Archetypal Analysis: Representation with Meaning
Talk by Catherine Potts (Graduate Student in Mathematical Sciences, MSU) - PhD Proposal Presentation
11/21/2019 Wilson Hall 1-144 3:10-4:00pm
In this PhD proposal, archetypal analysis (AA) is presented as a data model that features a representative dictionary consisting of archetypes which can be interpreted in the context of the original data set. An application of AA to calcium imaging of neuronal cell signals mediated by nanoparticles will be presented as part of joint work with the Kunze Lab. Future work improving the AA algorithm by introducing random projections or sketches to the data, and re-posing the problem in order to apply a primal-dual hybrid gradient method will be discussed as well as a comparison to Principal Component Analysis.
( h/t to Emily Stone at U of M, who introduced the concept of AA to us, at the Data Science Summit in Butte... )