Abstract:
Biomedical datasets are often wide: they contain many more features than examples. In thecase of rare diseases, this challenge is particularly pronounced. I’ll discuss how transferring unsupervised machine learning models to rare disease datasets can help to tackle this problem. Success on these types of problems requires high-quality features.
As single-cell RNA-seq is a rapidly growing area of methodological work, and one where we can finally get more examples than features, I’ll discuss a number of recent neural network based methods for feature construction on scRNA-seq data. I’ll discuss how performance comparisons in this space can be particularly challenging.
Coffee and snacks will be available before the Seminar begins
Tuesday, October 2, 2018 at 2:00pm to 3:00pm
CSM Building, Davis Auditorium
Community, Faculty, Postdocs, Staff, Students, Health Care Professionals
Medical Education, Graduate Education, Postdoctoral Affairs, Residencies, Fellowships
N/A
Casey S. Greene, PhD