mlsanityData to 7 Oct 2026

Understanding new tasks through the lens of training data via exponential tilting

arXiv 2205.13577 ↗First posted 26 May 2022cs.LG stat.ME stat.MLcited by 11

Not on the map. Why?

Abstract (arXiv)

Deploying machine learning models to new tasks is a major challenge despite the large size of the modern training datasets. However, it is conceivable that the training data can be reweighted to be more representative of the new (target) task. We consider the problem of reweighing the training samples to gain insights into the distribution of the target task. Specifically, we formulate a distribution shift model based on the exponential tilt assumption and learn train data importance weights minimizing the KL divergence between labeled train and unlabeled target datasets. The learned train data weights can then be used for downstream tasks such as target performance evaluation, fine-tuning, and model selection. We demonstrate the efficacy of our method on Waterbirds and Breeds benchmarks.

Key facts. Cited by 11 papers in the map (data through 7 Oct 2026). Builds on 46 of its 69 references that are in the map. First posted on arXiv 26 May 2022.

Builds on

46 of 69 references.

References · 46

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  3. 2015Domain-Adversarial Training of Neural Networkscited by 3,023
  4. 2020Unsupervised Learning of Visual Features by Contrasting Cluster Assignmentscited by 2,845
  5. 2019Invariant Risk Minimizationcited by 1,832
  6. 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,507
  7. 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
  8. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,232
  9. 2016Deep CORAL: Correlation Alignment for Deep Domain Adaptationcited by 1,194
  10. 2020In Search of Lost Domain Generalizationcited by 1,042
  11. 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 737
  12. 2022Fine-Tuning can Distort Pretrained Features and Underperform Out-of-Distributioncited by 722
34 more
  1. 2013Domain Generalization via Invariant Feature Representationcited by 622
  2. 2021Just Train Twice: Improving Group Robustness without Training Group Informationcited by 522
  3. 2019Domain Generalization via Model-Agnostic Learning of Semantic Featurescited by 405
  4. 2018Detecting and Correcting for Label Shift with Black Box Predictorscited by 364
  5. 2018Fairness Without Demographics in Repeated Loss Minimizationcited by 363
  6. 2020An Investigation of Why Overparameterization Exacerbates Spurious Correlationscited by 310
  7. 2020Environment Inference for Invariant Learningcited by 306
  8. 2018What is the Effect of Importance Weighting in Deep Learning?cited by 250
  9. 2018A DIRT-T Approach to Unsupervised Domain Adaptationcited by 215
  10. 2016Statistics of Robust Optimization: A Generalized Empirical Likelihood Approachcited by 182
  11. 2021Simple data balancing achieves competitive worst-group-accuracycited by 177
  12. 2020BREEDS: Benchmarks for Subpopulation Shiftcited by 155
  13. 2017No Classification without Representation: Assessing Geodiversity Issues in Open Data Sets for the Developing Worldcited by 152
  14. 2020Understanding Self-Training for Gradual Domain Adaptationcited by 149
  15. 2022Leveraging Unlabeled Data to Predict Out-of-Distribution Performancecited by 141
  16. 2020Domain Adaptation with Conditional Distribution Matching and Generalized Label Shiftcited by 124
  17. 2020A Unified View of Label Shift Estimationcited by 119
  18. 2019Regularized Learning for Domain Adaptation under Label Shiftscited by 118
  19. 2021Assessing Generalization of SGD via Disagreementcited by 116
  20. 2021Predicting with Confidence on Unseen Distributionscited by 116
  21. 2020Tilted Empirical Risk Minimizationcited by 103
  22. 2022Domain-Adjusted Regression or: ERM May Already Learn Features Sufficient for Out-of-Distribution Generalizationcited by 69
  23. 2019Transfer Learning for Nonparametric Classification: Minimax Rate and Adaptive Classifiercited by 64
  24. 2021Mandoline: Model Evaluation under Distribution Shiftcited by 55
  25. 2021DORO: Distributional and Outlier Robust Optimizationcited by 54
  26. 2021Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensemblescited by 54
  27. 2021Optimal Representations for Covariate Shiftcited by 51
  28. 2022Predicting Out-of-Distribution Error with the Projection Normcited by 50
  29. 2021A Theory of Label Propagation for Subpopulation Shiftcited by 48
  30. 2021On Tilted Losses in Machine Learning: Theory and Applicationscited by 42
  31. 2020Minimax optimal approaches to the label shift problem in non-parametric settingscited by 20
  32. 2021A linear adjustment based approach to posterior drift in transfer learningcited by 16
  33. 2019Maximum Likelihood with Bias-Corrected Calibration is Hard-To-Beat at Label Shift Adaptationcited by 3
  34. 2022Estimation of prediction error with known covariate shiftcited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 3; 2024: 4; 2025: 3; 2026: 1.

11 papers.

  1. 2023Large Language Model Routing with Benchmark Datasetscited by 166
  2. 2025SFT Doesn't Always Hurt General Capabilities: Revisiting Domain-Specific Fine-Tuning in LLMscited by 23
  3. 2024Learning In Reverse Causal Strategic Environments With Ramifications on Two Sided Marketscited by 6
  4. 2024Learning the Distribution Map in Reverse Causal Performative Predictioncited by 5
  5. 2023Invariance assumptions for class distribution estimationcited by 3
  6. 2024Positive and Unlabeled Data: Model, Estimation, Inference, and Classificationcited by 3
  7. 2023Addressing Distribution Shift in RTB Markets via Exponential Tiltingcited by 0
  8. 2024Exponential tilting of subweibull distributionscited by 0
  9. 2025A generalized approach to label shift: the Conditional Probability Shift Modelcited by 0
  10. 2025Unsupervised Domain Adaptation for Binary Classification with an Unobservable Source Subpopulationcited by 0
  11. 2026Conformal Prediction under Exponential-Tilt Joint Shiftcited by 0