تقرير
On consequences of finetuning on data with highly discriminative features
| العنوان: | On consequences of finetuning on data with highly discriminative features |
|---|---|
| المؤلفون: | Masarczyk, Wojciech, Trzciński, Tomasz, Ostaszewski, Mateusz |
| سنة النشر: | 2023 |
| المجموعة: | Computer Science |
| مصطلحات موضوعية: | Computer Science - Machine Learning |
| الوصف: | In the era of transfer learning, training neural networks from scratch is becoming obsolete. Transfer learning leverages prior knowledge for new tasks, conserving computational resources. While its advantages are well-documented, we uncover a notable drawback: networks tend to prioritize basic data patterns, forsaking valuable pre-learned features. We term this behavior "feature erosion" and analyze its impact on network performance and internal representations. Comment: NeurIPS 2023 -- UniReps Workshop |
| نوع الوثيقة: | Working Paper |
| URL الوصول: | http://arxiv.org/abs/2310.19537 |
| رقم الأكسشن: | edsarx.2310.19537 |
| قاعدة البيانات: | arXiv |
| الوصف غير متاح. |