Julia Kostin

About me

I am a PhD student at ETH Zurich in the Statistical Machine Learning Group, supervised by Prof. Fanny Yang. I am currently active in the fields of statistical learning, causal inference, information theory, mathematical signal processing and inverse problems.

Research interests

Statistics of overparametrized learning systems
Mathematics of high-dimensional data
Robustness of data-driven algorithms
Mathematical signal processing
Machine learning for neuroscientific discovery

News

06/2023 I started my PhD at ETH Zurich in the SML group.
03/2023 I graduated from the TopMath program at TU Munich.
03/2023 New preprint "How robust is randomized blind deconvolution via nuclear norm minimization against adversarial noise?" with Felix Krahmer and Dominik Stöger now available.

Publications and preprints

J. Kostin, F. Krahmer, and D. Stöger. How robust is randomized blind deconvolution via nuclear norm minimization against adversarial noise? [preprint]
M. Elminshawi, J. Kostin, E. A. P. Habets, and N. K. Sharma. Attended Talker Decoding Exclusively From Listening-State EEG.

Talks

Robustness of Low Rank Matrix Recovery under Adversarial Noise. 93rd Annual Meeting of the International Association of Applied Mathematics and Mechanics, Dresden, 02.06.2023.
Robustness Guarantees for Blind Deconvolution via Nuclear Norm Minimization. KU-LMU-TUM Joint Seminar on Mathematics of Data Science, Munich, 07.11.2022.
Convex Geometry Based Guarantees for Low-rank Matrix Recovery with Adversarial Noise. Approximation and geometry in high dimensions 2022, Bedlewo, Poland.

Teaching

In the past, I have tutored the following courses:

Winter Semester 2021/22: Linear Algebra for Electrical Engineering (TUM, Prof. Claudia Scheimbauer)
Summer Semester 2020: Multivariable Calculus and Topology (LMU, Prof. Peter Müller)
Winter Semester 2019/20: Measure Theory (LMU, Prof. Bernhard Leeb)
Summer Semester 2019: Calculus and Topology (LMU, Prof. Bernhard Leeb)