Secure, Privacy-Preserving Data Analytics and Machine Learning

[Research Statement] [Projects] [Members]

Research Statement

Advances in big data and artificial intelligence have demonstrated the range of revolutionary changes and potential societal benefits enabled by data analysis. However, the increased collection and analysis of sensitive information about individuals raises new and concerning privacy issues. As demonstrated by the increasing prevalence of insider attacks and data breaches, current approaches for data security and privacy cannot guarantee privacy for individuals while providing general-purpose access for the analyst.

Our work is focused on approaches for achieving formal privacy guarantees in real-world deployments. To this end, we make both theoretical and systems contributions in both privacy-preserving data analytics and privacy-preserving machine learning. Our work is designed to address the challenges of practical use, and our work is already being adopted to provide differential privacy protections for analysts at industrial partners like Uber.


The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networks

Nicholas Carlini, Chang Liu, Ulfar Erlingsson, Jernej Kos, Dawn Song.

The 28th USENIX Security Symposium (Usenix 2019).


Towards Efficient Data Valuation Based on the Shapley Value

Ruoxi Jia*, David Dao*, Boxin Wang, Frances Ann Hubis, Nick Hynes, Nezihe Merve Gurel, Bo Li, Ce Zhang, Dawn Song, Costas Spanos.

The 22nd International Conference on Artificial Intelligence and Statistics (AISTATS 2019).


Efficient Task-Specific Data Valuation for Nearest Neighbor Algorithms

Ruoxi Jia, David Dao, Boxin Wang, Frances Ann Hubis, Nezihe Merve Gurel, Bo Li, Ce Zhang, Costas Spanos, Dawn Song.

The Forty-fifth International Conference on Very Large Data Bases (VLDB 2019).


Data Capsule: A New Paradigm for Automatic Compliance with Data Privacy Regulations

Lun Wang, Joseph P. Near, Neel Somani, Peng Gao, Andrew Low, David Dao, Dawn Song.

VLDB 2019 Workshop: Poly'19.


Duet: An Expressive Higher-order Language and Linear Type System for Statically Enforcing Differential Privacy

Joseph P. Near, David Darais, Chike Abuah, Tim Stevens, Pranav Gaddamadugu, Lun Wang, Neel Somani, Mu Zhang, Nikhil Sharma, Alex Shan, Dawn Song.

SPLASH 2019 OOPSLA (Distinguished Paper Award).


A Demonstration of Sterling: A Privacy-Preserving Data Marketplace

Nick Hynes, David Dao, David Yan, Raymond Cheng, Dawn Song.

VLDB demo 2018.

Efficient Deep Learning on Multi-Source Private Data

Nick Hynes, Raymond Cheng, Dawn Song.

Towards Practical Differentially Private Convex Optimization

Om Thakkar, Lun Wang, Joseph P. Near, Roger Iyengar, Dawn Song, Abhradeep Thakurta.

IEEE S&P 2019 (Oakland).

Code: Github


Chorus: Differential Privacy via Query Rewriting

Noah Johnson, Joseph Near, Joseph Hellerstein, Dawn Song.

Press: USENIX Enigma | Uber Security + Privacy.


Towards Practical Differential Privacy for SQL Queries

Noah Johnson, Joseph Near, Dawn Song.

International Conference on Very Large Databases (VLDB), 2018.

Code: Github

Press: Wired | Gizmodo | iapp | NYU GovLab | Norton | Uber Security + Privacy | SmartCityNews | etcentric | ET Auto | CNet | WSJ | idownloadblog