【Big data workshop】Integrated Singular Value Decomposition and PCA for Large Matrices
2017/06/07
研討會日期 : 2017-06-07
時間 : 11:00
主講人 : Professor Su-Yun Huang
地點 : Conference Room B110
演講者簡介 : Professor Huang received her Ph.D. from Purdue University , USA in 1990. She is currently a research fellow at the Institute of Statistical Science, Academia Sinica. Her research interests are Statistical Machine Learning, Statistical Analysis for Biological Image Data, Nonparametric Inference, and Bayesian Decision
演講摘要 : The singular value decomposition (SVD) and PCA of large-scale matrices is a key tool in dimension reduction and big data analytics. The rapid growth in dimension and sample size has increased the need for developing efficient large-scale SVD/PCA algorithms. Randomized SVD/PCA based on one-time sketching has been studied, and its potential has been demonstrated in recent literature. Studies for randomized algorithms have become a new trend in numerical linear methods for big data. Instead of exploring different techniques for single random sketching, we propose a Monte Carlo type integrated SVD algorithm based on multiple random sketches. The proposed integration algorithm takes multiple random sketches and then integrates the results obtained from the multiple sketched subspaces. The integrated SVD can achieve higher accuracy and lower stochastic variations. Numerical results will be presented.
