Lune

Profiles2020-2025

Tom Goldstein

University of Maryland, College Park

30Papers
22Last author
51Unique co-authors
0First author
0Solo

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

Recent papers30

  1. Efficient Fine-Tuning and Concept Suppression for Pruned Diffusion ModelsAuthor 5 of 6CVPR 2025
  2. Understanding and Mitigating Copying in Diffusion ModelsAuthor 5 of 5NeurIPS 2023 · 265 citations
  3. GOAT: A Global Transformer on Large-scale GraphsAuthor 6 of 6ICML 2023 · 76 citations
  4. Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion ModelsAuthor 5 of 5CVPR 2023
  5. Provable Robustness against Wasserstein Distribution Shifts via Input RandomizationAuthor 3 of 4ICLR 2023
  6. Towards Transferable Adversarial Attacks on Vision TransformersAuthor 5 of 6AAAI 2022 · 156 citations
  7. Robust Optimization as Data Augmentation for Large-scale GraphsAuthor 8 of 8CVPR 2022 · 87 citations
  8. The Close Relationship Between Contrastive Learning and Meta-LearningAuthor 5 of 5ICLR 2022 · 16 citations
  9. Where do Models go Wrong? Parameter-Space Saliency Maps for ExplainabilityAuthor 6 of 6NeurIPS 2022 · 14 citations
  10. Can Neural Nets Learn the Same Model Twice? Investigating Reproducibility and Double Descent from the Decision Boundary PerspectiveAuthor 8 of 8CVPR 2022
  11. GradInit: Learning to Initialize Neural Networks for Stable and Efficient TrainingAuthor 6 of 6NeurIPS 2021 · 73 citations
  12. VQ-GNN: A Universal Framework to Scale up Graph Neural Networks using Vector QuantizationAuthor 7 of 7NeurIPS 2021 · 68 citations
  13. LowKey: Leveraging Adversarial Attacks to Protect Social Media Users from Facial RecognitionAuthor 7 of 7ICLR 2021 · 52 citations
  14. Gradient-Free Adversarial Training Against Image Corruption for Learning-based SteeringAuthor 5 of 6NeurIPS 2021 · 37 citations
  15. Center Smoothing: Certified Robustness for Networks with Structured OutputsAuthor 2 of 2NeurIPS 2021 · 23 citations
  16. Encoding Robustness to Image Style via Adversarial Feature PerturbationsAuthor 4 of 4NeurIPS 2021 · 23 citations
  17. Are Adversarial Examples Created Equal? A Learnable Weighted Minimax Risk for Robustness under Non-uniform AttacksAuthor 3 of 4AAAI 2021 · 21 citations
  18. WrapNet: Neural Net Inference with Ultra-Low-Precision ArithmeticAuthor 6 of 6ICLR 2021 · 16 citations
  19. FreeLB: Enhanced Adversarial Training for Natural Language UnderstandingAuthor 5 of 6ICLR 2020 · 502 citations
  20. MetaPoison: Practical General-purpose Clean-label Data PoisoningAuthor 5 of 5NeurIPS 2020 · 242 citations
  21. Universal Adversarial TrainingAuthor 6 of 6AAAI 2020 · 210 citations
  22. Certified Defenses for Adversarial PatchesAuthor 6 of 6ICLR 2020 · 194 citations
  23. Adversarially robust transfer learningAuthor 7 of 7ICLR 2020 · 130 citations
  24. Adversarially Robust Few-Shot Learning: A Meta-Learning ApproachAuthor 3 of 3NeurIPS 2020 · 107 citations
  25. Curse of Dimensionality on Randomized Smoothing for Certifiable RobustnessAuthor 3 of 4ICML 2020 · 102 citations
  26. Breaking Certified Defenses: Semantic Adversarial Examples with Spoofed robustness CertificatesAuthor 3 of 3ICLR 2020 · 57 citations
  27. Certifying Confidence via Randomized SmoothingAuthor 4 of 4NeurIPS 2020 · 44 citations
  28. Adversarial Attacks on Copyright Detection SystemsAuthor 3 of 3ICML 2020 · 38 citations
  29. Certifying Strategyproof Auction NetworksAuthor 3 of 4NeurIPS 2020 · 37 citations
  30. Detection as Regression: Certified Object Detection with Median SmoothingAuthor 6 of 6NeurIPS 2020 · 15 citations
Dusk over the sea between two cliffs drawn in fine vertical lines