p-ASL : Sign Language Recognition with Deep Learning (Lead: Marcos Acosta HMC '23) 🤞

• Extract hand and body landmarks using Google's MediaPipe
• Apply Deep Learning to recognize American Sign Language (ASL) signs in real-time
• Focus on the problem of correctly recognizing signs that involve motion
• Learn, explore, experiment, and have a bit of fun in a techy, supportive team
• Perfect for anyone interested in socially responsible AI

p-sync: Disillusioned with a chance of compassion, isolated with a chance of immunity: Forecasting Order in Chaotic Worlds (Lead: Jacob Zimmerman PO '23) 💉

• Complex systems are useful for modeling much of the dynamic world, balancing chaos and synchrony
• Such systems can be difficult to model using a simulated recreation, as a defining characteristic of such systems is the sensitivity to initial conditions.
• Reservoir computing has recently shown promise for rapidly forecasting the states of a chaotic complex system, without recreating the system itself.
• Our goal is to explore the potential of the technique for modeling diverse data, both from controlled, experimental simulations and from messy, complex systems in the world that connect people, ideas, and power.
• By developing a specialized tool for predicting the states of dynamic complex systems, we hope to take away lessons about the dynamic, chaotic nature of our worlds, and to contribute to a modern technological arsenal of instruments for taking the pulse of human activity, from predicting surges in the spread of viral misinformation and infectious diseases, to societal healing and moral unity.

p-web: Exploring Language Modeling Approaches Toward HTML (Lead: Andy Liu HMC '23)🗣️

• Building a neural network-based language model of the HTML markup language
• Scraping and processing a corpus of HTML to build such a model
• Training models for tasks such as filling in masked HTML tokens or generating new HTML from a prompt
• Leveraging new techniques used by models such as BERT or GPT-2