Older news

Older News

(03/07/23) New paper on arXiv: Exploring the Limits of Indiscriminate Data Poisoning Attacks.
(03/07/23) I will be a reader of Niki Hasrati's Master's thesis.
(03/02/23) New paper on arXiv: Choosing Public Datasets for Private Machine Learning via Gradient Subspace Distance.
(02/28/23) I will be an AC for NeurIPS 2023.
(02/18/23) I will be the Social Media Chair for SaTML 2024.
(02/16/23) Mahbod Majid wins a Waterloo Faculty of Mathematics Graduate Research Excellence Award for the paper "Effcient Mean Estimation with Pure Differential Privacy via a Sum-of-Squares Exponential Mechanism."
(02/13/23) I was recognized as a notable reviewer for SaTML 2023.
(02/06/23) New paper on arXiv: Private GANs, Revisited.
(02/06/23) I'm reviewing for ICML 2023 workshops.
(02/05/23) Robustness Implies Privacy in Statistical Estimation accepted to STOC 2023.
(01/31/23) New paper on arXiv: A Bias-Variance-Privacy Trilemma for Statistical Estimation.
(12/24/22) Indiscriminate Data Poisoning Attacks on Neural Networks published in TMLR.
(12/22/22) New paper on arXiv: Hidden Poison: Machine Unlearning Enables Camouflaged Poisoning Attacks.
(12/20/22) A few talks in early 2023: AI4OPT seminar in Georgia Tech, differential privacy tutorial at SaTML in Raleigh, Rising Stars in AI Symposium at KAUST.
(12/20/22) I will be an AC for ICML 2023.
(12/18/22) I will be an AC for FAccT 2023.
(12/13/22) Two new papers on arXiv: Robustness Implies Privacy in Statistical Estimation, and Considerations for Differentially Private Learning with Large-Scale Public Pretraining.
(11/15/22) I was quoted in a Gizmodo article on AI-generated deepfakes.
(10/20/22) Papers accepted to NeurIPS 2022 workshops DistShift, SyntheticData4ML, Trustworthy and Socially Responsible ML, ML Safety.
(10/14/22) I'll be speaking at a session on robustness and privacy at JSM 2023.
(10/13/22) I will be a mentor for a the Fall 2022 Learning Theory Alliance Mentorship Workshop.
(10/12/22) I am a member of the senior PC for COLT 2023.
(9/29/22) Recently received an unrestricted gift from Apple to support work on private estimation. Thanks Apple!
(9/26/22) I am a member of the senior PC for ALT 2023.
(9/20/22) I'll be giving talks this term at Columbia Stats, Canadian AI Federated Learning Workshop (Toronto), Cornell CS Theory, ICSDS (Florence) and virtually at the US Census, Rutgers Business, Berkeley BLISS, and LinkedIn. Drop me a line if you're in any of the areas!
(9/19/22) I will be on the program committee for USENIX Security 2023.
(9/14/22) Two papers (1, 2) accepted to NeurIPS 2022.
(8/17/22) New paper on arXiv: Private Estimation with Public Data.
(8/15/22) I will be a reviewer for ICBINB 2022.
(8/11/22) I will be an AC for ICLR 2023.
(8/9/22) I was on Nathan Harms' PhD thesis committee.
(7/19/22) I'm an ethical reviewer for NeurIPS 2022.
(7/11/22) I'm giving a few talks in the upcoming weeks, at an operations research summer school (7/21, virtual), our Fields workshop on privacy (7/29, in Toronto), at Facebook (8/3, virtual), at University of Victoria (8/25, Victoria), and UBC (8/29, Vancouver).
(6/28/22) I am serving on the PhD thesis commitees of Bailey Kacsmar and Tim Dockhorn and Yiwei Lu.
(6/27/22) I will be giving an invited talk at ICORS on robust estimation for random graphs.
(6/27/22) I will be on the Master's thesis committees of Haolin Yu and Emily Lepert.
(6/24/22) I gave a talk at Google on private fine-tuning of large language models.
(6/6/22) New paper on arXiv: Per-Instance Privacy Accounting for Differentially Private Stochastic Gradient Descent.
