paper-with-me

홈 › Papers

Knowledge Distillation: Bad Models Can Be Good Role Models

2022-03-28 · Gal Kaplun, Eran Malach, Preetum Nakkiran, Shai Shalev-Shwartz

Large neural networks trained in the overparameterized regime are able to fit noise to zero train error. Recent work \citep{nakkiran2020distributional} has empirically observed that such networks behave as "conditional samplers" from the noisy distribution. That is, they replicate the noise in the train data to unseen examples. We give a theoretical framework for studying this conditional sampling behavior in the context of learning theory. We relate the notion of such samplers to knowledge distillation, where a student network imitates the outputs of a teacher on unlabeled data. We show that samplers, while being bad classifiers, can be good teachers. Concretely, we prove that distillation from samplers is guaranteed to produce a student which approximates the Bayes optimal classifier. Finally, we show that some common learning algorithms (e.g., Nearest-Neighbours and Kernel Machines) can generate samplers when applied in the overparameterized regime.

📄 PDF Abstract BibTeX arXiv:2203.14649

Code (0)

등록된 구현이 없습니다.

Tasks

Knowledge DistillationLearning Theory

Similar Papers 제목 키워드 기반

ORC: Network Group-based Knowledge Distillation using Online Role Change

2022-06-01 · ICCV 2023 1 · Junyong Choi, Hyeon Cho, Seokhwa Cheung, Wonjun Hwang

In knowledge distillation, since a single, omnipotent teacher network cannot solve all problems, multiple teacher-based knowledge distillations have been studied recently. However, sometimes their improvements are not as…

Knowledge Distillation

Knowledge Distillation with Adversarial Samples Supporting Decision Boundary

2018-05-15 · Byeongho Heo, Minsik Lee, Sangdoo Yun, Jin Young Choi

Many recent works on knowledge distillation have provided ways to transfer the knowledge of a trained network for improving the learning process of a new one, but finding a good technique for knowledge distillation is st…

Adversarial AttackKnowledge Distillation

The Role of Masking for Efficient Supervised Knowledge Distillation of Vision Transformers

2023-02-21 · Seungwoo Son, Jegwang Ryu, Namhoon Lee, Jaeho Lee

Knowledge distillation is an effective method for training lightweight vision models. However, acquiring teacher supervision for training samples is often costly, especially from large-scale models like vision transforme…

Knowledge Distillation

What Makes a Good Dataset for Knowledge Distillation?

2024-11-19 · CVPR 2025 1 · Logan Frank, Jim Davis

Knowledge distillation (KD) has been a popular and effective method for model compression. One important assumption of KD is that the teacher's original dataset will also be available when training the student. However, …

Continual LearningKnowledge DistillationModel Compression

Improved Knowledge Distillation for Pre-trained Language Models via Knowledge Selection

2023-02-01 · Chenglong Wang, Yi Lu, Yongyu Mu, Yimin Hu 외

Knowledge distillation addresses the problem of transferring knowledge from a teacher model to a student model. In this process, we typically have multiple types of knowledge extracted from the teacher model. The problem…

Knowledge Distillation