mlsanityData to 14 Oct 2026

Adversarial Robustness in One-Stage Learning-to-Defer

arXiv 2510.10988 ↗First posted 13 Oct 2025stat.ML cs.LGcited by 5

Not on the map. Why?

Abstract (arXiv)

Learning-to-Defer (L2D) enables hybrid decision-making by routing inputs either to a predictor or to external experts. While promising, L2D is highly vulnerable to adversarial perturbations, which can not only flip predictions but also manipulate deferral decisions. Prior robustness analyses focus solely on two-stage settings, leaving open the end-to-end (one-stage) case where predictor and allocation are trained jointly. We introduce the first framework for adversarial robustness in one-stage L2D, covering both classification and regression. Our approach formalizes attacks, proposes cost-sensitive adversarial surrogate losses, and establishes theoretical guarantees including H, (R , F), and Bayes consistency. Experiments on benchmark datasets confirm that our methods improve robustness against untargeted and targeted attacks while preserving clean performance.

Key facts. Cited by 5 papers in the map (data through 14 Oct 2026). Builds on 47 of its 96 references that are in the map. First posted on arXiv 13 Oct 2025.

Builds on

47 of 96 references.

References · 47

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  3. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  4. 2018Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labelscited by 997
  5. 2023FrugalGPT: How to Use Large Language Models While Reducing Cost and Improving Performancecited by 853
  6. 2021MedMNIST v2 -- A large-scale lightweight benchmark for 2D and 3D biomedical image classificationcited by 690
  7. 2019Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC)cited by 669
  8. 2017Adversarial Examples for Evaluating Reading Comprehension Systemscited by 571
  9. 2017Robust Loss Functions under Label Noise for Deep Neural Networkscited by 443
  10. 2020Uncovering the Limits of Adversarial Training against Norm-Bounded Adversarial Examplescited by 278
  11. 2020Consistent Estimators for Learning to Defer to an Expertcited by 271
  12. 2017Predict Responsibly: Improving Fairness and Accuracy by Learning to Defercited by 233
35 more
  1. 2023Cross-Entropy Loss Functions: Theoretical Analysis and Applicationscited by 210
  2. 2020MedMNIST Classification Decathlon: A Lightweight AutoML Benchmark for Medical Image Analysiscited by 206
  3. 2025Universal Model Routing for Efficient LLM Inferencecited by 86
  4. 2021Towards Unbiased and Accurate Deferral to Multiple Expertscited by 72
  5. 2023Who Should Predict? Exact Algorithms For Learning to Defer to Humanscited by 70
  6. 2022Learning to Defer to Multiple Experts: Consistent Surrogate Losses, Confidence Calibration, and Conformal Ensemblescited by 60
  7. 2021Calibration and Consistency of Adversarial Surrogate Lossescited by 47
  8. 2022Sample Efficient Learning of Predictors that Complement Humanscited by 46
  9. 2021Combining Human Predictions with Model Probabilities via Confusion Matrices and Calibrationcited by 40
  10. 2022Forming Effective Human-AI Teams: Building Machine Learning Models that Complement the Capabilities of Multiple Expertscited by 40
  11. 2023Principled Approaches for Learning to Defer with Multiple Expertscited by 38
  12. 2023Predictor-Rejector Multi-Class Abstention: Theoretical Analysis and Algorithmscited by 36
  13. 2023Learning to Reject with a Fixed Predictor: Application to Decontextualizationcited by 32
  14. 2023Theoretically Grounded Loss Functions and Algorithms for Score-Based Multi-Class Abstentioncited by 32
  15. 2024Regression with Multi-Expert Deferralcited by 29
  16. 2023Ranking with Abstentioncited by 27
  17. 2024Learning to Defer to a Population: A Meta-Learning Approachcited by 27
  18. 2023In Defense of Softmax Parametrization for Calibrated and Consistent Learning to Defercited by 26
  19. 2024Realizable H-Consistent and Bayes-Consistent Loss Functions for Learning to Defercited by 26
  20. 2024A Universal Growth Rate for Learning with Smooth Surrogate Lossescited by 24
  21. 2024Cardinality-Aware Set Prediction and Top-k Classificationcited by 23
  22. 2024Multi-Label Learning with Stronger Consistency Guaranteescited by 22
  23. 2024H-Consistency Guarantees for Regressioncited by 21
  24. 2024Enhanced H-Consistency Boundscited by 18
  25. 2024A Two-Stage Learning-to-Defer Approach for Multi-Task Learningcited by 17
  26. 2024Trustworthy and Practical AI for Healthcare: A Guided Deferral System with Large Language Modelscited by 16
  27. 2025Adversarial Robustness in Two-Stage Learning-to-Defer: Algorithms and Guaranteescited by 16
  28. 2025Mastering Multiple-Expert Routing: Realizable H-Consistency and Strong Guarantees for Learning to Defercited by 16
  29. 2024Optimal Query Allocation in Extractive QA with LLMs: A Learning-to-Defer Framework with Theoretical Guaranteescited by 15
  30. 2025Balancing the Scales: A Theoretical and Algorithmic Framework for Learning from Imbalanced Datacited by 15
  31. 2022Counterfactual Inference of Second Opinionscited by 14
  32. 2024A Causal Framework for Evaluating Deferring Systemscited by 13
  33. 2021Learning-to-defer for sequential medical decision-making under uncertaintycited by 12
  34. 2025Why Ask One When You Can Ask k? Learning-to-Defer to the Top-k Expertscited by 7
  35. 2025One-Stage Top-k Learning-to-Defer: Score-Based Surrogates with Theoretical Guaranteescited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026Learning-to-Defer in Non-Stationary Time Series via Switching State-Space Modelscited by 6
  2. 2026Consistent Learning-to-Defer with Expert-Conditional Advicecited by 3
  3. 2026Optimized Deferral for Imbalanced Settingscited by 2
  4. 2026Linear-Core Surrogates: Smooth Loss Functions with Linear Rates for Classification and Structured Predictioncited by 2
  5. 2026Online Learning-to-Defer with Varying Expertscited by 1