Enregistré dans:
Détails bibliographiques
Auteurs principaux: Cooper, Nicholas, Chen, Lijun, Dwivedy, Sailesh, Gurari, Danna
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2511.14981
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866908664848187392
author Cooper, Nicholas
Chen, Lijun
Dwivedy, Sailesh
Gurari, Danna
author_facet Cooper, Nicholas
Chen, Lijun
Dwivedy, Sailesh
Gurari, Danna
contents Knowledge distillation (KD) methods can transfer knowledge of a parameter-heavy teacher model to a light-weight student model. The status quo for feature KD methods is to utilize loss functions based on logits (i.e., pre-softmax class scores) and intermediate layer features (i.e., latent representations). Unlike previous approaches, we propose a feature KD framework for training the student's backbone using feature-based losses exclusively (i.e., without logit-based losses such as cross entropy). Leveraging recent discoveries about the geometry of latent representations, we introduce a knowledge quality metric for identifying which teacher layers provide the most effective knowledge for distillation. Experiments on three image classification datasets with four diverse student-teacher pairs, spanning convolutional neural networks and vision transformers, demonstrate our KD method achieves state-of-the-art performance, delivering top-1 accuracy boosts of up to 15% over standard approaches. We publically share our code to facilitate future work at https://github.com/Thegolfingocto/KD_wo_CE.
format Preprint
id arxiv_https___arxiv_org_abs_2511_14981
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Logit-Based Losses Limit the Effectiveness of Feature Knowledge Distillation
Cooper, Nicholas
Chen, Lijun
Dwivedy, Sailesh
Gurari, Danna
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
I.2.6
Knowledge distillation (KD) methods can transfer knowledge of a parameter-heavy teacher model to a light-weight student model. The status quo for feature KD methods is to utilize loss functions based on logits (i.e., pre-softmax class scores) and intermediate layer features (i.e., latent representations). Unlike previous approaches, we propose a feature KD framework for training the student's backbone using feature-based losses exclusively (i.e., without logit-based losses such as cross entropy). Leveraging recent discoveries about the geometry of latent representations, we introduce a knowledge quality metric for identifying which teacher layers provide the most effective knowledge for distillation. Experiments on three image classification datasets with four diverse student-teacher pairs, spanning convolutional neural networks and vision transformers, demonstrate our KD method achieves state-of-the-art performance, delivering top-1 accuracy boosts of up to 15% over standard approaches. We publically share our code to facilitate future work at https://github.com/Thegolfingocto/KD_wo_CE.
title Logit-Based Losses Limit the Effectiveness of Feature Knowledge Distillation
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
I.2.6
url https://arxiv.org/abs/2511.14981