Our upcoming event for the Statistics Colloquium Series is scheduled for Monday, September 28th from 12:00 – 1:00pm (ET) and will be an in-person presentation at Maxwell-Dworkin, Room 119. Lunch will be provided to guests following the talk. This week's speaker will be Anna Seigal of Assistant Professor of Applied Mathematics in the School of Engineering and Applied Sciences at Harvard, and an Affiliate in the Department of Statistics.
Factorizations for data analysis
Abstract: Factorizations underpin data analysis methods. A classical example is principal component analysis (PCA), which uses the eigendecomposition of the covariance matrix to find axes of variation in a dataset. Modern datasets record information across contexts and modalities. One example is the contrastive setting, which has two contexts: a foreground dataset, representing an experimental group, and a background dataset, representing a control group. The goal is to explain the foreground after “subtracting off" the background. With more than two contexts, we seek factors that can appear across multiple contexts. I will talk about 'multi-context principal component analysis' (MCPCA), a generalization of PCA to find axes of variation and the combinations of contexts in which they appear. I’ll also talk about work to relax the assumption that unobserved variables in a system are mutually independent. Both projects estimate parameters in latent variable models using new tensor decompositions. We’ll see applications to study gene expression data and contextualized word embeddings from language models.
Speakers Bio: Anna Seigal is an Assistant Professor of Applied Mathematics in the School of Engineering and Applied Sciences at Harvard, and an Affiliate in the Department of Statistics. Her research is in applied algebra. She was previously a Junior Fellow at the Society of Fellows at Harvard University. Before that, she held a Hooke Research Fellowship in the Mathematical Institute at the University of Oxford and a Junior Research Fellowship at The Queen's College. She received her PhD from UC Berkeley in 2019. She is a recipient of the SIAM Richard C. DiPrima Prize and an Alfred P. Sloan Research Fellowship in Mathematics.