AFLB
From ResearchWiki
(→Sep 16) |
(→Fall 2010) |
||
| Line 25: | Line 25: | ||
=== Sep 23 === | === Sep 23 === | ||
* Parasaran will talk about spatially-aware meta-clustering | * Parasaran will talk about spatially-aware meta-clustering | ||
| + | |||
| + | === Sep 30 === | ||
| + | * Jeff will give a talk on the following joint work with Pankaj Agarwal and Hai Yu. | ||
| + | |||
| + | Stability or ε-Kernels. | ||
| + | |||
| + | Given a point set P of n points in R^d, an ε-kernel K ⊆ P is a coreset of P that approximates the convex hull of P. As a result, it has been found very useful in the past 10 years for many geometric approximation problems. | ||
| + | In this talk I will discuss the stability of ε-kernels in two senses. | ||
| + | The dynamic stability shows how to maintain the ε-kernel as points are inserted and removed from P. We show how to efficiently maintain K with a constant number of changes for each point added or removed from P. | ||
| + | The approximation stability studies how the size of K can change as ε changes. This reveals structure about the optimal shape of K and how stabile it is in different dimension. In summary, our results show that K is stable for d=2,3, but for higher dimensions it can be quite unstable with respect to changes in ε. | ||
| + | |||
| + | [http://www.cs.utah.edu/~jeffp/papers/arXiv1003.5874v1+.pdf Stability of eps-Kernels] | ||
== Papers for discussion == | == Papers for discussion == | ||
Revision as of 22:10, 29 September 2010
The Algorithms For Lunch Bunch
Thu @ Noon (starting Aug 26, 2010)
Venue: the graphics annex
Contents |
Fall 2010
Aug 26
Suresh will talk about the recent P vs NP kerfuffle, along with some background on the problem.
Sep 2
We have a double bill:
- Avishek will talk about adaptive methods for multi-task learning (abstract) (paper)
- Suresh will talk about information-theoretic repair for databases violating integrity constraints
Sep 9
- Rasmus will talk about JL on the sphere and the simplex
Sep 16
- Qiushi will talk about recent work on implementations of the Johnson-Lindenstrauss Transform
Sep 23
- Parasaran will talk about spatially-aware meta-clustering
Sep 30
- Jeff will give a talk on the following joint work with Pankaj Agarwal and Hai Yu.
Stability or ε-Kernels.
Given a point set P of n points in R^d, an ε-kernel K ⊆ P is a coreset of P that approximates the convex hull of P. As a result, it has been found very useful in the past 10 years for many geometric approximation problems. In this talk I will discuss the stability of ε-kernels in two senses. The dynamic stability shows how to maintain the ε-kernel as points are inserted and removed from P. We show how to efficiently maintain K with a constant number of changes for each point added or removed from P. The approximation stability studies how the size of K can change as ε changes. This reveals structure about the optimal shape of K and how stabile it is in different dimension. In summary, our results show that K is stable for d=2,3, but for higher dimensions it can be quite unstable with respect to changes in ε.
Papers for discussion
Previous Semesters
Contact
If you are interested in giving a talk at AFLB or have questions, please feel free to send a mail to moeller@cs.utah.edu, praman@cs.utah.edu or avishek@cs.utah.edu. If you are planning to give a talk, we would really appreciate if you have an abstract ready a week before the talk is scheduled.
