PhD
Chair for Statistical Learning and Data Science at LMU
LMU Munich
Department of Statistics
Ludwigstr. 33
80539 München
Biosketch
Lisa obtained a Bachelor’s and Master’s degree in Statistics at LMU Munich. She joined the Chair for Statistical Learning and Data Science as a student assistant and proceeded to start her PhD on “uncertainty quantification in machine learning” in February 2022. Since October of the same year, she has been a member of relAI.
relAI Research
Uncertainty quantification in machine learning
My research revolves around predictive uncertainty in machine learning and deep learning in particular. Providing trustworthy predictions requires (1) a meaningful representation and (2) faithful quantification of uncertainty. To address the former, I’ve been working on facilitating sampling-based inference in neural networks, which is computationally demanding, by leveraging parameter symmetries. With regards to the latter, I’ve contributed to challenging, and improving upon, the current de facto standard of entropy-based uncertainty measures. Both endeavors are severely inhibited by a lack of ground-truth uncertainty to compare against, suggesting that uncertainty quantification will remain an active field of research for some time to come.