mlsanityData to 7 Oct 2026

Perspectives on Incorporating Expert Feedback into Model Updates

arXiv 2205.06905 ↗First posted 13 May 2022cs.LGcited by 7

Not on the map. Why?

Abstract (arXiv)

Machine learning (ML) practitioners are increasingly tasked with developing models that are aligned with non-technical experts' values and goals. However, there has been insufficient consideration on how practitioners should translate domain expertise into ML updates. In this paper, we consider how to capture interactions between practitioners and experts systematically. We devise a taxonomy to match expert feedback types with practitioner updates. A practitioner may receive feedback from an expert at the observation- or domain-level, and convert this feedback into updates to the dataset, loss function, or parameter space. We review existing work from ML and human-computer interaction to describe this feedback-update taxonomy, and highlight the insufficient consideration given to incorporating feedback from non-technical experts. We end with a set of open questions that naturally arise from our proposed taxonomy and subsequent survey.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 73 of its 147 references that are in the map. First posted on arXiv 13 May 2022.

Builds on

73 of 147 references.

References · 73

  1. 2017Deep reinforcement learning from human preferencescited by 3,846
  2. 2018The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networkscited by 2,250
  3. 2019CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparisoncited by 1,410
  4. 2017Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)cited by 1,309
  5. 2020Concept Bottleneck Modelscited by 1,022
  6. 2019Machine Unlearningcited by 1,020
  7. 2017Deep Bayesian Active Learning with Image Datacited by 812
  8. 2014Certifying and removing disparate impactcited by 715
  9. 2016Hyperband: A Novel Bandit-Based Approach to Hyperparameter Optimizationcited by 688
  10. 2019Pseudo-Labeling and Confirmation Bias in Deep Semi-Supervised Learningcited by 507
  11. 2019Learning the Difference that Makes a Difference with Counterfactually-Augmented Datacited by 428
  12. 2019Unlabeled Data Improves Adversarial Robustnesscited by 415
38 more
  1. 2017Snorkel: Rapid Training Data Creation with Weak Supervisioncited by 370
  2. 2018All Models are Wrong, but Many are Useful: Learning a Variable's Importance by Studying an Entire Class of Prediction Models Simultaneouslycited by 364
  3. 2017Right for the Right Reasons: Training Differentiable Models by Constraining their Explanationscited by 293
  4. 2020MCUNet: Tiny Deep Learning on IoT Devicescited by 269
  5. 2018Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shiftcited by 240
  6. 2019Human uncertainty makes classification more robustcited by 231
  7. 2018Empirical Risk Minimization under Fairness Constraintscited by 226
  8. 2019Explainable Machine Learning in Deploymentcited by 210
  9. 2018Online Learning: A Comprehensive Surveycited by 203
  10. 2020Formalizing Trust in Artificial Intelligence: Prerequisites, Causes and Goals of Human Trust in AIcited by 166
  11. 2018FairGAN: Fairness-aware Generative Adversarial Networkscited by 164
  12. 2019Predictive Multiplicity in Classificationcited by 162
  13. 2020Debugging Tests for Model Explanationscited by 146
  14. 2022Jury Learning: Integrating Dissenting Voices into Machine Learning Modelscited by 115
  15. 2019Human-Centered Tools for Coping with Imperfect Algorithms during Medical Decision-Makingcited by 109
  16. 2018A Kernel Theory of Modern Data Augmentationcited by 106
  17. 2021Human-in-the-Loop for Data Collection: a Multi-Target Counter Narrative Dataset to Fight Online Hate Speechcited by 81
  18. 2018Imbalanced Deep Learning by Minority Class Incremental Rectificationcited by 65
  19. 2021Characterizing Fairness Over the Set of Good Models Under Selective Labelscited by 65
  20. 2021Editing a classifier by rewriting its prediction rulescited by 64
  21. 2019Metric Learning for Individual Fairnesscited by 51
  22. 2016Learning Optimized Risk Scorescited by 47
  23. 2017Occam's razor is insufficient to infer the preferences of irrational agentscited by 44
  24. 2017Machine Teaching: A New Paradigm for Building Machine Learning Systemscited by 41
  25. 2017Iterative Machine Teachingcited by 40
  26. 2021Finding and Fixing Spurious Patterns with Explanationscited by 38
  27. 2018Human-in-the-Loop Interpretability Priorcited by 35
  28. 2019Regularizing Black-box Models for Improved Interpretabilitycited by 35
  29. 2018TED: Teaching AI to Explain its Decisionscited by 33
  30. 2021Discovering and Validating AI Errors With Crowdsourced Failure Reportscited by 28
  31. 2018Active Fairness in Algorithmic Decision Makingcited by 27
  32. 2021Soliciting Stakeholders' Fairness Notions in Child Maltreatment Predictive Systemscited by 27
  33. 2020Enforcing Interpretability and its Statistical Impacts: Trade-offs between Accuracy and Interpretabilitycited by 24
  34. 2019Fairness Warnings and Fair-MAML: Learning Fairly with Minimal Datacited by 23
  35. 2019A mathematical theory of cooperative communicationcited by 23
  36. 2019Recommendations and User Agency: The Reachability of Collaboratively-Filtered Informationcited by 21
  37. 2020Explainable Active Learning (XAL): An Empirical Study of How Local Explanations Impact Annotator Experiencecited by 20
  38. 2021Knowledge-Adaptation Priorscited by 20

50 most cited of 73 All 73 (JSON)

Built on by

Cited by year

Citing papers per year: 2022: 3; 2023: 2; 2024: 2.

7 papers.

  1. 2022Eliciting and Learning with Soft Labels from Every Annotatorcited by 43
  2. 2023Understanding Frontline Workers' and Unhoused Individuals' Perspectives on AI Used in Homeless Servicescited by 29
  3. 2022Human-in-the-Loop Mixupcited by 8
  4. 2024Unpacking Human-AI interactions: From interaction primitives to a design spacecited by 7
  5. 2022Identifying the Context Shift between Test Benchmarks and Production Datacited by 5
  6. 2023FeedbackLogs: Recording and Incorporating Stakeholder Feedback into Machine Learning Pipelinescited by 3
  7. 2024FI-CBL: A Probabilistic Method for Concept-Based Learning with Expert Rulescited by 1