[an error occurred while processing this directive] [an error occurred while processing this directive] [an error occurred while processing this directive] [an error occurred while processing this directive] PhD Candidate [an error occurred while processing this directive] Northwestern University [an error occurred while processing this directive] [an error occurred while processing this directive] [an error occurred while processing this directive] [an error occurred while processing this directive]

Cyber-Physical Systems and Design Automation

Human-Computer Interaction

[an error occurred while processing this directive] Re-imagining Wearable Visual Observation [an error occurred while processing this directive] Sensor-based computational models have the potential to revolutionize behavioral research by providing day-to-day insights on human behavior, advancing our understanding of behavioral models to enhance health and wellbeing. The development and validation of computational models to detect daily human behaviors using wearable devices require labeled data collected from the natural field environment. However, existing groundtruth measures are inadequate, explaining the shortage in labeled human activity datasets. Wearable cameras are extensively used to extract fine-grained groundtruth labels to build machine-learned models for other less invasive sensing modalities. Naturally, the wearers and bystanders might feel uncomfortable using wearable cameras, creating a catch-22 paradox around privacy and utility.

In my poster, I will present part of my work that responds to this groundtruth challenge which prevents wearable behavioral sensing systems from realizing their full potential. My research demonstrates several efficient and practical solutions to improve visual groundtruth systems by investigating the privacy-, energy-, burden- information tradeoff space. These include (1) modifying recording affordances, (2) introducing privacy-by-default obfuscation, and (3) implementing efficient foreground information extraction. By improving current visual groundtruth methods, I aim to enable the building and verification of sensor-based behavioral models for health and wellbeing applications. [an error occurred while processing this directive] Rawan Alharbi is a Ph.D. candidate in Computer Science at Northwestern University and part of the interdisciplinary Health Aware Bits (HABits) lab that bridges Computer Science and Health Sciences research. Rawan’s research focuses on the design and development of practical, long-lived wearable technologies to enhance wellbeing. Her interdisciplinary work has been published in top-tier computer science conferences (UbiComp/IMWUT, CHI, PerCom) and behavioral science conferences (Obesity and Society of Behavioral Medicine). Rawan is currently on the job market looking for faculty positions in CS departments and iSchools. [an error occurred while processing this directive] Personal home page [an error occurred while processing this directive] [an error occurred while processing this directive]