Button to scroll to the top of the page.

Events

Monthly View
By Month
Weekly View
By Week
Daily View
Today
Search
Search
Junior Applied Math And Probability
Download as iCal file
Ruhui Jin, RLM 9.166: Dimensionality reduction and its applications in numerical linear algebra
Friday, March 01, 2019, 02:00pm - 03:00pm
Randomized algorithm has proved its success on reducing the computationcost, memory and storage for solving linear algebra problems.With linear "sketching" technique, one can compress an over constrainedsystem to a much smaller setting. The idea of dimensionalityreduction can serve as a guide to find suitable number of measurementsfor the sketch matrices. This talk will focus on a well-knownclass of random projections supporting fast matrix-vector multiplications:fast Johnson-Lindestrauss transform (FJLT), which enable low-distortion\ell_2 embedding with high probability. It has been studied thatthe fast JL embedding used for sketching the least-square problemscan help output fairly good approximations of the optimal solutions.We will also study a derandomized version of the JL embedding,designed not only for matrix problems, but for higher-order tensorCP decompositions (tensor low-rank approximation).
Location: RLM 9.166

Math Calendar Login