Our next meeting will be held on Wednesday 15th of March, at 5.00pm, in LG.11, David Hume Tower (on the lower ground floor - map below). As usual, the meeting will be followed by drinks and chat in the Potting Shed. Meetings are open for all to attend and newcomers/beginners are very welcome.
Our speaker is Benjamin Erichson from the School of Mathematics and Statistics at the University of St Andrews, who will talk about:
Randomized Matrix Decompositions using ‘rsvd’
Matrix decompositions play an important role for machine learning, statistical computing and elsewhere. However, massive datasets pose a computational challenge for traditional (deterministic) algorithms.
Over the past two decades, the powerful idea of randomness as a computational strategy to find low-rank approximations has emerged. The basic idea is to employ a degree of randomness to derive from a high-dimensional matrix a smaller matrix, which captures the essential information. Subsequently, the smaller matrix can be used to efficiently compute a near-optimal low-rank approximation.
In this talk we give an overview of randomized matrix algorithms and present the ‘rsvd’ package. Specifically, this package provides routines to compute the randomized singular value decomposition (rSVD) and randomized principal component analysis (rPCA). We will show several examples and demonstrate the computational advantage.
For any newcomers, here’s a map of where we’ll be.