AFLB
From ResearchWiki
(→Oct 21) |
(→Oct 21) |
||
| Line 43: | Line 43: | ||
=== Oct 21 === | === Oct 21 === | ||
* Amirali will discuss some related problems dealing with epsilon-nets and centrality. Some representative slides/papers. | * Amirali will discuss some related problems dealing with epsilon-nets and centrality. Some representative slides/papers. | ||
| - | + | ( http://russell.lums.edu.pk/~nabil/Other/math-lec4.pdf (a stronger version of this problem is still open)) | |
| - | + | ( http://cccg.ca/proceedings/2010/paper42.pdf (only the part on axis-parallel rectangles)) | |
| - | + | ( http://russell.lums.edu.pk/~nabil/Research/Papers/centerpointgen.pdf (skip any lower bounds related stuff)) | |
| - | + | ( http://www.cccg.ca/proceedings/2005/75.pdf (focus on the upper bounds, dealing with convex sets)) | |
== Papers for discussion == | == Papers for discussion == | ||
Revision as of 00:11, 21 October 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
- Meta-Issues: Suresh will take questions on research, being in academia / industry, and other issues.
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.
Oct 7
Brainstorm: Suresh will throw out two problems that came up in recent conversations.
Oct 21
- Amirali will discuss some related problems dealing with epsilon-nets and centrality. Some representative slides/papers.
( http://russell.lums.edu.pk/~nabil/Other/math-lec4.pdf (a stronger version of this problem is still open)) ( http://cccg.ca/proceedings/2010/paper42.pdf (only the part on axis-parallel rectangles)) ( http://russell.lums.edu.pk/~nabil/Research/Papers/centerpointgen.pdf (skip any lower bounds related stuff)) ( http://www.cccg.ca/proceedings/2005/75.pdf (focus on the upper bounds, dealing with convex sets))
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.