(5/25/22) I will be on the program committee of FOCS 2023.
(5/17/22) New paper on arXiv: New Lower Bounds for Private Estimation and a Generalized Fingerprinting Lemma.
(5/15/22) One paper accepted to ICML 2022 with a long talk.
(5/14/22) Three papers (1, 2, 3) accepted to COLT 2022.
(5/11/22) I gave an invited tutorial at IWCIT 2022 on differential privacy. Slides are available here.
(4/22/22) Calibration with Privacy in Peer Review accepted to ISIT 2022.
(4/21/22) New paper on arXiv: Indiscriminate Data Poisoning Attacks on Neural Networks.
(4/20/22) I will be on the program committee of SaTML 2023.
(4/1/22) I will be a reader of Xinda Li's Master's thesis, and on Kelly Ramsay's PhD thesis committee.
(3/31/22) One paper accepted to FORC 2022, non-archival track.
(3/30/22) I will be an area chair for NeurIPS 2022.
(3/28/22) I'm giving talks at Waterloo's Probability seminar and University of Washington's theory seminar.
(3/18/22) Organizing two workshops at ICML 2022: TPDP 2022 and UpML 2022.
(3/15/22) I gave a talk at the Simons Data Privacy Reunion.
(3/1/22) I will give a (virtual) talk at Huawei Montreal on March 11.
(2/28/22) I will attending a Apple workshop on differential privacy on April 5 and 6.
(2/24/22) I gave a (virtual) talk at UMass Amherst's ML and Friends Lunch.
(2/8/22) I organized a junior-senior lunch mentoring session at FOCS 2021.
(2/3/22) Efficient Mean Estimation with Pure Differential Privacy via a Sum-of-Squares Exponential Mechanism accepted to STOC 2022.
(1/27/22) New paper on arXiv: Calibration with Privacy in Peer Review.
(1/20/22) Differentially Private Fine-tuning of Language Models accepted to ICLR 2022.
(12/20/21) Oral presentation at PPAI 2022 for Calibration with Privacy in Peer Review: A Theoretical Study.
(12/19/21) The Discrete Gaussian for Differential Privacy to appear in the Journal of Privacy and Confidentiality.
(12/6/21) I was a reader of Harry Sivasubramaniam's Master's thesis.
(12/1/21) The Role of Adaptive Optimizers for Honest Private Hyperparameter Selection accepted to AAAI 2022 (oral presentation).
(11/28/21) New paper on arXiv: Efficient Mean Estimation with Pure Differential Privacy via a Sum-of-Squares Exponential Mechanism.
(11/17/21) Best reviewer award at CCS 2021.
(11/9/21) New paper on arXiv: Robust Estimation for Random Graphs.
(11/9/21) New paper on arXiv: The Role of Adaptive Optimizers for Honest Private Hyperparameter Selection.
(11/8/21) New paper on arXiv: A Private and Computationally-Efficient Estimator for Unbounded Gaussians.
(10/27/21) A few talks coming up: at a Google Federated Learning workshop, an IDEAL robustness workshop, the ML Collective Reading Group , and a BIRS workshop.
(10/27/21) Unbiased Statistical Estimation and Valid Confidence Intervals Under Differential Privacy posted to arXiv.
(10/14/21) Differentially Private Fine-tuning of Language Models posted to arXiv.
(9/28/21) Two papers (1, 2) accepted to NeurIPS 2021.
(9/27/21) I will be a senior member of the program committee for COLT 2022.
(9/4/21) I will be on the program committee for ICBINB 2021.
(8/19/21) Our work was feature in a WIRED magazine article on machine unlearning.
(8/19/21) I will be on the program committee for PriML 2021.
(8/10/21) I will be a Senior Program Committee member for AAAI 2022.
(7/10/21) I will be on the PC for ALT 2022.
(7/9/21) I will be on Tosca Lechner's and Jimit Majmudar's Ph.D. thesis committees.
(6/28/21) I will be a reader for Thomas Humphries's Master's thesis.
(6/28/21) New paper on arXiv: The Price of Tolerance in Distribution Testing.
