Research
Racial Bias in Facial Recognition: The Role of Training Data and Human Prejudice
This project integrates psychological measures of prejudice with deep learning pipelines to examine racial bias in facial recognition. Dataset imbalance and annotator prejudice drive racial bias in facial recognition more than model design.
NLP-Driven Insights into Queer Sentiment During Legal Milestones
A longitudinal NLP study of emotional sentiment among openly queer Twitter users from 2012–2017, capturing shifts around two major U.S. Supreme Court rulings on same-sex marriage
Blame the Bot? Evaluating Racial Bias from Humans vs AI
A behavioral study examining how people evaluate racial bias from humans versus AI systems, revealing shifts in emotional response and blame attribution based on bias type and perceiver identity.
Gender Differences in Response to East Asian Erotic Stimuli: A Validated Image Dataset
This study presents the East Asian Erotic Picture Dataset to address the lack of representative stimuli in sexuality research. Ratings from Chinese college students revealed gender differences in arousal and attractiveness responses to opposite-sex images, with females favoring semi-nude male stimuli. The dataset offers a validated resource for studying sexual function and cultural influences in East Asian populations.
