Seminar Scientific Computing (Dr. Tim Breitenbach)

Abstract:

In this talk, we explain and showcase our new data science methods ranging from statistics to integer optimization with three real world scenarios from life science and cloud computing. First, we use our newly developed principal feature analysis for selecting genes from a single-cell data set whose expression are essential for generating the different p henotypes f romt he data set. Furthermore, the procedure demonstrates one notion of explainable AI. Second, periodic patterns in time series representing the utilization of cloud computing resources are identified with autocorrelation and ML models. The periodic part of the time series is quantified w ith a newly defined c riterion. Third, we demonstrate how integer optimization can provide a systematic way to analyze genomic data to design synthetic DNA pieces that are used to report phenotypes in glioblastoma research and thus supporting to find new effective therapies.

All lectures in the 'Seminar Scientific Computing'