(6/16/21) I will be an area chair for ICLR 2022.
(6/2/21) New paper on arXiv: Improved Rates for Differentially Private Stochastic Convex Optimization with Heavy-Tailed Data.
(5/28/21) I will be on Guojun Zhang's Ph.D. thesis committee.
(5/8/21) PAPRIKA accepted to ICML 2021.
(4/27/21) I will be a reader for Nivasini Ananthakrishnan's Master's thesis.
(4/20/21) I am editor for ALT Highlights.
(4/9/21) I gave a talk at the Boston-area DP Seminar.
(4/2/21) Rachel Cummings and I are co-chairing the 2021 workshop on Theory and Practice of Differential Privacy (TPDP 2021), to be held at ICML 2021.
(4/1/21) PAPRIKA accepted to FORC 2021 (non-archival track).
(4/1/21) I gave a talk at the TrustML Seminar.
(3/24/21) I will be a reader for Lingyi Zhang's Master's thesis.
(3/18/21) I will be an area chair for NeurIPS 2021.
(3/8/21) New paper on arXiv: Remember What You Want to Forget: Algorithms for Machine Unlearning.
(3/2/21) I'm on the PC for SODA 2022.
(3/1/21) I'm giving a talk at the (virtual) London Symposium on Information Theory (LSIT), held on May 19 to 21.
(3/1/21) I'm giving a talk at a conference on robustness and privacy on March 22 or 23.
(2/19/21) I'm a reviewer for ICML 2021 workshops.
(1/27/21) I'm giving a talk at Google on March 5.
(1/27/21) I'm organizing a workshop on Distributed and Private Machine Learning at ICLR 2021.
(1/18/21) I gave a talk at Waterloo's ML + Logic Seminar.
(1/17/21) I will be a social chair for NeurIPS 2021.
(1/11/21) I will be giving a talk at McGill's statistics seminar.
(1/7/21) I hosted Graduating Bits at ITCS 2021, and helped out with a few other things behind the scenes.
(1/4/21) I started mirroring my videos on Bilibili, the "Chinese Youtube". A popular WeChat account wrote about my joining (in Chinese).
(12/22/20) I am now a faculty affiliate at the Vector Institute.
(12/21/20) One paper accepted to ALT 2021.
(12/16/20) I will be attending a (virtual) differential privacy workshop at Google in February.
(12/9/20) I was a mentor at Women in Machine Learning.
(12/7/20) I will be on the program committee of ESA 2021.
(12/6/20) Aloni Cohen and I will be running a social on Data Privacy: Academia, Industry, Policy, and Society at NeurIPS 2020.
(12/4/20) I will be on the program committee of the Privacy and Anonymity track for CCS 2021.
(11/22/20) Private Hypothesis Selection accepted to IEEE Transactions on Information Theory.
(11/20/20) I gave a talk on differentially private statistics at a reading group at the Simons Institute.
(10/24/20) Two papers (1, 2) to be presented at the workshop PPML 2020.
(10/20/20) New paper on arXiv: On the Sample Complexity of Privately Learning Unbounded High-Dimensional Gaussians.
(10/19/20) New paper on arXiv: Enabling Fast Differentially Private SGD via Just-in-Time Compilation and Vectorization. Code is available here.
(10/18/20) Excited to be speaking at the Trustworthy ML Seminar on April 1, 2021.
(9/30/20) One paper accepted to SODA 2021.
(9/26/20) I'm organizing a differential privacy workshop at the Fields Institute in June 2022.
(9/26/20) I will be giving a talk at University of Toronto's Theory Seminar on October 16.
(9/25/20) Three papers accepted to NeurIPS 2020: one as spotlight (1), two as posters (1, 2).
(9/16/20) I will be an area chair for COLT 2021.
(8/23/20) I will be visiting Wharton's Statistics department and giving a talk on September 2.
(8/21/20) I visited Northwestern's IDEAL and gave a talk.
(8/19/20) I will be on the program committee for FORC 2021.
(8/19/20) I gave a talk at CMU's theory lunch. A recording of the full version of this talk is available here.
