Seal     

Xi Wu

I joined Google in 2016. I received my Ph.D. in Computer Science from University of Wisconsin-Madison, advised by Jeffrey F. Naughton and Somesh Jha. I was awarded a Google PhD Fellowship in 2016 (Privacy and Security).

Bilevel Relational Algebra (BiRA). Modern data is big and complex. At Google, I lead a team to build novel analytic infrastructure to make the search for data insights easy and efficient. A main theme of my work is Bilevel Relational Algebra (BiRA), which augments the classic relational algebra with BilevelRelation and bilevel algebraic operators. BiRA is superior in declarative power for a wide range of insight-analytic questions that have a "search+eval" flavor, and is more amenable to systematic query optimizations. Please refer to our manuscript for details.

Publications

Stratified Adversarial Robustness with Rejection
Jiefeng Chen, Jayaram Raghuram, Jihye Choi, Xi Wu, Yingyu Liang, Somesh Jha
ICML 2023, arXiv 2023, AAAI Workshop on Adversarial Machine Learning and Beyond 2022 (Oral Presentation and Best Paper Award)
(code)
The Trade-off between Universality and Label Efficiency of Representations from Contrastive Learning
Zhenmei Shi, Jiefeng Chen, Kunyang Li, Jayaram Raghuram, Xi Wu, Yingyu Liang, Somesh Jha
ICLR 2023 (Spotlight)
(code)
Towards Evaluating the Robustness of Neural Networks Learned by Transduction
Jiefeng Chen, Xi Wu, Yang Guo, Yingyu Liang, Somesh Jha
ICLR 2022, arXiv 2021
(code)
Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles
Jiefeng Chen, Frederick Liu, Besim Avci, Xi Wu, Yingyu Liang, Somesh Jha
NeurIPS 2021, arXiv 2021
(code)
ATOM: Robustifying Out-of-distribution Detection Using Outlier Mining
Jiefeng Chen, Yixuan Li, Xi Wu, Yingyu Liang and Somesh Jha
ECML 2021, arXiv 2020, ICML UDL 2020
(code)
DIFF: A Relational Interface to Large-Scale Data Explanation
Firas Abuzaid, Peter Kraft, Sahhana Suri, Edward Gan, Eric Xu, Atul Shenoy, Avsin Anathanarayan, John Sheu, Erik Meijer, Xi Wu, Jeffrey F. Naughton, Peter Bailis, Matei Zaharia
The VLDB Journal (2021), VLDB 2019 (Invited to "Best of VLDB 2019" Special Issue)
Concise Explanations for Neural Networks using Adversarial Training
Prasad Chalasani, Jiefeng Chen, Amrita Roy Chowdhury, Somesh Jha, Xi Wu
ICML 2020, arXiv 2018
(code)
Robust Attribution Regularization
Jiefeng Chen, Xi Wu, Vaibhav Rastogi, Yingyu Liang, Somesh Jha
NeurIPS 2019, arXiv 2019
(code, slides, poster, Alta Cognita)
Towards Understanding Limitations of Pixel Discretization Against Adversarial Attacks
Jiefeng Chen, Xi Wu, Vaibhav Rastogi, Yingyu Liang, Somesh Jha
EuroS&P 2019, arXiv 2018
(code)
Tuple-oriented Compression for Large-scale Mini-batch Stochastic Gradient Descent
Fengan Li, Lingjiao Chen, Yijing Zeng, Arun Kumar, Xi Wu, Jeffrey F. Naughton, Jignesh M. Patel
SIGMOD 2019, arXiv 2017
Reinforcing Adversarial Robustness using Model Confidence Induced by Adversarial Training
Xi Wu, Uyeong Jang, Jiefeng Chen, Lingjiao Chen, Somesh Jha
ICML 2018, arXiv 2017
(slides)
Bolt-on Differential Privacy for Scalable Stochastic Gradient Descent-based Analytics
Xi Wu, Fengan Li, Arun Kumar, Kamalika Chaudhuri, Somesh Jha, Jeffrey F. Naughton
SIGMOD 2017, arXiv 2016
(slides)
Objective Metrics and Gradient Descent Algorithms for Adversarial Examples in Machine Learning
Uyeong Jang, Xi Wu, Somesh Jha
ACSAC 2017
A Study of Stability in Data Privacy
Advisors: Jeffrey F. Naughton, Somesh Jha.
Ph.D. Thesis, UW-Madison, August 2016
ProQuest
A Methodology for Modeling Model-Inversion Attacks
Xi Wu, Matthew Fredrikson, Somesh Jha, Jeffrey F. Naughton
CSF 2016
(slides)
Distillation as a Defense to Adversarial Perturbations against Deep Neural Networks
Nicolas Papernot, Patrick McDaniel, Xi Wu, Somesh Jha, Ananthram Swami
S&P (Oakland) 2016, arXiv 2015
A Completeness Theory for Polynomial (Turing) Kernelization
with Danny Hermelin, Stephan Kratsch, Karolina Soltys, Magnus Wahlstrom.
Algorithmica 2015, IPEC 2013
Uncertainty Aware Query Execution Time Prediction
Wentao Wu, Xi Wu, Hakan Hacigümüs, Jeffrey F. Naughton.
VLDB 2014
Weak Compositions and Their Applications to Polynomial Lower Bounds for Kernelization
with Danny Hermelin
SODA 2012, ECCC 2011
(slides)
COREMU: A Scalable and Portable Parallel Full-system Emulator
Zhaoguo Wang, Ran Liu, Yufei Chen, Xi Wu, Haibo Chen, Binyu Zang
PPoPP 2011
Extended Islands of Tractability for Parsimony Haplotyping
with Rudolf Fleischer, Jiong Guo, Rolf Niedermeier, Johannes Uhlmann, Yihui Wang, Mathias Weller
CPM 2010
Experimental Study of FPT Algorithms for the Directed Feedback Vertex Set Problem
with Rudolf Fleischer, Liwei Yuan
ESA 2009
(slides)
Control Flow Obfuscation with Information Flow Tracking
Haibo Chen, Liwei Yuan, Xi Wu, Bo Huang, Pen-chung Yew, Binyu Zang
MICRO 2009
From Speculation to Security: Practical and Efficient Information Flow Tracking using Speculative Hardware
Haibo Chen, Xi Wu, Liwei Yuan, Binyu Zang, Pen-chung Yew, Frederic T. Chong
ISCA 2008

