Welcome to the academic year 2026-2027. Join us for the first talk of the Fall Semester Monday next week!
Computer Science Seminar Series
Tracing the Signs: Multimodal Foundation Models for Ancient Writing
Morris Alper, Ph.D.
Ungar Building, Room UB305 (3rd Floor), Coral Gables Campus
Monday 08/31/2026 at 5:05PM | Zoom: https://miami.zoom.us/j/95533396867
Refreshments at 4:30PM
Abstract
Modern vision-language models (VLMs) can read handwriting, parse complex documents, and
translate text across dozens of languages. But when applied to photographs of ancient i
nscriptions, whether carved, incised, or impressed into clay, metal, and rock thousands
of years ago, they fail. Our ongoing research studies how foundation VLMs can be
adapted out-of-domain to handle complex ancient writing systems, with applications
to cuneiform, ancient North Arabian inscriptions, and potentially many others.
I will show how we can adapt diffusion models for pose estimation to estimate the
internal configurations of signs for paleography and improved optical character
recognition, efforts to create new benchmarks, and our vision for future models
bridging visual knowledge with ancient languages and epigraphy.
Bio
Morris Alper is an Assistant Professor in Computer Science at the University of Miami.
His research focuses on multimodal learning and AI applications for linguistics,
applied across textual, visual, and audio modalities. He joined the faculty this
month; previously, he was a postdoc at CMU's Language Technologies Institute,
after receiving his PhD in multimodal machine learning from Tel Aviv University.
