• Collaborated with two graduate students to develop a toolkit that implements Kalman filter and regression modeling, supports data cleaning, interpolation and extrapolation, and provides data analysis tools and visualization.
• Explored effectiveness of Kalman filter and regression for capturing spatio-temporal correlations and cleaning sensor network data.

Research Assistant, Summer 2003
Department of Computer Science, Swarthmore College, PA

• Researched prediction and abstraction mechanisms for unsupervised robot learning.
• Developed a robot control system consisting of self-organizing maps and neural networks for learning and associating sensorimotor data with the environment.
• Participated in the Emergent Systems Working Group discussions at Bryn Mawr College.

Open Source Project, Summer 2003
Pyro (Python Robotics)

• Worked on enhancements to the self-organizing map module of Pyro, an open source, python-based, integrated programming environment used in undergraduate and graduate level robotics and artificial intelligence courses.

Senior Research Paper, Spring 2003
A Minimally Supervised Malay Affix Learner

• Developed a minimally supervised system for learning Malay affixation, using an algorithm that focused on identifying and extracting morphological relationships from text corpora via orthographic analysis, and building an affix inventory via a semantic-based approach.

• Collaborated with a fellow student to research place recognition and localization.
• Developed an artificial neural control system for controlling robot behavior and extracting regularities from time-series sensorimotor information. Used a simple recurrent network to learn motor control, and a self-organizing map to classify the network's hidden states into coherent segments corresponding to a robot's spatial location and behavior.