Linear Probing
الاختبار الخطي
بروتوكول تقييم يُدرَّب فيه مصنّف خطي وحيد فوق سمات مُدرَّبة مسبقاً مُجمَّدة. يختبر مدى قابلية التمثيلات المُتعلَّمة للفصل الخطي، لكنه لا يكشف عن قوة السمات اللاخطية.
An evaluation protocol where a single linear classifier is trained on frozen pre-trained features. It tests how linearly separable the learned representations are, but does not reveal the strength of non-linear features.
Also translated asالتقييم بالمصنّف الخطي، الفحص الخطي
First appears in this corpus in: Momentum Contrast for Unsupervised Visual Representation Learning (2020)
Appears in these papers
- Emerging Properties in Self-Supervised Vision Transformers2021in the sky ✦
- Emerging Properties in Self-Supervised Vision Transformers2021in the sky ✦
- DINOv2: Learning Robust Visual Features Without Supervision2023in the sky ✦
- DINOv2: Learning Robust Visual Features Without Supervision2023in the sky ✦
- Masked Autoencoders Are Scalable Vision Learners2022in the sky ✦
- Momentum Contrast for Unsupervised Visual Representation Learning2020in the sky ✦
- A Time Series Is Worth 64 Words: Long-Term Forecasting with Transformers2023in the sky ✦
- A Time Series Is Worth 64 Words: Long-Term Forecasting with Transformers2023in the sky ✦
- A Simple Framework for Contrastive Learning of Visual Representations2020in the sky ✦
- A Simple Framework for Contrastive Learning of Visual Representations2020in the sky ✦