(8/17/20) I will be on the program committee for PPML 2020.
(8/10/20) My lab, The Salon, has a website!
(8/10/20) I recorded a talk on CoinPress. I gave a similar talk at JSM 2020 and the Harvard Privacy Tools group meeting.
(8/9/20) Five papers (1, 2, 3, 4, 5) will be presented at the workshop TPDP 2020.
(7/18/20) We launched DifferentialPrivacy.org!
(7/12/20) I have been awarded an NSERC Discovery Grant, including an Accelerator Supplement. See these articles for some more details.
(7/12/20) I will be a reader of Kaiwen Wu's Master's Thesis.
(7/10/20) I will be a reader of Beracira Chen's Master's Thesis.
(7/9/20) I will be an area chair for ICLR 2021.
(6/30/20) We wrote a "Behind the Screens" look at STOC 2020, a document which may be useful to other virtual conference organizers.
(6/18/20) I will be on the program committee for ALT 2021.
(6/15/20) I will be on the program committee for TPDP 2020.
(6/11/20) New paper on arXiv: CoinPress: Practical Private Mean and Covariance Estimation. Code is available here.
(5/31/20) One paper accepted to ICML 2020.
(5/29/20) I will serve on Vikrant Singhal's thesis committee.
(5/25/20) Two papers (1, 2) accepted to COLT 2020.
(5/13/20) I gave a talk on robustness at Waterloo's Algorithms and Complexity seminar.
(5/11/20) I will be a reader of Amur Ghose's Master's Thesis.
(5/9/20) I will be a reader of Shubhankar Mohapatra's Master's thesis.
(5/5/20) I am now a general co-chair for STOC 2020, helping out with the transition to a virtual conference.
(4/17/20) Jonathan Ullman and I wrote a brief survey of differentially private statistics: A Primer on Private Statistics.
(4/1/20) New paper on arXiv: The Discrete Gaussian for Differential Privacy. Code available here.
(3/9/20) Xi He and I were awarded a 2020 Resources from Research Groups grant from Compute Canada. Thank you Compute Canada!
(3/7/20) I was recognized as one of six top instructors for graduate courses in Computer Science in Fall 2019 (for CS 761).
(2/27/20) New paper on arXiv: PAPRIKA: Private Online False Discovery Rate Control. Code is available here.
(2/25/20) INSPECTRE: Privately Estimating the Unseen accepted to the Journal of Privacy and Confidentiality, Special Issue for TPDP 2018.
(2/25/20) I will be speaking in a session on differential privacy (organized by Weijie Su) at the 2020 Joint Statistical Meetings.
(02/23/20) Three new papers on arXiv! All on differential privacy, and include learning MRFs, mean estimation for heavy-tailed distributions, and hypothesis selection (in the local model).
(2/10/20) I will serve on Amit Levi's thesis committee.
(1/23/20) I will be on the program committee of Random 2020.
(1/7/20) I will be organizing a session at ITA 2020.
(1/3/20) I will be a reader of Sachin Vernekar's Master's thesis.
(1/2/20) I will be a reader of Sushant Agarwal's Master's thesis.
(11/18/19) New paper (Random Restrictions of High-Dimensional Distributions and Uniformity Testing with Subcube Conditioning) posted to arXiv.
(10/16/19) I will be serving as a guest editor for the special issue of TALG for SODA 2020.
(10/8/19) Shubhankar Mohapatra won one of three $1,000 best poster prizes at University of Waterloo's Cybersecurity and Privacy Institute Annual Conference, presenting on a project with me and Xi He.
(10/3/19) I will be an area chair for ICML 2020.
(10/3/19) Clément Canonne and I are organizing a workshop titled "A TCS Quiver" at FOCS 2019.
(10/1/19) Two posters at PriML 2019, on privately learning graphical models, and testing multivariate distributions.
(9/9/19) New paper (Differentially Private Algorithms for Learning Mixtures of Separated Gaussians) posted to arXiv.
(9/3/19) Two papers (Private Hypothesis Selection, Differentially Private Algorithms for Learning Mixtures of Separated Gaussians) accepted to NeurIPS 2019.
