Fisher-DARTS: A Neural Architecture Search Framework with Fisher Information Optimization
Differentiable Neural Architecture Search has emerged as a powerful paradigm for automated network design, yet it suffers from a fundamental optimization inconsistency problem: Architectures optimized under continuous relaxation often fail to maintain their performance after discretization. To addre...
Сохранить в:
| Главные авторы: | , |
|---|---|
| Формат: | Artigo |
| Язык: | Inglês |
| Опубликовано: |
MDPI AG
2026-04-01
|
| Серии: | Applied Sciences |
| Предметы: | |
| Online-ссылка: | https://www.mdpi.com/2076-3417/16/8/3808 |
| Метки: |
Нет меток, Требуется 1-ая метка записи!
|