
Before coming
to Rochester I worked for a year at Adobe India and
even before that I was at IIT Kanpur pursuing a
Master's degree.
Research Interests
Active learning for
generative models; Active prediction for graphical
models; Nonparametric density estimation.
I am advised by Daniel
Stefankovic; Dan Gildea is also on my committee.
Publications
- Learning mixtures of Gaussians
with maximum-a-posteriori oracle, 14th International
Conference on Articial Intelligence and Statistics (AI &
Statistics 2011) [pdf].
- Approximating L1-distances between mixture distributions
using random
projections, 6th Workshop on Analytic Algorithmics and
Combinatorics, 2009 (with Daniel Stefankovic) [pdf] (preprint).
- Density estimation in linear time, 21st Annual Conference
on Learning Theory, 2008
(with Daniel Stefankovic) [pdf].
Working Papers
- Active prediction in graphical models, under review (with
Daniel Stefankovic) [pdf]
(older version).
- A note on active learning for smooth problems [pdf].
I can be reached at smahalan
at cs dot rochester dot edu.