Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2508.13445 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918127116222464 |
|---|---|
| author | Park, Heewon Joe, Mugon Kim, Miru Kwon, Minhae |
| author_facet | Park, Heewon Joe, Mugon Kim, Miru Kwon, Minhae |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_13445 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ASAP: Unsupervised Post-training with Label Distribution Shift Adaptive Learning Rate Park, Heewon Joe, Mugon Kim, Miru Kwon, Minhae Machine Learning 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. |
| title | ASAP: Unsupervised Post-training with Label Distribution Shift Adaptive Learning Rate |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2508.13445 |