Lune

Profiles2019-2026

Soheil Feizi

Department of Computer Science, University of Maryland

28Papers
25Last author
41Unique co-authors
0First author
0Solo

First and last author come from byline position, which some venues assign alphabetically.

Recent papers28

  1. Revisiting the Past: Data Unlearning with Model State HistoryAuthor 4 of 4ICLR 2026 · 2 citations
  2. Adversarial Paraphrasing: A Universal Attack for Humanizing AI-Generated TextAuthor 5 of 5NeurIPS 2025 · 28 citations
  3. DyePack: Provably Flagging Test Set Contamination in LLMs Using BackdoorsAuthor 4 of 4EMNLP 2025
  4. Tool Preferences in Agentic LLMs are UnreliableAuthor 8 of 8EMNLP 2025
  5. Decomposing and Interpreting Image Representations via Text in ViTs Beyond CLIPAuthor 3 of 3NeurIPS 2024 · 26 citations
  6. CUDA: Convolution-Based Unlearnable DatasetsAuthor 3 of 3CVPR 2023
  7. Provable Robustness against Wasserstein Distribution Shifts via Input RandomizationAuthor 4 of 4ICLR 2023
  8. Segment and Complete: Defending Object Detectors against Adversarial Patch Attacks with Robust Patch DetectionAuthor 5 of 5CVPR 2022 · 99 citations
  9. Policy Smoothing for Provably Robust Reinforcement LearningAuthor 3 of 3ICLR 2022 · 62 citations
  10. A Comprehensive Study of Image Classification Model Sensitivity to Foregrounds, Backgrounds, and Visual AttributesAuthor 4 of 4CVPR 2022 · 42 citations
  11. Toward Efficient Robust Training against Union of ℓp\ell_p Threat ModelsAuthor 3 of 3NeurIPS 2022 · 2 citations
  12. Perceptual Adversarial Robustness: Defense Against Unseen Threat ModelsAuthor 3 of 3ICLR 2021 · 217 citations
  13. Improving Deep Learning Interpretability by Saliency Guided TrainingAuthor 2 of 3NeurIPS 2021 · 121 citations
  14. Low Curvature Activations Reduce Overfitting in Adversarial TrainingAuthor 3 of 4ICCV 2021 · 49 citations
  15. Winning Lottery Tickets in Deep Generative ModelsAuthor 3 of 3AAAI 2021 · 44 citations
  16. Deep Partition Aggregation: Provable Defenses against General Poisoning AttacksAuthor 2 of 2ICLR 2021 · 22 citations
  17. Sample Efficient Detection and Classification of Adversarial Attacks via Self-Supervised EmbeddingsAuthor 2 of 2ICCV 2021 · 20 citations
  18. Influence Functions in Deep Learning Are FragileAuthor 3 of 3ICLR 2021 · 15 citations
  19. Benchmarking Deep Learning Interpretability in Time Series PredictionsAuthor 4 of 4NeurIPS 2020 · 249 citations
  20. (De)Randomized Smoothing for Certifiable Defense against Patch AttacksAuthor 2 of 2NeurIPS 2020 · 188 citations
  21. Robust Optimal Transport with Applications in Generative Modeling and Domain AdaptationAuthor 3 of 3NeurIPS 2020 · 141 citations
  22. Robustness Certificates for Sparse Adversarial Attacks by Randomized AblationAuthor 2 of 2AAAI 2020 · 114 citations
  23. Curse of Dimensionality on Randomized Smoothing for Certifiable RobustnessAuthor 4 of 4ICML 2020 · 102 citations
  24. Dual Manifold Adversarial Robustness: Defense against Lp and non-Lp Adversarial AttacksAuthor 5 of 5NeurIPS 2020 · 70 citations
  25. Second-Order Provable Defenses against Adversarial AttacksAuthor 2 of 2ICML 2020 · 64 citations
  26. Certifying Confidence via Randomized SmoothingAuthor 3 of 4NeurIPS 2020 · 44 citations
  27. Maximum Likelihood Embedding of Logistic Random Dot Product GraphsAuthor 3 of 3AAAI 2020 · 8 citations
  28. Normalized Wasserstein for Mixture Distributions With Applications in Adversarial Learning and Domain AdaptationAuthor 3 of 3ICCV 2019 · 53 citations
Dusk over the sea between two cliffs drawn in fine vertical lines