Recognition of a hand-drawn PowerPoint slide.
I am a PhD student in the UC Berkeley EECS department, studying under Professor Richard Newton. My research interests include sketch-based user interfaces, embedded systems design, visual languages, and information visualization.
The primary focus of my research is a generic algorithm for probabilistic, incremental, two-dimensional parsing. I am using the algorithm to perform robust recognition of syntactic structures like bulleted/numbered lists and diagrams within a sketch-based version of Microsoft's PowerPoint presentation software. I am very fortunate to be collaborating with Heloise Hse on low-level shape recognition, Hanna Pasula on a probabilistic framework in which to model hand-sketched diagrams, Andrew Begel on parsing algorithms for ambiguous input streams, Niraj Shah on the PowerPoint application, and Anoop Sinha on multi-modal interfaces combining speech and pen.
I am also actively involved with the Mescal project, in which we are developing a new methodology for embedded systems design. Our goal is to provide a programmer's model and software development environment that allows for efficient implementation of an interesting set of applications onto a family of fully-programmable architectures / microarchitectures. I am working on the applications end of things, trying to write the world's fastest software-only network router.
Finally, I work on the Diva toolkit from time to time. Diva is an architecture for visualizing and interacting with dynamic information spaces, co-developed with Heloise Hse, Steve Neuendorffer, and John Reekie. My MS report generalizes the design of Diva's graph package into the concept of visualization protocols, a policy that enables the construction of reusable visualization components.