Summer 2004: I worked on streaming algorithms with my advisor Mitsu Ogihara. A streaming algorithm is one that can work with huge streams of data (presented as items) and compute functions of this data (probably, approximately) using minimal space and time per item. Since there is typically an enormous amount of data, even linear storage is infeasible. To read more on this, refer to a great survey by Muthukrishnan.
Spring 2005: Worked with Mitsu Ogihara in designing a novel data structure, based upon the counting Bloom filter, to maintain counts of a large number of items.
Summer/Fall 2005: Worked with Mitsu Ogihara, Vyas Sekar, Hui Zhang, and Jim Xu on estimating the entropy in a streaming context. Estimating the entropy of a stream is a natural problem that arises in networking when the administrator wants to monitor features of network traffic or detect anomalies. We presented algorithms based on the celebrated Alon--Matias--Szegedy algorithm for computing the frequency moments over a stream. We also empiricially demonstrated that sieving out the high-frequency items from the stream increased the performance considerably. The technical report version of this paper can be found here. "Data Streaming Algorithms for Estimating Entropy of Network Traffic" has been accepted to SIGMETRICS 2006.
Spring 2006: I was at Georgia Tech, working with Jim Xu on various streaming problems. I also sat in on a class on advanced networking taught by Nick Feamster and one on randomized algorithms taught by Eric Vigoda.
Summer/Fall 2006: I worked with Mitsu Ogihara and Jim Xu on designing a novel system to keep track of and estimate counts of high-frequency items or elephants in a stream.
Spring 2007: I worked with Chuck Zhao, Mitsu Ogihara, Jim Xu, Oliver Spatscheck, and Jia Wang on an algorithm for estimating the entropy of origin-destination pairs in a network. "A Data Streaming Algorithm for Estimating Entropies of OD Flows" was accepted to IMC 2007.
Summer 2007: I interned at the IBM T.J. Watson lab in the area of "problem determination." I was mentored by Anca Sailer and my manager was Dennis G. Shea. I worked with Anca Sailer and Mark Brodie on a project we called SPIRIT (Service for Providing Infrastructure Recommendations for IT). A short paper on this work was accepted to NOMS 2008, a full version is under submission, and two patents have been filed.
Fall 2007: I am working on two projects with my co-advisor, Jim Xu. One problem is in the field of anomaly detection, jointly with collaborators at AT&T Research, Jia Wang and Qi Zhao. The second project is in distributed iceberg detection, in collaboration with John Byers at Boston University.
In the Spring of 2005 I was awarded the Edward Peck Curtis Award for Excellence in Teaching by a Graduate Student.
CSC400 Graduate Seminar
with
Randal Nelson
(Fall 2003)
CSC444 Logical Foundations of AI
with
Len Schubert
(Fall 2003)
CSC446 Mathematical Foundations of AI
with
Dana Ballard
(Spring 2004)
CSC447 Natural Language Processing
with
Len Schubert
(Fall 2004)
CSC455 Advanced Programming Systems
with
Chen Ding
(Fall 2003)
CSC456 Operating Systems
with
Kai Shen
(Spring 2004)
CSC 458: Parallel & Distributed Systems
with
Sandhya Dwarkadas
(Spring 2005)
CSC484 Advanced Algorithms
with
Joel Seiferas
(Spring 2004)
CSC486 Computational Complexity
with
Lane Hemaspaandra
(Fall 2003)
CSC487 Randomized, Parallel, and Other Advanced Modes of Computation
with
Joel Seiferas
(Fall 2004)