Weekly Bulletin
The FIM provides a Newsletter called FIM Weekly Bulletin, which is a selection of the mathematics seminars and lectures taking place at ETH Zurich and at the University of Zurich. It is sent by e-mail every Tuesday during the semester, or can be accessed here on this website at any time.
Subscribe to the Weekly Bulletin
FIM Weekly Bulletin
×
Modal title
Modal content
| Monday, 23 June | |||
|---|---|---|---|
| — no events scheduled — |
| Tuesday, 24 June | |||
|---|---|---|---|
| Time | Speaker | Title | Location |
| 16:15 - 17:30 |
Prof. Dr. Alex Weincall_made UC Davis |
Abstract
An order-3 tensor is an n-by-n-by-n array of real numbers. We consider
the task of decomposing a given tensor as the sum of rank-1 tensors,
using the minimal number of terms. This task has various applications
in statistics and data science, such as learning the latent parameters
of certain statistical models from the empirical third moment tensor.
Under the standard assumption that the tensor components are
"generic," a classical method called simultaneous diagonalization or
"Jennrich's algorithm" can decompose tensors of rank up to r <= n in
polynomial time. A recent result of Koiran (2024) improves this to r
<= 4n/3, and we improve this further to r <= 2n. The algorithm is
based on a non-trivial procedure for "flattening" tensors to matrices.
We also give a matching impossibility result, showing that no
flattening of the style we consider can surpass 2n. This may suggest a
fundamental barrier for fast algorithms.
DACO SeminarOvercomplete Tensor Decomposition via Koszul-Young Flatteningsread_more |
HG G 19.1 |
| Wednesday, 25 June | |||
|---|---|---|---|
| Time | Speaker | Title | Location |
| 09:25 - 09:45 |
Emmanuel Candès Stanford University |
Abstract
The mosaic permutation test: An exact and nonparametric goodness-of-fit test for factor |
HG F 3 |
| 09:50 - 10:10 |
Axel Munk Georg-August-Universität Göttingen |
Abstract
Statistical optimal transport- theory powering applications |
HG F 3 |
| 10:15 - 10:35 |
Abstract
tbd |
HG F 3 |
|
| 11:05 - 11:25 |
Richard J. Samworth University of Cambridge |
Abstract
Deep learning with missing data |
HG F 3 |
| 11:30 - 11:50 |
Mathias Drton Technische Universität München |
Abstract
Causal Modeling with Stationary Processes |
HG F 3 |
| 11:55 - 12:15 |
Ming Yuan Columbia University |
Abstract
A cumulant approach to linear regression |
HG F 3 |
| 12:20 - 12:40 |
Po-Ling Loh University of Cambridge |
Abstract
On the Benefits of Accelerated Optimization in Robust and Private Estimation |
HG F 3 |
| 14:00 - 14:20 |
Tony Cai University of Pennsylvania |
Abstract
Title T.B.A. |
HG F 3 |
| 14:25 - 14:45 |
Peter J. Bickel University of California, Berkeley |
Abstract
An old conjecture of David Blackwell |
HG F 3 |
| 14:50 - 15:10 |
Klaus-Robert Müller Technische Universität Berlin |
Abstract
Title T.B.A. |
HG F 3 |
| 15:40 - 16:00 |
Cun-Hui Zhang Rutgers University |
Abstract
Precise regret and adaptive inference in multi-armed bandits |
HG F 3 |
| 16:05 - 16:25 |
Zijian Guo Rutgers University |
Abstract
Multi-Source Learning with Minimax Optimization: From Adversarial Robustness to Causal Invariance |
HG F 3 |
| 16:30 - 16:50 |
Annie Qu University of California, Irvine |
Abstract
Representation Retrieval Learning for Heterogeneous Data Integration |
HG F 3 |
| Thursday, 26 June | |||
|---|---|---|---|
| Time | Speaker | Title | Location |
| 09:00 - 09:20 |
Bin Yu University of California, Berkeley |
Abstract
Implicit Regularization: GD in Boosting and Learning Rates in DL |
HG F 3 |
| 09:25 - 09:45 |
Rainer von Sachs Université catholique de Louvain |
Abstract
Time-varying degree-corrected stochastic block models |
HG F 3 |
| 09:50 - 10:10 |
Bernhard Schölkopf Max-Planck-Institut für Intelligente Systeme |
Abstract
Title T.B.A. |
HG F 3 |
| 10:15 - 10:35 |
Abstract
tbd |
HG F 3 |
|
| 11:05 - 11:25 |
David Blei Columbia University |
Abstract
A Bayesian Approach to Invariant Causal Prediction |
HG F 3 |
| 11:30 - 11:50 |
Dominik Rothenhäusler Stanford University |
Abstract
Data quality or data quantity? Prioritizing data collection under distribution shifts |
HG F 3 |
| 11:55 - 12:15 |
Regina Y. Liu Rutgers University |
Abstract
Fusion Learning: Combining Inferences from Diverse Data Sources |
HG F 3 |
| Friday, 27 June | |||
|---|---|---|---|
| Time | Speaker | Title | Location |
| 09:00 - 09:20 |
Harrison Huibin Zhou Yale University |
Abstract
From Score Estimation to Sampling |
HG F 3 |
| 09:25 - 09:45 |
Jelle Goeman Leiden University Medical Center |
Abstract
Title T.B.A. |
HG F 3 |
| 09:50 - 10:10 |
Rajen Shah University of Cambridge |
Abstract
Average partial effect estimation using double machine learning |
HG F 3 |
| 10:15 - 10:35 |
Liza Levina University of Michigan |
Abstract
Towards Interpretable and Trustworthy Network-Assisted Prediction |
HG F 3 |
| 11:05 - 11:25 |
Jianqing Fan Princeton University |
Abstract
When can weak latent factors be statistically inferred? |
HG F 3 |
| 11:30 - 11:50 |
Martin Wainwright Massachusetts Institute of Technology |
Abstract
Inference for black-box prediction |
HG F 3 |