Bridging Distribution Shift and AI Safety: Conceptual and Methodological Synergies
Not on the map. Why?
Abstract (arXiv)
This paper bridges distribution shift and AI safety through a comprehensive analysis of their conceptual and methodological synergies. While prior discussions often focus on narrow cases or informal analogies, we establish two types connections between specific causes of distribution shift and fine-grained AI safety issues: (1) methods addressing a specific shift type can help achieve corresponding safety goals, or (2) certain shifts and safety issues can be formally reduced to each other, enabling mutual adaptation of their methods. Our findings provide a unified perspective that encourages deeper integration between distribution shift and AI safety research.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 157 of its 233 references that are in the map. First posted on arXiv 28 May 2025.
Builds on
157 of 233 references.
References · 157
- 2016Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localizationcited by 6,733
- 2015Domain-Adversarial Training of Neural Networkscited by 3,044
- 2016Equality of Opportunity in Supervised Learningcited by 2,007
- 2019A Survey on Bias and Fairness in Machine Learningcited by 1,917
- 2019Invariant Risk Minimizationcited by 1,854
- 2016Concrete Problems in AI Safetycited by 1,642
- 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,515
- 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,432
- 2017Towards A Rigorous Science of Interpretable Machine Learningcited by 1,416
- 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,375
- 2015Learning Transferable Features with Deep Adaptation Networkscited by 1,329
- 2017BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chaincited by 1,106
38 more
- 2017Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoningcited by 1,055
- 2021A General Language Assistant as a Laboratory for Alignmentcited by 1,055
- 2019Aleatoric and Epistemic Uncertainty in Machine Learning: An Introduction to Concepts and Methodscited by 913
- 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 742
- 2016Fair prediction with disparate impact: A study of bias in recidivism prediction instrumentscited by 672
- 2013Domain Generalization via Invariant Feature Representationcited by 623
- 2018Fine-Pruning: Defending Against Backdooring Attacks on Deep Neural Networkscited by 580
- 2015Causal inference using invariant prediction: identification and confidence intervalscited by 536
- 2024Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Trainingcited by 529
- 2021Just Train Twice: Improving Group Robustness without Training Group Informationcited by 523
- 2018Spectral Signatures in Backdoor Attackscited by 484
- 2019STRIP: A Defence Against Trojan Attacks on Deep Neural Networkscited by 459
- 2018Detecting Backdoor Attacks on Deep Neural Networks by Activation Clusteringcited by 456
- 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 422
- 2017Wild Patterns: Ten Years After the Rise of Adversarial Machine Learningcited by 414
- 2019Contrastive Adaptation Network for Unsupervised Domain Adaptationcited by 407
- 2019A Survey of Deep Learning Techniques for Autonomous Drivingcited by 403
- 2022Last Layer Re-Training is Sufficient for Robustness to Spurious Correlationscited by 381
- 2019Hidden Trigger Backdoor Attackscited by 371
- 2018Detecting and Correcting for Label Shift with Black Box Predictorscited by 366
- 2019Selection via Proxy: Efficient Data Selection for Deep Learningcited by 339
- 2020Keystroke Dynamics as Part of Lifeloggingcited by 334
- 2020Performative Predictioncited by 327
- 2021Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networkscited by 324
- 2017Unified Deep Supervised Domain Adaptation and Generalizationcited by 323
- 2020Environment Inference for Invariant Learningcited by 306
- 2019Label-Consistent Backdoor Attackscited by 296
- 2021Distribution-Free, Risk-Controlling Prediction Setscited by 288
- 2021Anti-Backdoor Learning: Training Clean Models on Poisoned Datacited by 263
- 2019Risks from Learned Optimization in Advanced Machine Learning Systemscited by 258
- 2019Adversarial Domain Adaptation with Domain Mixupcited by 250
- 2023Large Language Model Alignment: A Surveycited by 243
- 2020Domain Generalization using Causal Matchingcited by 225
- 2021Invariance Principle Meets Information Bottleneck for Out-of-Distribution Generalizationcited by 220
- 2023On the Robustness of ChatGPT: An Adversarial and Out-of-distribution Perspectivecited by 211
- 2021Trustworthy AI: From Principles to Practicescited by 207
- 2022Improving Out-of-Distribution Robustness via Selective Augmentationcited by 202
- 2017Central Moment Discrepancy (CMD) for Domain-Invariant Representation Learningcited by 192
50 most cited of 157 All 157 (JSON)
Built on by
Citing papers per year: 2025: 2.
2 papers.