Safety, Robustness & Interpretability

Adversarial robustness, AI safety, hallucination, interpretability/explainability.

A topic in the Science corpus, drawn from 2,034 papers.

Trusted voices in Safety, Robustness & Interpretability

Ordered by the trust Safety, Robustness & Interpretability's own network concentrates on each researcher, most first.

  1. Ian J. Goodfellow β€” trusted by 1,700 researchers here
  2. Rob Fergus β€” trusted by 1,274 researchers here
  3. Christian Szegedy β€” trusted by 1,098 researchers here
  4. Nicholas Carlini β€” trusted by 843 researchers here
  5. Matthew D. Zeiler β€” trusted by 1,070 researchers here
  6. Cynthia Dwork β€” trusted by 765 researchers here
  7. David A. Wagner β€” trusted by 760 researchers here
  8. Jonathon Shlens β€” trusted by 828 researchers here
  9. Nicolas Papernot β€” trusted by 649 researchers here
  10. Patrick Drew McDaniel β€” trusted by 602 researchers here
  11. Percy Liang β€” trusted by 666 researchers here
  12. Sameer Singh β€” trusted by 570 researchers here

Most-cited papers in Safety, Robustness & Interpretability

Ordered by the recorded citation count from the source corpus; a citation is not an endorsement.

  1. A Unified Approach to Interpreting Model Predictions (2017) β€” 37,902 citations
  2. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization (2016) β€” 28,183 citations
  3. Explaining and Harnessing Adversarial Examples (2014) β€” 22,661 citations
  4. Visualizing and Understanding Convolutional Networks (2013) β€” 17,221 citations
  5. Intriguing properties of neural networks (2013) β€” 16,794 citations
  6. Towards Deep Learning Models Resistant to Adversarial Attacks (2017) β€” 15,165 citations
  7. Anomaly detection: A survey (2009) β€” 12,600 citations
  8. Towards Evaluating the Robustness of Neural Networks (2016) β€” 10,045 citations
  9. Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead (2018) β€” 9,528 citations
  10. k-Anonymity: A Model for Protecting Privacy (2002) β€” 9,049 citations
  11. Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AI (2019, Information Fusion) β€” 9,028 citations
  12. On Calibration of Modern Neural Networks (2017) β€” 8,827 citations
  13. Deep Learning with Differential Privacy (2016) β€” 8,077 citations
  14. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜 (2021) β€” 7,969 citations
  15. Differential Privacy (2006) β€” 7,564 citations
  16. On the Opportunities and Risks of Foundation Models (2021) β€” 6,923 citations
  17. Adversarial examples in the physical world (2016) β€” 6,799 citations
  18. What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision? (2017) β€” 6,295 citations
  19. A Survey on Bias and Fairness in Machine Learning (2019, ACM Computing Surveys) β€” 6,091 citations
  20. Dissecting racial bias in an algorithm used to manage the health of populations (2019) β€” 5,573 citations
  21. DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks (2015) β€” 5,469 citations
  22. Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification (2018) β€” 5,467 citations
  23. Explanation in Artificial Intelligence: Insights from the Social Sciences (2017) β€” 5,450 citations
  24. Membership Inference Attacks Against Machine Learning Models (2016) β€” 5,446 citations
  25. Towards A Rigorous Science of Interpretable Machine Learning (2017) β€” 5,317 citations
  26. Equality of Opportunity in Supervised Learning (2016) β€” 5,258 citations
  27. Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI) (2018) β€” 5,195 citations
  28. Survey of Hallucination in Natural Language Generation (2022) β€” 5,175 citations
  29. A Survey of Methods for Explaining Black Box Models (2018) β€” 5,098 citations
  30. Striving for Simplicity: The All Convolutional Net (2014) β€” 5,049 citations
  31. On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation (2015) β€” 4,977 citations
  32. Learning Important Features Through Propagating Activation Differences (2017) β€” 4,716 citations
  33. Fairness through awareness (2011) β€” 4,555 citations
  34. The mythos of model interpretability (2016) β€” 4,511 citations
  35. A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks (2016) β€” 4,454 citations
  36. The Limitations of Deep Learning in Adversarial Settings (2015) β€” 4,317 citations
  37. Sorting Things Out: Classification and Its Consequences (2001) β€” 4,185 citations
  38. Differential Privacy: A Survey of Results (2008) β€” 4,153 citations
  39. Practical Black-Box Attacks against Machine Learning (2016) β€” 4,113 citations
  40. Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings (2016) β€” 3,784 citations
  41. Understanding Black-box Predictions via Influence Functions (2017) β€” 3,706 citations
  42. Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples (2018) β€” 3,560 citations
  43. Deep neural networks are easily fooled: High confidence predictions for unrecognizable images (2014) β€” 3,543 citations
  44. Adversarial Machine Learning at Scale (2016) β€” 3,411 citations
  45. Boosting Adversarial Attacks with Momentum (2017) β€” 3,379 citations
  46. Constitutional AI: Harmlessness from AI Feedback (2022, arXiv.org) β€” 3,377 citations
  47. Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures (2015) β€” 3,376 citations
  48. Fake News Detection on Social Media: A Data Mining Perspective (2017) β€” 3,316 citations
  49. Distillation as a Defense to Adversarial Perturbations Against Deep Neural Networks (2015, IEEE Symposium on Security and Privacy) β€” 3,315 citations
  50. Concrete Problems in AI Safety (2016) β€” 3,295 citations

Most active researchers

By papers in this topic: a volume measure, not trust.