Professor Michael Trosset has been awarded funding to support his development of a statistical theory that researchers at Johns Hopkins University (JHU) are applying to large language models (LLMs) for artificial intelligence. LLMs rely on constructing representations of word associations. Trosset, an expert on mathematical representations of pairwise associations, has developed a theory of "continuous multidimensional scaling" that is being used to study how LLMs behave as the number of models and/or the number of queries to those models increases.
Statistics faculty member developing theory for study of large language models.
Tuesday, February 17, 2026
