Social Network Clustering of Sparse Data

Start: 12/09/2011 - 1:15pm
End  : 12/09/2011 - 2:15pm

Applied Math Seminar

Blake Hunter, UCLA


Trillions of devices around the world are continuously
producing exabytes of data every day. The impact of search engines
has been enormous, but there is also a parallel development in the
applications of these methods to other related problems concerning the
extraction of knowledge from large datasets. Data mining is the
mathematics, methodologies and procedures used to extract knowledge
from large datasets. While this includes topics related to search
engines it is mainly devoted to the more general problem of finding
features and structure in a dataset. There are many active scientific
fields, including pure and applied mathematics, statistics, computer
science and engineering with numerous applications such as finance,
the social sciences, and the humanities. Spectral embedding uses
eigenfunctions of a Laplace operator (or related graph affinity
matrix) for extracting the underlying global structure of a dataset.
This talk will give an introduction to spectral embeddings.
Applications presented will include clustering LA street gang members
based on sparse observations of where and who they are seen with, and
automatically topic detection of Twitter tweets and Amazon product