Thema der Dissertation:
Design and multi-criteria optimization of cell classifier circuits in cancer therapy Thema der Disputation:
How to apply machine learning to biological data and sleep well: pitfalls and mitigation strategies
Design and multi-criteria optimization of cell classifier circuits in cancer therapy Thema der Disputation:
How to apply machine learning to biological data and sleep well: pitfalls and mitigation strategies
Abstract: Machine learning (ML) becomes omnipresent in almost every subfield of computational biology including genomics, proteomics and synthetic biology. However, estimation of an ML model's performance is usually as good as the underlying assumptions, as well as the employed data and evaluation strategies. Although machine learning provides an array of powerful approaches for solving various problems, their misuse may bring incorrect insights. In this talk, I will discuss the common pitfalls of applying machine learning to different biological questions touching on problems such as information leakage or distributional differences and present potential mitigation strategies.
Zeit & Ort
21.07.2023 | 13:00
Seminarraum 032
(Fachbereich Mathematik und Informatik, Arnimallee 6, 14195 Berlin)