paper-with-me

Papers

Revisiting Knowledge Distillation under Distribution Shift

2023-12-25 · Songming Zhang, Ziyu Lyu, Xiaofeng Chen

Knowledge distillation transfers knowledge from large models into small models, and has recently made remarkable achievements. However, few studies has investigated the mechanism of knowledge distillation against distribution shift. Distribution shift refers to the data distribution drifts between training and testing phases. In this paper, we reconsider the paradigm of knowledge distillation by reformulating the objective function in shift situations. Under the real scenarios, we propose a unified and systematic framework to benchmark knowledge distillation against two general distributional shifts including diversity and correlation shift. The evaluation benchmark covers more than 30 methods from algorithmic, data-driven, and optimization perspectives for five benchmark datasets. Overall, we conduct extensive experiments on the student model. We reveal intriguing observations of poor teaching performance under distribution shifts; in particular, complex algorithms and data augmentation offer limited gains in many cases.

📄 PDF Abstract BibTeX arXiv:2312.16242

Code (1)

zzhangsm/ookd 공식 구현 pytorch

Tasks

Data AugmentationDiversityKnowledge Distillation

Methods 이 논문이 사용한 방법론

Knowledge Distillation A very simple way to improve the performance of almost any machine learning algorithm is to train many different models on the same data and then to average their predictions.…

Similar Papers 제목 키워드 기반

Momentum Adversarial Distillation: Handling Large Distribution Shifts in Data-Free Knowledge Distillation

2022-09-21 · Kien Do, Hung Le, Dung Nguyen, Dang Nguyen 외

Data-free Knowledge Distillation (DFKD) has attracted attention recently thanks to its appealing capability of transferring knowledge from a teacher network to a student network without using training data. The main idea…

Data-free Knowledge DistillationKnowledge Distillation

Understanding the Effect of Data Augmentation on Knowledge Distillation

2023-05-21 · Ziqi Wang, Chi Han, Wenxuan Bao, Heng Ji

Knowledge distillation (KD) requires sufficient data to transfer knowledge from large-scale teacher models to small-scale student models. Therefore, data augmentation has been widely used to mitigate the shortage of data…

Data AugmentationKnowledge Distillation

De-confounded Data-free Knowledge Distillation for Handling Distribution Shifts

2024-03-28 · CVPR 2024 1 · Yuzheng Wang, Dingkang Yang, Zhaoyu Chen, Yang Liu 외

Data-Free Knowledge Distillation (DFKD) is a promising task to train high-performance small models to enhance actual deployment without relying on the original training data. Existing methods commonly avoid relying on pr…

Causal InferenceData-free Knowledge DistillationKnowledge Distillation

Distribution Shift Matters for Knowledge Distillation with Webly Collected Images

2023-07-21 · ICCV 2023 1 · Jialiang Tang, Shuo Chen, Gang Niu, Masashi Sugiyama 외

Knowledge distillation aims to learn a lightweight student network from a pre-trained teacher network. In practice, existing knowledge distillation methods are usually infeasible when the original training data is unavai…

Contrastive LearningData-free Knowledge DistillationKnowledge DistillationManagement+1

Revisiting Data-Free Knowledge Distillation with Poisoned Teachers

2023-06-04 · Junyuan Hong, Yi Zeng, Shuyang Yu, Lingjuan Lyu 외

Data-free knowledge distillation (KD) helps transfer knowledge from a pre-trained model (known as the teacher model) to a smaller model (known as the student model) without access to the original training data used for t…

Backdoor Defense for Data-Free Distillation with Poisoned TeachersData-free Knowledge DistillationKnowledge Distillation