(8/26/19) I will be teaching CS 761 - Randomized Algorithms in Fall 2019.
(8/22/19) Robust Estimators in High Dimensions without the Computational Intractability was invited to Communications of the ACM as a Research Highlight.
(8/10/19) Two posters at TPDP 2019 (on privately learning graphical models, and testing multivariate distributions), and one oral presentation (Private Hypothesis Selection).
(8/7/19) I will be on the program committee of ICALP 2020.
(8/2/19) I will be on the program committee of PriML 2019, a workshop at NeurIPS 2019.
(7/13/19) I will be giving a talk at NTUA's Corelab Seminar on July 18.
(7/9/19) Testing Ising Models accepted to IEEE Transactions on Information Theory.
(7/2/19) I will be attending and organizing a session at WALE 2019.
(7/1/19) I started as an Assistant Professor at University of Waterloo.
(6/17/19) I started a blog.
(6/2/19) New paper (Private Hypothesis Selection) posted to arXiv.
(6/1/19) I was recognized as a top 5% reviewer for ICML 2019.
(5/28/19) New paper (Private Identity Testing for High-Dimensional Distributions) posted to arXiv.
(5/14/19) Code for Sever (in ICML 2019) has been made public.
(4/30/19) Robust Estimators in High-Dimensions Without the Computational Intractability published in SIAM Journal on Computing (Special Section on FOCS 2016).
(4/25/19) I gave a talk on robust and private estimation at Google Seattle's Cerebra Journal Club.
(4/21/19) One paper (Sever: A Robust Meta-Algorithm for Stochastic Optimization) accepted to ICML 2019.
(4/18/19) One paper (Privately Learning High-Dimensional Distributions) accepted to COLT 2019.
(4/8/19) Audra gave a talk about our paper on private hypothesis testing.
(3/22/19) I will be on the program committee of TPDP 2019.
(3/8/19) I gave a talk at the Simons Institute workshop on Data Privacy: Foundations and Applications. Video is available here.
(2/21/19) I will be giving a talk at the Berkeley Theory Lunch on March 20, 2019.
(2/10/19) I will be giving a talk at ITA 2019 on February 14, 2019.
(2/8/19) One paper (The Structure of Optimal Private Tests for Simple Hypotheses) accepted to STOC 2019.
(1/29/19) I gave a talk on private statistics from a TCS perspective, as part of the private statistics mini-course in the Data Privacy Boot Camp. Video is here, and the slides are here.
(1/23/19) I am a maintainer of the CS Theory Blog Aggregator.
(1/22/19) I will be giving a talk at MIT's Algorithms and Complexity Seminar on February 19, 2019.
(1/21/19) I will be giving a talk at Berkeley's BLISS Seminar on February 25, 2019.
(1/14/19) I will be on the PC of SODA 2020.
(12/19/18) A Chasm Between Identity and Equivalence Testing with Conditional Queries published in Theory of Computing.
(12/13/18) I will be giving a talk during the Simons Institute Data Privacy Boot Camp on January 29, 2019.
(12/10/18) I will be giving a talk at Caltech's Mathematics of Information Seminar on January 22, 2019.
(11/27/18) New paper (The Structure of Optimal Private Tests for Simple Hypotheses) posted to arXiv.
(11/13/18) One paper (Sever: A Robust Meta-Algorithm for Stochastic Optimization) to be presented at SECML 2018, oral presentation.
(11/2/18) Video of my talk today on Realizing Robustness is available.
(10/25/18) I will be giving a talk on Realizing Robustness at the Simons Institute workshop on Robust and High-Dimensional Statistics on November 2.
(9/27/18) One paper (Anaconda: A Non-Adaptive Conditional Sampling Algorithm for Distribution Testing) accepted to SODA 2019.
(9/3/18) I was recognized as one of the 30% highest-scoring reviewers for NeurIPS 2018.
(8/31/18) My Ph.D. thesis on Modern Challenges in Distribution Testing is complete and submitted! Besides my own work, it should serve as a good survey of recent works in the field.
(8/22/18) Two papers (1, 2) to be presented at TPDP 2018.
