix in O(mn log l) arithmetic operations where typically l << min(m, n). This dissertation introduces an alternative, which we call the Subsampled …

cuny-grad Repository record for Novel Fast Algorithms For Low Rank Matrix Approximation (opens in a new tab)