ETH Zurich
Seminar for Statistics (SfS)
Rämistrasse 101
8092 Zurich
Schweiz
Email: matthias.loeffler [at] stat.math.ethz.ch
I was a postdoc at the Seminar for Statistics at ETH Zürich, mentored by Prof. Sara van de Geer. As of August 2022 I work at QuantCo.
I am interested in the mathematics of data science, in particular high-dimensional statistics and theoretical machine learning. Currently, I am particularly interested in models with a discrete aspect, such as clustering or classification, and instrumental variable models.
Together with Mona Azadkia, Yuansi Chen, Geoffrey Chinot and Armeen Taeb I organized the virtual Young Data Science Researcher Seminar Zurich.
Short CV
Publications & Preprints
2019-2022: Postdoc at the Seminar for Statistics at ETH Zürich, mentor: Prof. Sara van de Geer
2019: PhD in Mathematics, University of Cambridge.
Thesis on Statistical inference in high-dimensional matrix models, advisor: Prof. Richard Nickl
2015: Master of Advanced Studies in Mathematical Statistics (Part III ), University of Cambridge.
Part III Essay about low rank recovery problems in Statistics, advisor: Prof. Richard Nickl
2014: Bachelor of Science in Mathematics with Minor in Economics, Ruprecht-Karls-Universität Heidelberg. Bachelor’s thesis about Lepski’s method for a variable choice of bandwidth (in German), advisor: Prof. Rainer Dahlhaus
11. Computationally efficient sparse clustering. (with A.S. Wein and A.S. Bandeira). Information & Inference, to appear
10. AdaBoost and robust one-bit compressed sensing. (with G. Chinot, F. Kuchelmeister and S. van de Geer). Mathematical Statistics and Learning, to appear
9. On the robustness of minimum-norm interpolators and regularized empirical risk minimizers. (with G. Chinot and S. van de Geer). Annals of Statistics, to appear
8. Spectral thresholding for the estimation of Markov chain transition operators. (with A. Picard). Electronic Journal of Statistics, 15(2) (2021), 6281-6310
7. Reconstruction of the neutrino mass as a function of redshift. (with C.S. Lorenz, L. Funcke and E. Calabrese). Physical Review D, 104(12) (2021), 123518
6. Optimality of Spectral clustering in the Gaussian mixture model. (with A.Y. Zhang and H.H. Zhou). Annals of Statistics, 49(5) (2021) 2506-2530
5. Efficient Estimation of Linear Functionals of Principal Components. (with V. Koltchinskii and R. Nickl). Annals of Statistics, 48(1) (2020) 464-490
4. Wald Statistics in high-dimensional PCA. ESAIM: Probability & Statistics 23 (2019) 662–671
3. Constructing confidence sets for the matrix completion problem. (with A. Carpentier and O. Klopp). Nonparametric Statistics -3rd ISNPS, Avignon, 2016,
Springer Proceedings in Mathematics and Statistics (2018)
2. Adaptive confidence sets for matrix completion. (with A. Carpentier, O. Klopp and R. Nickl). Bernoulli, 24(4A) (2018) 2429-2460
1. Comments on: 'High-dimensional simultaneous inference with the bootstrap‘. (with R. Nickl). TEST, 26(4) (2017) 731-733
Seminar & Conference Talks
AdaBoost learns sparse halfspaces robustly
56th CISS, Princeton, March 2022
AdaBoost and robust one-bit compressed sensing
Meeting in Mathematical Statistics 2021, CIRM, Marseille, France, December 2021
AdaBoost and robust one-bit compressed sensing
Conference on Mathematics of Machine Learning, Center for Interdisciplinary Research (ZiF), Bielefeld University, August 2021
Optimality of Spectral Clustering in the Gaussian mixture model
MHC 2021 - Conference on Mixtures, Hidden Markov Models & Clustering, Paris, France, June 2021
Optimality of Spectral Clustering in the Gaussian mixture model
Spring webinar series in statistics, Collegio Carlo Alberto, April 2021
Optimality of Spectral Clustering in the Gaussian mixture model
Stochastics Seminar, GeorgiaTech, March 2021
Optimality of Spectral Clustering in the Gaussian mixture model
Colloquium on Mathematical Statistics and Stochastic Processes, Universität Hamburg, January 2021
Computationally efficient sparse clustering
CMStatistics 2020, London, United Kingdom, December 2020
Computationally efficient sparse clustering
Forschungsseminar Mathematische Statistik, Humbold University, November 2020
Linear functionals in PCA and clustering
Workshop on high-dimensional covariance operators and their applications, Berlin, Germany, September 2019
Linear functionals in PCA and clustering
Statistics Seminar Université Paris Nanterrre, Paris, February 2019
Spectral Thresholding for the estimation of Markov chain transition operators
Meeting in Mathematical Statistics 2018, Frejus, France, December 2018
Spectral Thresholding for Markov chain transition operators
Forschungsseminar Mathematische Statistik, Humbold University, December 2018
Spectral Thresholding for the estimation of Markov chain transition operators
Probability and Statistics Seminar, Universität Mannheim, November 2018
Linear functionals in PCA and clustering
Statistics Seminar, Rutgers University, October 2018
Confidence sets for matrix completion
Young Researcher’s Meeting in Statistics, Berlin, Germany, September 2016