Geometric and Multiscale Methods for Statistical Learning
Presidential Lectures are free public colloquia centered on four main themes: Biology, Physics, Mathematics and Computer Science, and Neuroscience and Autism Science. These curated, high-level scientific talks feature leading scientists and mathematicians and are intended to foster discourse and drive discovery among the broader NYC-area research community. We invite those interested in the topic to join us for this weekly lecture series.
Large, high-dimensional datasets appear in a wide variety of applications. Extracting information from these datasets and performing machine-learning tasks on them can be challenging for both fundamental statistical reasons and because of computational barriers.
In this lecture, Mauro Maggioni will discuss a family of ideas, algorithms and results for learning from high-dimensional data. These methods rely on the idea that complex, high-dimensional data has geometric structures that, once discovered, assist in a variety of tasks, including statistical learning and data visualization. He will focus on multiscale decompositions that can be used to solve problems such as dictionary learning, classification and regression. These decompositions lead to the construction of novel probabilistic models for data, new notions of learning and approximation of high-dimensional stochastic systems.
If this lecture is videotaped, it will be posted here after production.