The Fred Hutch Biostatistics Program hosts seminars featuring presentations by Fred Hutch and outside scientists to share their latest developments and recent research. Each seminar includes an hour-long presentation and discussion during which speakers showcase their work and findings.
This seminar will be in-person.
Please contact email@example.com if this is not practical and you would like a Zoom link.
In traditional statistical analysis, including experimental design, features are typically well-defined by a set of random variables. However, this is not the case anymore in modern data analysis, where we often deal with a large number of random variables. Recent statistical advancements over the past two to three decades have predominantly concentrated on variable selection techniques. Nonetheless, in many scenarios, features are not clearly defined in the data and need to be obtained through learning methods before we conduct any analysis.
In this talk, we revisit a statistical methodology known as Sliced Inverse Regression (SIR) and leverage its potential to automatically learn features in two applications. The first application involves individualized treatment rules, wherein SIR defines a low-dimensional feature space for estimating conditional average treatment effects, so that we can make an individualized treatment rule. In our second application, we use SIR to learn invariant features across multiple data environments, facilitating out-of-distribution generalization. We showcase the relevance of this established statistical model and method within the contemporary realm of feature representation learning.