Manuscripts

Bilevel Relations and Their Applications to Data Insights
Xi Wu, Xiangyao Yu, Shaleen Deep, Ahmed Mahmood, Uyeong Jang, Stratis Viglas, Somesh Jha, John Cieslewicz, Jeffrey F. Naughton
arXiv 2023
Two Heads are Better than One: Towards Better Adversarial Robustness by Combining Transduction and Rejection
Nils Palumbo, Yang Guo, Xi Wu, Jiefeng Chen, Yingyu Liang, Somesh Jha
arXiv 2023
Towards Adversarial Robustness via Transductive Learning
Jiefeng Chen, Yang Guo, Xi Wu, Tianqi Li, Qicheng Lao, Yingyu Liang, Somesh Jha
arXiv 2021
Robust Out-of-distribution Detection for Neural Networks
Jiefeng Chen, Yixuan Li, Xi Wu, Yingyu Liang, Somesh Jha
arXiv 2020
Representation Bayesian Risk Decompositions and Multi-Source Domain Adaptation
Xi Wu, Yang Guo, Jiefeng Chen, Yingyu Liang, Somesh Jha, Prasad Chalasani
arXiv 2020
Rearchitecting Classification Frameworks For Increased Robustness
Varun Chandrasekaran, Brian Tang, Nicolas Papernot, Kassem Fawaz, Somesh Jha, Xi Wu
arXiv 2019
Revisiting Differentially Private Regression: Lessons From Learning Theory and their Consequences
Xi Wu, Matthew Fredrikson, Wentao Wu, Somesh Jha, Jeffrey F. Naughton
arXiv 2015