mlsanityData to 14 Oct 2026

ASAP: Unsupervised Post-training with Label Distribution Shift Adaptive Learning Rate

arXiv 2508.13445 ↗First posted 19 Aug 2025cs.LGcited by 3

Not on the map. Why?

Abstract (arXiv)

In real-world applications, machine learning models face online label shift, where label distributions change over time. Effective adaptation requires careful learning rate selection: too low slows adaptation and too high causes instability. We propose ASAP (Adaptive Shift Aware Post-training), which dynamically adjusts the learning rate by computing the cosine distance between current and previous unlabeled outputs and mapping it within a bounded range. ASAP requires no labels, model ensembles, or past inputs, using only the previous softmax output for fast, lightweight adaptation. Experiments across multiple datasets and shift scenarios show ASAP consistently improves accuracy and efficiency, making it practical for unsupervised model adaptation.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 25 of its 43 references that are in the map. First posted on arXiv 19 Aug 2025.

Builds on

25 of 43 references.

References · 25

  1. 2017Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithmscited by 4,349
  2. 2018Detecting and Correcting for Label Shift with Black Box Predictorscited by 366
  3. 2021Online Adaptation to Label Distribution Shiftcited by 44
  4. 2023Test-Time Domain Adaptation by Learning Domain-Aware Batch Normalizationcited by 31
  5. 2022DLTTA: Dynamic Learning Rate for Test-time Adaptation on Cross-domain Medical Imagescited by 28
  6. 2022Adapting to Online Label Shift with Provable Guaranteescited by 26
  7. 2024Enhancing Zero-Shot Vision Models by Label-Free Prompt Distribution Learning and Bias Correctingcited by 26
  8. 2024Protected Test-Time Adaptation via Online Entropy Matching: A Betting Approachcited by 23
  9. 2023Label Shift Adapter for Test-Time Adaptation under Covariate and Label Shiftscited by 21
  10. 2023Fairness through Aleatoric Uncertaintycited by 16
  11. 2024Tolerant Algorithms for Learning with Arbitrary Covariate Shiftcited by 15
  12. 2023Causality and Independence Enhancement for Biased Node Classificationcited by 13
13 more
  1. 2023Online Label Shift: Optimal Dynamic Regret meets Practical Algorithmscited by 11
  2. 2024FuseFL: One-Shot Federated Learning through the Lens of Causality with Progressive Model Fusioncited by 11
  3. 2024Bridging Multicalibration and Out-of-distribution Generalization Beyond Covariate Shiftcited by 10
  4. 2024One-Shot Sequential Federated Learning for Non-IID Data by Enhancing Local Model Diversitycited by 8
  5. 2024Generalizable Person Re-identification via Balancing Alignment and Uniformitycited by 8
  6. 2024Learning Fair Invariant Representations under Covariate and Correlation Shifts Simultaneouslycited by 7
  7. 2023SAILOR: Structural Augmentation Based Tail Node Representation Learningcited by 6
  8. 2023MaxEnt Loss: Constrained Maximum Entropy for Calibration under Out-of-Distribution Shiftcited by 6
  9. 2023On the Benefits of Public Representations for Private Transfer Learning under Distribution Shiftcited by 5
  10. 2023Continual Learning in Predictive Autoscalingcited by 4
  11. 2023Class Distribution Shifts in Zero-Shot Learning: Learning Robust Representationscited by 4
  12. 2023CASUAL: Conditional Support Alignment for Domain Adaptation with Label Shiftcited by 3
  13. 2023Fairness under Covariate Shift: Improving Fairness-Accuracy tradeoff with few Unlabeled Test Samplescited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Fed-ADE: Adaptive Learning Rate for Federated Post-adaptation under Distribution Shiftcited by 2
  2. 2026Multi2: Hierarchical Multi-Agent Decision-Making with LLM-Based Agents in Interactive Environmentscited by 2
  3. 2026Post-Hoc Merging is Not Enough: Many-Shot Model Merging with Loss-Gap Balancingcited by 1