Contact information

ｔｓｓ＠ｒｉ․ｃｍｕ․ｅｄｕ
My GPG key

cell phone:	412 608 4056
lab phone:	412 268 3922
office phone:	412 268 1420

AIM:	tstepleton
Skype:	tstepleton

Tom Stepleton's webpage

I'm a Ph.D. student at the Robotics Institute at Carnegie Mellon. My research interests include computational and biological vision, object learning and recognition, unsupervised learning, nonparametric Bayesian statistics, and information technologies for developing communities. My advisor is Dr. Tai Sing Lee. Naturally, this webpage is under development.

NB: Stepleton is pronounced with a long E: the first part sounds like the word steep.

Jump to: publications, code, other items.

0. Curriculum vitae

1. Publications

T. Stepleton, Toward Versatile Structural Modification for Bayesian Nonparametric Time Series Models. Ph.D. thesis, CMU-RI-TR-10-16, defended 6 April 2010.
T. Stepleton, Z. Ghahramani, G. Gordon, T.S. Lee, The Block Diagonal Infinite Hidden Markov Model". AISTATS 12, 2009.
T.S. Lee, T. Stepleton, B. Potetz, J. Samonds, "Neural Coding of Scene Statistics for Surface and Object Inference". In Object Categorization: Perspectives from Human and Machine Vision, edited by S. Dickenson, A. Leonardis, B. Schiele, M. Tarr. Cambridge University Press, 2008.
N. Kalra, T. Lauwers, D. Dewey, T. Stepleton, M.B. Dias, "Design of a Braille Writing Tutor to Combat Illiteracy". Information Systems Frontiers, in press.
A. Stein, T. Stepleton, M. Hebert, "Towards Unsupervised Whole Object Segmentation: Combining Automated Matting with Boundary Detection". CVPR 2008.
T. Lauwers, N. Kalra, , D. Dewey, T. Stepleton, M.B. Dias, "Iterative Design of a Braille Writing Tutor to Combat Illiteracy". ICTD 2007.
T. Stepleton, "Predicting and Evaluating the Power of Shared Features". Workshop on Empirical Evaluation Methods in Computer Vision, at CVPR 2005.
T. Stepleton, T.S. Lee, "Using Co-occurrence and Segmentation to Learn Feature-based Object Models from Video". WACV 2005.
B. Maxwell, N. Fairfield, N. Johnson, P. Malla, P. Dickson, S. Kim, S. Wojtkowski, T. Stepleton, "A Real-Time Vision Module for Interactive Perceptual Agents". Machine Vision and Applications, 14:72-82, 2003.
T. Stepleton, Work-Augmented Lazines with the Los Task Request System". USENIX LISA 2002. (Best paper award)

2. Code

A partial implementation of Bai and Sapiro's fast image matting method from ICCV 2007. Templated C++ (with Boost), MATLAB interface, OpenCV interface. Includes:
- A lightweight plotting library for OpenCV
- Glue code for using 2D MATLAB and OpenCV matrices with Boost::Graph
An implementation of a generalized Swendsen-Wang graph partition sampling algorithm described by Barbu and Zhu in 2003 and used in subsequent works. Templated C++.
Code for computing Stirling numbers of the first kind. C++ with a MATLAB MEX interface. Caches results so no number is computed more than once. Requires the GNU MP bignum library.
An implementation of the Adaptive Rejection Sampling algorithm for 1-D density functions. Templated C++ with a MATLAB MEX interface.

Yes, all of my code looks like this...

3. Other items

I collaborated with TechBridgeWorld to develop the Adaptive Braille Writing Tutor project. In Summer 2007 I worked with undergraduate Daniel Dewey to improve the adaptivity of the Tutor. Some slides about mentoring undergrads at TechBridgeWorld and why it's a good idea. The source code for the library I developed for the Tutor is not online yet, but you can browse the library documentation here.
I am on Robotics Institute Ph.D. research qualifier committees for Jean-François Lalonde and Daniel Leeds. Previously I was on Ph.D. qualifier and Masters committees for Nathaniel Fairfield and Kevin Yoon respectively.
I have TA'd for the undergraduate course 15-212: Principles of Programming. Some recitation slides are available online.

This article about Tom Stepleton is a stub. Tom Stepleton can help Tom Stepleton's academic webpage by expanding it.