Natural language research announcements

["Samdani, Rajhans" <rsamdan2 AT illinois.edu>]

Please see the seminar announcement below.

Best,

Rajhans

---

From: Zhai, Chengxiang
Sent: Wednesday, August 07, 2013 12:53 PM
To: Samdani, Rajhans
Subject: FW: Special Yahoo!-DAIS seminar talk: Sayan Ranu, 10-11am, Aug. 9, 2013

Hi Rajhans,

Could you please help advertising the following talk among relevant AI students?

Thanks,
-Cheng

---

From: dais-bounces AT cs.uiuc.edu [dais-bounces AT cs.uiuc.edu] on behalf of Zhai, Chengxiang [czhai AT illinois.edu]
Sent: Monday, August 05, 2013 10:14 AM
To: dais AT cs.uiuc.edu
Subject: [dais] Special Yahoo!-DAIS seminar talk: Sayan Ranu, 10-11am, Aug. 9, 2013

Hope you are all having a great summer.  There will be three special Yahoo!-DAIS seminar talks right before the start of the new semester. The first is the following talk by Dr. Sayan Ranu from IBM Research - India this Friday at 10-11am, Aug. 9, in 3405SC (please see below for details). 

The next two will be given by Prof. Jie Tang from Tsinghua Universsity at 2-3pm on Aug. 15 (Thursday), and Prof. Huan Liu from Arizona State University 2-3pm on Aug. 22 (Thursday), respectively.

Hope to see you at these seminar talks!

-Cheng

########  Special Yahoo!-DAIS Seminar Talk #######

SPEAKER: Sayan Ranu, IBM Research - India
TIME:  10am-11am, Friday, Aug. 9, 2013
PLACE:  3405 Siebel Center


TITLE: Indexing similarity queries over noisy trajectories.


ABSTRACT:
The ability to accurately quantify similarity between two trajectories is
central to a multitude of analytical tasks on spatio-temporal databases. In
this talk, we will look at the impact of noise on existing trajectory
matching techniques and discuss a new distance function called Edit
Distance with Projections (EDwP) to combat these challenges. EDwP
formalizes a threshold-free distance function that is robust to variance in
sampling rates. Additionally, by deploying the idea of projections, EDwP
goes beyond matching only the sampled points to align trajectories.

To enable efficient trajectory retrievals using EDwP, we will discuss an
index structure called TrajTree, which derives its pruning power by
employing the unique combination of bounding boxes with Lipschitz
embedding. Finally, I will present some preliminary work at IBM on how
trajectories can be better inferred from partial observations in the
presence of an underlying road network model and their applications in the
telecommunication domain.

BIO:
Sayan Ranu is a researcher at the Information Management group at IBM
Research, Bangalore. Sayan works on spatio-temporal data analytics with a
particularly strong focus on noisy trajectory databases. Prior to joining
IBM, Sayan completed his Ph.D. from the University of California, Santa
Barbara under the guidance of Prof. Ambuj K. Singh. His dissertation
focused on graph mining and indexing and their subsequent applications to
drug discovery.

_______________________________________________
dais mailing list
dais AT cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/dais