NYU Media Research Lab
719 Broadway, 12 FL
New York, NY 10003
ccp252 [ AT ] cs.nyu.edu
|
|
Chris Pennock NYU Media Research Lab 719 Broadway, 12 FL New York, NY 10003 ccp252 [ AT ] cs.nyu.edu |
Bio Sketch: I am a Master's student at the Media Research Lab at New York University, finishing May 2006. My academic interests are in Machine Learning and modeling human behavior. |
 |
RapunselI am currently working on Rapunsel, an NSF-funded project to encourage middle-school girls to program in Java by letting them compete in dance competitions with a Java-controlled dance bot.My work has been primarily on the code editor/interpreter and on lesson plans. |
 |
Gaze Model for Simulated PedestriansAs an addition to a human pedestrian behavioral simulation (Shao 05), I built an algorithm and API to add gazing behaviors to all higher level behaviors.This project started as a project for Demetri Terzopoulos's Artificial Life class. Currently in preparation for submission to SCA. An early draft is here. |
|
Multiplayer Tabletop game using Computer VisionFor Ken Perlin's Advanced Multimedia class, I led a group to build this real-time, competitive multiplayer game with a shared physical interface played on a projected tabletop display.A video of the game in action is here. A nice discussion of the lessons learned during the project is here. |
|
Statistical Arbitrage Stock Trading using Time-Delay Neural NetworksAs a final project for the Yann Lecun's Neural Networks class, I built a learning machine using Time-Delay Neural Networks (TDNN) that could predict the direction of a stock's closing price relative to the previous close, based on the previous year of activity. The algorithm was correct for greater than 50% of the sample stocks, and for greater than 50% of the days - an impressive result. The paper is here. |
|
Sign Language RecognizerUsing a 4-dimensional version of the Lucas-Kanade algorithm, we compared prototype sign language characters to video and still input of real hands signing. Restricting input to a subset of 6 characters, we achieved 100% correct classification on signs performed by hands other than the prototype owners, across a range of rotations. This was the final project for the Computer Vision class.The full results can be viewed here. |
|
Renderers: Wireframe, Ray Tracer and Z-BuffferAs projects for a Computer Graphics class taught by Ken Perlin, I created a series of rendering engines and graphic demos to showcase them. The best demo was Gladiators: Predator vs. Prey which used the Wireframe engine. See the rest of the demos here. |
 
|
Massachusetts Museum of Contemporary ArtIn 2002, I built an interactive, self-curated exhibit for MassMoCA.Launch the Demo: (your show here) From the brochure: Welcome to (your show here), an interactive forum that lets you create your own exhibition. You are invited to sit at the computer terminal, browse a database of twentieth-century art images, choose up to five, write a curatorial statement, and title the show. The digital images are instantly projected at the scale of the original objects, creating a gallery of virtual works of art at the click of a button. The exhibition's duration is fleeting, since each show is replaced by that of the next "visiting curator," but a print-out of your selections can be posted on the bulletin board near the gallery entrance. |
|
CMLabsCMLabs is a group of AI-focused computer scientists building tools for rapid prototying of modular AI systems. These software tools include Psyclone, a messaging server to manage communcation between modules. Psyclone offers an XML-based language to describe the behavior of and communcation between modules, allowing a single portable file to describe the behavior of an entire system. Also important is A.I.R, a set of networking libraries to allow code in Java, C++ and soon LISP to communicate with the server.We also host Mindmakers.org, a site for sharing modules and projects based on CMLabs technology. This technology has been used sucessfully in Professor Thorisson's class at Columbia, Embodied Agents in Augmented & Virtual Realities. A page describing their successes building a virtual agent, MIRAGE, can be seen here. Multiple publications have resultsed from thsi research (see "Publications", below). |