k-NN as a Simple and Effective Estimator of Transferability
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918051920740352 |
|---|---|
| author | Sorkhei, Moein Matsoukas, Christos Haslum, Johan Fredin Konuk, Emir Smith, Kevin |
| author_facet | Sorkhei, Moein Matsoukas, Christos Haslum, Johan Fredin Konuk, Emir Smith, Kevin |
| contents | How well can one expect transfer learning to work in a new setting where the domain is shifted, the task is different, and the architecture changes? Many transfer learning metrics have been proposed to answer this question. But how accurate are their predictions in a realistic new setting? We conducted an extensive evaluation involving over 42,000 experiments comparing 23 transferability metrics across 16 different datasets to assess their ability to predict transfer performance. Our findings reveal that none of the existing metrics perform well across the board. However, we find that a simple k-nearest neighbor evaluation -- as is commonly used to evaluate feature quality for self-supervision -- not only surpasses existing metrics, but also offers better computational efficiency and ease of implementation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_18528 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | k-NN as a Simple and Effective Estimator of Transferability Sorkhei, Moein Matsoukas, Christos Haslum, Johan Fredin Konuk, Emir Smith, Kevin Machine Learning Computer Vision and Pattern Recognition How well can one expect transfer learning to work in a new setting where the domain is shifted, the task is different, and the architecture changes? Many transfer learning metrics have been proposed to answer this question. But how accurate are their predictions in a realistic new setting? We conducted an extensive evaluation involving over 42,000 experiments comparing 23 transferability metrics across 16 different datasets to assess their ability to predict transfer performance. Our findings reveal that none of the existing metrics perform well across the board. However, we find that a simple k-nearest neighbor evaluation -- as is commonly used to evaluate feature quality for self-supervision -- not only surpasses existing metrics, but also offers better computational efficiency and ease of implementation. |
| title | k-NN as a Simple and Effective Estimator of Transferability |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.18528 |