Abstract
David Woodruff (CMU): Dimensionality Reduction Techniques for Randomized Linear Algebra
I will give an abridged version of my Algorithms for Big Data course at Carnegie Mellon University, in particular, I will cover fast dimensionality based reduction techniques for problems such as regression and low rank approximation. These involve using random matrices to reduce a large instance of a problem to a much smaller instance. Other techniques include sampling schemes to approximately preserve structural properties of one's data. I will also talk about these problems in various models, such as the distributed and streaming models of computation.
Back to the Agenda