ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
Google Calendar Link (updated every monday): https://calendar.google.com/calendar/u/0?cid=YXF1cTUzOGZzZWl1c2hidWJrOXN0aHVzb2NAZ3JvdXAuY2FsZW5kYXIuZ29vZ2xlLmNvbQTalks in group meeting should be ~30min
2
3
(MAD+ roughly biweekly, 16-18 Fridays)Notes
4
Zurich Time15:00-16:0016:00-17:00 (except Tuesdays)17:00-18:00 (except Tuesdays)Talks in group meeting should be ~30min
Talks are sometimes moved if there is a seminar
5
18 January 2021MondayInformal Meeting
6
22 January 2021Friday
7
25 January 2021Monday
8
29 January 2021Friday
9
01 February 2021Monday
10
5 February 2021FridayGroup MeetingDylan https://arxiv.org/pdf/2101.04036.pdfShrey Aryan
11
8 February 2021Mondayavoid if possible (just 15-16)
12
12 February 2021FridayYDS Seminar (Tengyao Wang)avoid if possibleavoid if possible
https://math.ethz.ch/sfs/news-and-events/young-data-science.html?s=fs21
13
15 February 2021Mondayavoid if possible (just 15-16)Group Meetingavoid if possible
14
19 February 2021FridayYDS Seminar (Evgenii Chzhen)Seminar by Amir Asadi. https://ethz.zoom.us/j/511336187
Title: Neural Networks and Multiscale Entropies (Zoom https://ethz.zoom.us/j/511336187)
Abstract: One of the most striking properties of our world is the enormous variety of length scales in its structure. Complex physical phenomena and real-world signals and images contain structures at different scales. In this talk, we study distributions that, given a mean constraint, simultaneously maximize entropies at different scales. This is motivated by the goal of analyzing neural networks while exploiting their distinguishing characteristic, that is, their multi-level architecture. Neural networks have found remarkable success in many areas of science such as image classification, yet finding a proper theoretical ground for designing their architecture and training procedure remains an active research area. In this talk, we define the notion of multiscale entropic regularization and demonstrate how it naturally applies to the multilevel architecture of neural networks. We then propose simulating a multiscale extension of the Gibbs-Boltzmann distribution as a training procedure that is an alternative to stochastic gradient descent and is based on the chain rule of entropy rather than the chain rule of derivatives. This information-theoretic approach is inspired by both multiscale techniques of chaining in high-dimensional probability and renormalization group theory in statistical physics. It can also be viewed as an extension of the maximum entropy results of Jaynes (1957) and Csiszár (1975) and a Bayesian expansion of the renormalization group.

Joint work with Emmanuel Abbe (EPFL).

Bio:
Amir R. Asadi is a final year Ph.D. candidate in the Department of Electrical and Computer Engineering at Princeton University where he is advised by Emmanuel Abbe. He received B.Sc. degrees in Electrical Engineering and in Mathematics from Sharif University of Technology in Iran, both in 2015, and M.A. degree in Electrical Engineering from Princeton University in 2017. The recipient of the Anthony Ephremides Fellowship from Princeton University in 2016, his research is on the interface of machine learning, information theory, and high-dimensional probability.
15
22 February 2021Mondayavoid if possible (just 15-16)
Seminar announced by Fanny (see post on channel)
group meeting / informal meeting (Afonso can't make it)
Start of Semester
16
26 February 2021FridayNikita ZhivotovskiyYDS Seminar (Feng Ruan) FOCM seminar
http://focm-society.org/events.php and https://math.ethz.ch/sfs/news-and-events/young-data-science.html?s=fs21
17
1 March 2021Mondayavoid if possible (just 15-16)Geoffrey Chinot
Group Meeting and chat about Free Probability Reading Group
Toward a general analysis for minimum norm interpolators
18
5 March 2021FridayYDS Seminar (Aaditya Ramdas) at 14:30Antoine Maillard (at 15:30)Journal Club Preparation Session
Series of papers on the optimization of spin glasses : https://arxiv.org/abs/1812.04588, https://arxiv.org/abs/2001.00904, https://arxiv.org/abs/2009.11481
19
8 March 2021Mondayavoid if possible (just 15-16)Journal Club Talk (Bruno Loureiro)Journal Club Talk (Bruno Loureiro)
20
FridayKonstantin Donhauser
YDS (Christos Thrampoulidis) Seminar at 16:30-17:30
avoid if possible (because YDS ends only at 17:30)
21
15 March 2021Mondayavoid if possible (just 15-16)Journal Club Preparation SessionReserved for Free ProbabilityDST Offset -- (Afonso can't make it)
22
FridayNikita ZhivotovskiyYDS Seminar (Stephen Bates)DST Offset