(8/17/18) I gave a talk at the Workshop on Computational Efficiency and High-Dimensional Robust Statistics. Slides from my talk are available here.
(7/17/18) New paper (Anaconda: A Non-Adaptive Conditional Sampling Algorithm for Distribution Testing) posted to arXiv.
(6/19/18) I have been named a Microsoft Research Fellow during my Simons-Berkeley Research Fellowship at the Simons Institute.
(6/18/18) I successfully defended my PhD thesis! Video of the defense is available here.
(5/11/18) One paper (INSPECTRE: Privately Estimating the Unseen) accepted to ICML 2018.
(5/3/18) Video from my talk at the BIRS Workshop on Mathematical Foundations of Data Privacy is now available here.
(5/1/18) New paper (Privately Learning High-Dimensional Distributions) posted to arXiv.
(5/1/18) One paper (Actively Avoiding Nonsense in Generative Models) accepted to COLT 2018.
(4/27/18) Video from my talk at the CRM workshop on Modern Challenges in Learning Theory is now available here.
(4/20/18) I will join the University of Waterloo's Cheriton School of Computer Science in July 2019!
(4/20/18) I will be a Simons-Berkeley Research Fellow at the Simons Institute for the Theory of Computing for the Fall 2018 semester program on Foundations of Data Science and the Spring 2019 semester program on Data Privacy: Foundations and Applications.
(4/12/18) I will be attending a BIRS workshop on Mathematical Foundations of Data Privacy from April 29, 2018 to May 4, 2018.
(4/11/18) I gave a talk at MIT's LIDS and Stats tea on April 11, 2018.
(3/31/18) I will be giving an invited talk at a workshop on Modern Challenges of Learning Theory on April 26, 2018.
(3/23/18) I gave an invited talk at CISS 2018 on March 23, 2018, in a session organized by Jiantao Jiao and Tsachy Weissman.
(3/15/18) Code for Being Robust (in High Dimensions) Can Be Practical posted to GitHub.
(3/8/18) Code for Concentration of Multilinear Functions of the Ising Model with Applications to Network Data posted to GitHub.
(3/7/18) New paper (Sever: A Robust Meta-Algorithm for Stochastic Optimization) posted to arXiv.
(3/1/18) New paper (INSPECTRE: Privately Estimating the Unseen) posted to arXiv.
(2/16/18) Code for INSPECTRE posted to GitHub.
(2/23/18) I will be attending a workshop on Computational Efficiency & High-Dimensional Robust Statistics from August 13-17, 2018.
(2/20/18) New paper (Actively Avoiding Nonsense in Generative Models) posted to arXiv.
(2/16/18) Code for Priv'IT posted to GitHub.
(2/15/18) I will be giving a talk at the Boston University Computer Science Seminar on February 21, 2018.
(1/5/18) I will be giving a talk at the University of Waterloo Computer Science Seminar on February 1, 2018.
(1/5/18) I will be giving a talk at the McGill Computer Science Seminar on February 5, 2018.
(11/2/17) A Chasm Between Identity and Equivalence Testing with Conditional Queries to appear in the journal Theory of Computing (previously in RANDOM 2015).
(10/31/17) I will be giving a talk at the Boston University Theory Seminar on November 17, 2017.
(9/29/17) Three papers (1, 2, and 3) accepted to SODA 2018.
(9/26/17) I will be giving a talk at Cornell Theory Tea on September 28, 2017.
(9/20/17) I will be giving a talk at the McMaster University Computing and Software Department Seminar on October 5, 2017.
(9/14/17) I will be giving a talk at the MIT Theory Lunch on October 26, 2017.
(9/14/17) I will be be giving a talk on distribution testing at the UMass Amherst Theory Seminar on October 24, 2017.
(9/7/17) Clément Canonne and I are organizing a workshop on frontiers in distribution testing at FOCS 2017, on October 14, 2017.
(9/4/17) One paper (Concentration of Multilinear Functions of the Ising Model with Applications to Network Data) accepted to NIPS 2017.
(8/1/17) I will be be giving a talk on distribution testing at the Cornell Theory Seminar on September 25, 2017.