Columbia HomeRobust Statistics and Privacy Workshop
Thursday, October 5th - Friday, October 6th
Time: See Program Schedule
Locations:
Faculty House Garden Room 2 on Thursday, October 5
Room C03 SSW on Friday, October 6
Description
The goal of this workshop will be to present recent developments in robust statistics and differential privacy, including algorithmic foundations, concentration results, inference and applications to high dimensional statistics and machine learning.
Thursday, October 5, 2023
Location: Faculty House, Garden Room 2
Friday, October 6, 2023
Location: School of Social Work Building, Room C03 (C-level)
Speakers
- Jason Altschuler (University of Pennsylvania)
- Victor-Emmanuel Brunel (ENSAE-ParisTech)
- Samuel Hopkins (MIT)
- Gautam Kamath (University of Waterloo)
- Po-Ling Loh (Cambridge University)
- Gabor Lugosi (University Pompeu Fabra)
- Stanislav Minsker (University of Southern California)
- Roberto Imbuzeiro Oliveira (IMPA)
- Lekshmi Ramesh (Columbia University)
- Zoraida Rico (Columbia University)
- Elvezio Ronchetti (University of Geneva)
- Adam Smith (Boston University)
- Weijie Su (University of Pennsylvania)
- Yi Yu (University of Warwick)
Thursday, October 5
Book of Abstracts
Location: Faculty House, Garden Room 2
Coffee: 9:30-10:00am
10:00 – 12:00 Session 1
Elvezio Ronchetti (University of Geneva)
Some Lessons from Classical Robust Statistics with an Outlook to Some New Developments
Victor Emmanuel Brunel (ENSAE ParisTech)
Geodesically convex M-estimation in metric spaces
Stanislav Minsker (University of Southern California)
Improved performance guarantees for the median of means estimator
12:00-1:30
Lunch Break
1:30-2:50 Session 2
Yi Yu (University of Warwick)
Robust mean change point testing in high-dimensional data with heavy tails
Lekshmi Ramesh (Columbia University)
Empirical Risk Minimization Under User-Level Local Privacy Constraints
2:50-3:10
Coffee Break
3:10-4:30 Session 3
Weijie Su (University of Pennsylvania)
Gaussian Differential Privacy and How to Enhance Census Data Privacy for Free
Samuel Hopkins (MIT)
The Full Landscape of Robust Mean Testing: Sharp Separations between Oblivious and Adaptive Contamination.
Friday, October 6
Book of Abstracts
Location: Room C03 (C-level), School of Social Work Building
Coffee: 9:30 – 10:00
10:00 – 12:00 Session 1
Gabor Lugosi (Universitat Pompeu Fabra)
On the quality of randomized approximations of Tukey’s depth.
Roberto Oliveira (IMPA)
Trimmed sample means for uniform mean estimation and regression
Zoraida Rico (Columbia University)
Fine bounds on covariance estimation
12:00-1:30
Lunch Break
1:30-2:50 Session 2
Po-Ling Loh (Cambridge University)
Differentially private penalized M-estimation via noisy optimization
Jason Altschuler (University of Pennsylvania)
Shifted divergences for sampling, privacy, and beyond
2:50-3:10
Coffee Break
3:10-4:30 Session 3
Gautam Kamath (University of Waterloo)
Robust Estimators for Private Estimation
Adam Smith (Boston University)
Fast, Private (Sub) Gaussian Estimation