23
22 March 2021Mondayavoid if possible (just 15-16)DST Offset
24
FridayYDS Seminar (Dana Yang)Journal Club Talk (Sidhanth Mohanty)Journal Club Talk (Sidhanth Mohanty)
DST Offset // FOCM Seminar this day: http://focm-society.org/events.php
25
29 March 2021Mondayavoid if possible (just 15-16)Fanny Yang ITS Colloquium (16:15)Free probability RG, after Fanny's talk
https://eth-its.ethz.ch/activities/its-science-colloquium/fan-yang.html
26
Friday
27
05 April 2021MondayEaster Break
28
Friday
29
12 April 2021Mondayavoid if possible (just 15-16)Reserved for Free ProbabilityAfonso - NCK (Informal)
30
FridayYDS Seminar (Spencer Frei) (16:30)MAD+ seminar
31
19 April 2021MondaySechseläuten
32
Friday
Reserved for Free Probability /YDS Seminar (Yuqi Gu)
MDA+ Lorin CrawfordMDA+ Lorin Crawford
33
26 April 2021Mondayavoid if possible (just 15-16)Journal Club prepMartin Uhrin [EPFL]
34
Friday
Journal Club: Maria Refinetti (https://arxiv.org/pdf/2102.11742.pdf)
Journal Club: Maria Refinetti (https://arxiv.org/pdf/2102.11742.pdf)
FOCM Seminarhttp://focm-society.org/events.php
35
3 May 2021Mondayavoid if possible (just 15-16)Anastasis TalkAlexandre Talk
36
Tuesday
On Tuesdays the slots are 11am-12pm and 12pm-1pm:
Informal Group Meeting (Starts at 11:15am)
Xingyu Xu (guest talk)
37
7 May 2021FridayYDS Seminar (Cheng Mao)Inbar MAD+Inbar MAD+Will likely include gather.town gathering
38
10 May 2021Mondayavoid if possible (just 15-16)Journal Club prep (Gabriel)Reserved for Free Probability
39
Tuesday
On Tuesdays the slots are 11am-12pm and 12pm-1pm:
Matthias Talk - AdaBoost and 1-bit-CSGather.Town informal gathering
40
FridayYDS Seminar (Boris Muzellec)
Journal Club: Tsigler (https://arxiv.org/pdf/1906.11300.pdf)
Journal Club: Tsigler (https://arxiv.org/pdf/1906.11300.pdf)
41
17 May 2021Mondayavoid if possible (just 15-16)Reserved for Free Probability - Chapter 8Afonso can't join on the 16h-17h slot
42
Tuesday
On Tuesdays the slots are 11am-12pm and 12pm-1pm:
Afonso - NCKMittagseminar (12:15pm)
43
FridayYDS Seminar (Johannes Wiesel)No meetingNo Meeting
44
24 May 2021Monday
Joel Tropp is giving a colloquium in CS: https://inf.ethz.ch/news-and-events/events/event-detail.joel-tropp-finding-structure-with-randomness.61221.html
Whitsuntide
45
Tuesday
On Tuesdays the slots are 11am-12pm and 12pm-1pm:
Introduction to the resolvent method in RMT
Mittagseminar (12:15pm)
46
FridayYDS Seminar (Morgane Austern)Siddharth Setlur (Interlacing Families)FOCM Seminarhttp://focm-society.org/events.php
47
31 May 2021Mondayavoid if possible (just 15-16)No meetingReserved for Free Probability
48
Tuesday
On Tuesdays the slots are 11am-12pm and 12pm-1pm:
No meetingMittagseminar (12:15pm)
49
4 June 2021Friday
Community detection using t-SNE /YDS Seminar (16:30)
Konstantin https://arxiv.org/pdf/2104.04244.pdf
End of Semester
50
51
52
53
54
55
56
57
58
2020
59
60
Monday 27 April 2020Group TestingFriday 1 May 2020
61
Monday 4 May 2020Davin ChooFriday 8 May 2020Afonso can't attend on May 8
62
Monday 11 May 2020Friday 15 May 2020Nikita Zhivotovskiy
63
Monday 18 May 2020Friday 22 May 2020
64
Monday 25 May 2020Friday 29 May 2020Jean-Baptiste Seby
65
Monday 1 June 2020Friday 5 June 2020Chat
66
Tuesday 9 June 2020Guarantees for EM
67
Tuesday 16 June 2020Average Hardness Reductions
68
Tuesday 23 June 2020Compressed Sensing w Neural Nets.
69
Wednesday 1 July 20203PM: Informal
70
Monday 6 July 2020Orthogonal group synchronization
71
Monday 13 July 2020LP relaxations for clustering
72
Monday 20 July 2020MLE & Moment Method in Low SNR GMMs
73
Monday 27 July 2020Novel min-max forms of linear inverse problems
74
Monday 3 August 2020
75
Monday 10 August 2020
76
Monday 17 August 2020
77
Monday 24 August 2020
78
Monday 31 August 2020
79
Monday 7 September 2020Multireference alignment in high dimensions
80
Wednesday October 21 2020Concentration Ineq for Random Tensors
81
Wednesday October 28 2020A Simple Measure of Conditional Dependence
82
Wednesday November 4 2020Phase synchronization and non-convexity
83
Wednesday, November 11, 2020Phase Retrieval with Structured Sensing Matrices
84
Wednesday November 18 2020
85
Wednesday November 25 2020
Group testing and local search: is there a computational-statistical gap?
86
Wednesday December 2 2020
Reading on the stochastic localization proof techniques for the KLS conjecture, based on the paper Eldan’s Stochastic Localization and the KLS Hyperplane Conjecture:
An Improved Lower Bound for Expansion
,
87
Wednesday December 9 2020
88
Wednesday, December 16, 2020
89
90
91
92
93
94
95
96
97
98
99
100