paper-with-me

Papers

CAKD: A Correlation-Aware Knowledge Distillation Framework Based on Decoupling Kullback-Leibler Divergence

2024-10-17 · Zao Zhang, Huaming Chen, Pei Ning, Nan Yang, Dong Yuan

In knowledge distillation, a primary focus has been on transforming and balancing multiple distillation components. In this work, we emphasize the importance of thoroughly examining each distillation component, as we observe that not all elements are equally crucial. From this perspective,we decouple the Kullback-Leibler (KL) divergence into three unique elements: Binary Classification Divergence (BCD), Strong Correlation Divergence (SCD), and Weak Correlation Divergence (WCD). Each of these elements presents varying degrees of influence. Leveraging these insights, we present the Correlation-Aware Knowledge Distillation (CAKD) framework. CAKD is designed to prioritize the facets of the distillation components that have the most substantial influence on predictions, thereby optimizing knowledge transfer from teacher to student models. Our experiments demonstrate that adjusting the effect of each element enhances the effectiveness of knowledge transformation. Furthermore, evidence shows that our novel CAKD framework consistently outperforms the baseline across diverse models and datasets. Our work further highlights the importance and effectiveness of closely examining the impact of different parts of distillation process.

📄 PDF Abstract BibTeX arXiv:2410.14741

Code (0)

등록된 구현이 없습니다.

Tasks

Binary ClassificationKnowledge DistillationTransfer Learning

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.…
Focus 설명 없음

Similar Papers 제목 키워드 기반

CaKDP: Category-aware Knowledge Distillation and Pruning Framework for Lightweight 3D Object Detection

2024-01-01 · CVPR 2024 1 · Haonan Zhang, Longjun Liu, Yuqi Huang, Zhao Yang 외

Knowledge distillation (KD) possesses immense potential to accelerate the deep neural networks (DNNs) for LiDAR-based 3D detection. However in most of prevailing approaches the suboptimal teacher models and insuffici…

3D Object DetectionKnowledge Distillationobject-detectionObject Detection

Cross Architecture Distillation for Face Recognition

2023-06-26 · Weisong Zhao, Xiangyu Zhu, Zhixiang He, Xiao-Yu Zhang 외

Transformers have emerged as the superior choice for face recognition tasks, but their insufficient platform acceleration hinders their application on mobile devices. In contrast, Convolutional Neural Networks (CNNs) cap…

Face RecognitionKnowledge Distillation

TAS: Distilling Arbitrary Teacher and Student via a Hybrid Assistant

2024-10-16 · Guopeng Li, Qiang Wang, Ke Yan, Shouhong Ding 외

Most knowledge distillation (KD) methodologies predominantly focus on teacher-student pairs with similar architectures, such as both being convolutional neural networks (CNNs). However, the potential and flexibility of K…

Knowledge DistillationTransfer Learning

Revisiting Cross-Architecture Distillation: Adaptive Dual-Teacher Transfer for Lightweight Video Models

2025-11-12 · Ying Peng, Hongsen Ye, Changxin Huang, Xiping Hu 외 arxiv

Vision Transformers (ViTs) have achieved strong performance in video action recognition, but their high computational cost limits their practicality. Lightweight CNNs are more efficient but suffer from accuracy gaps. Cro…

Knowledge DistillationAction Recognition

YOLO-IOD: Towards Real Time Incremental Object Detection

2025-12-28 · Shizhou Zhang, Xueqiang Lv, Yinghui Xing, Qirui Wu 외 arxiv

Current methods for incremental object detection (IOD) primarily rely on Faster R-CNN or DETR series detectors; however, these approaches do not accommodate the real-time YOLO detection frameworks. In this paper, we firs…

parameter-efficient fine-tuningKnowledge DistillationIncremental LearningObject Detection