Social Ai Lab Research Assistant
CurrentUnder the mentorship of Assistant Professor Roy Ka-Wei Lee, I worked on multimodal LMMs just before my senior year and investigated to what extent do these models comprehend humor.Some of the experiences that I gathered:• Conducted extensive literature review about the theories in humor and examined how they can help Large Multimodal Models (LMMs) understand humor.• Collected data and built a dataset of 28000 comics for research whilst ensuring the anonymity of identifiable information.• Crafted surveys for human and LMM annotations for quantitative and qualitative analysis.• Designed the methodology and experiments to conduct a comparative study between humans and LMMs to assess the gap in understanding humor.