Label Smoothing

تنعيم التسميات

أسلوب تنظيم يستبدل الأهداف الصلبة (ترميز أحادي ساخن) بمزيج مرجّح من التسمية الحقيقية والتوزيع المنتظم على جميع الفئات، ما يمنع الشبكة من اكتساب ثقة مفرطة ويُحسّن التعميم والمعايرة.

A regularization technique that replaces hard one-hot targets with a weighted mixture of the true label and the uniform distribution over all classes, preventing the network from becoming over-confident and improving generalization and calibration.

Also translated asتمليس التسميات، تنعيم الملصقات

First appears in this corpus in: Distilling the Knowledge in a Neural Network (2015)

Appears in these papers

ALIGN: Scaling Up Visual and Vision-Language Representation Learning with Noisy Text Supervision2021 in the sky ✦
ALIGN: Scaling Up Visual and Vision-Language Representation Learning with Noisy Text Supervision2021 in the sky ✦
BART: Denoising Sequence-to-Sequence Pre-Training for Natural Language Generation, Translation, and Comprehension2019 in the sky ✦
A ConvNet for the 2020s2022 in the sky ✦
Training Data-Efficient Image Transformers & Distillation Through Attention2021 in the sky ✦
Training Data-Efficient Image Transformers & Distillation Through Attention2021 in the sky ✦
Distilling the Knowledge in a Neural Network2015 in the sky ✦
When Does Label Smoothing Help?2019 in the sky ✦
When Does Label Smoothing Help?2019 in the sky ✦
mixup: Beyond Empirical Risk Minimization2018 in the sky ✦
mixup: Beyond Empirical Risk Minimization2018 in the sky ✦