Sotabase

Career

· Research Scientist, OpenAI2023–
· Cadence Design Systems Professor of Computing, Massachusetts Institute of Technology2021–
· Professor, Massachusetts Institute of Technology2020–
· Associate Professor of Computer Science with Tenure, Massachusetts Institute of Technology2018–2020
· Associate Professor of Computer Science without Tenure, Massachusetts Institute of Technology2017–2018
· Assistant Professor, Massachusetts Institute of Technology2015–2017
· Director, Massachusetts Institute of Technology2015–
· Assistant Professor, EPFL2012–2015
· Postdoctoral Researcher, Microsoft Research New England2011–2012
· MS in Computer Science, Massachusetts Institute of Technology2007–2009
· PhD in Computer Science, Massachusetts Institute of Technology2007–2011
· Licencjat Degree (B.Sc. equivalent) in Theoretical Physics, University of Wroclaw2007–
· Magister Degree (B.Sc.+M.Sc. equivalent) in Computer Science, University of Wroclaw2006–

Publications (219)

Towards Deep Learning Models Resistant to Adversarial Attacks
International Conference on Learning Representations · 2017
13,928
cited
Adversarial Examples Are Not Bugs, They Are Features
Neural Information Processing Systems · 2019
2,017
cited
Robustness May Be at Odds with Accuracy
International Conference on Learning Representations · 2018
1,888
cited
How Does Batch Normalization Help Optimization? (No, It Is Not About Internal Covariate Shift)
Neural Information Processing Systems · 2018
1,674
cited
On Evaluating Adversarial Robustness
arXiv.org · 2019
963
cited
On Adaptive Attacks to Adversarial Example Defenses
Neural Information Processing Systems · 2020
916
cited
Spectral Signatures in Backdoor Attacks
Neural Information Processing Systems · 2018
896
cited
Adversarially Robust Generalization Requires More Data
Neural Information Processing Systems · 2018
843
cited
Exploring the Landscape of Spatial Robustness
International Conference on Machine Learning · 2017
548
cited
Label-Consistent Backdoor Attacks
arXiv.org · 2019
472
cited
Do Adversarially Robust ImageNet Models Transfer Better?
Neural Information Processing Systems · 2020
468
cited
Noise or Signal: The Role of Image Backgrounds in Object Recognition
International Conference on Learning Representations · 2020
438
cited
Prior Convictions: Black-Box Adversarial Attacks with Bandits and Priors
International Conference on Learning Representations · 2018
411
cited
A Rotation and a Translation Suffice: Fooling CNNs with Simple Transformations
arXiv.org · 2017
381
cited
Electrical flows, laplacian systems, and faster approximation of maximum flow in undirected graphs
Symposium on the Theory of Computing · 2010
370
cited
Dataset Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defenses
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2020
360
cited
Implementation Matters in Deep Policy Gradients: A Case Study on PPO and TRPO
arXiv.org · 2020
294
cited
Implementation Matters in Deep RL: A Case Study on PPO and TRPO
International Conference on Learning Representations · 2020
286
cited
Clean-Label Backdoor Attacks
2018
261
cited
Navigating Central Path with Electrical Flows: From Flows to Matchings, and Back
IEEE Annual Symposium on Foundations of Computer Science · 2013
259
cited
Sotabase
Aleksander Madry | Researcher Profile | Sotabase | Sotabase