Tze Yi Tiong
Research Mentor(s): Alanson Sample
Mentor Department: Computer Science and Engineering
Authors: Tze Yi Tiong, Yasha Iravantchi, Alanson Sample
Session: Session 1 (9:00am – 9:50am)
Presentation Type: Poster 69
Abstract
In recent years, concerns about home appliances and Internet of Things (IoT) devices, such as smartphones and robotic vacuums, eavesdropping on users have raised significant privacy issues. These devices often collect sensitive audio data, leading to potential invasions of privacy. Our project, PUFFIN (Privacy-preserving Ubiquitous Functional Forecasting & INference), aims to develop a high-bandwidth audio system that monitors daily activities at home via inaudible frequencies to preserve user privacy. One key application of PUFFIN is in the homes of individuals with Multiple Sclerosis (MS), a chronic disease affecting the central nervous system. By tracking essential activities such as bathroom visits and meal breaks, PUFFIN could help caregivers and medical professionals better understand daily activity patterns without infringing on personal privacy. Building upon an earlier version of PUFFIN published by my research mentor, our team seeks to enhance the audio system’s cost-effectiveness and improve the robustness of its machine learning (ML) model.
