Correlation Congruence for Knowledge Distillation
Motivation 目前大多数KD框架的pipeline 一张图片送入teacher和student得到各自的feature和output 然后定义loss，拉进feature或者output的距离 但是他们都是直接在一个整图级别上做的，student很难学习到teacher的那种处理类内和类间的能力 如下图：白色空心圈是期望学习到的效果 左边是期望类间可以拉开，右边是期望类内可以聚拢 Meth
相关文章
- 1. Awesome Knowledge-Distillation
- 2. Knowledge Distillation
- 3. Structured Knowledge Distillation for Semantic Segmentation
- 4. Knowledge Distillation for Segmentation 笔记
- 5. knowledge distillation 论文阅读之：Triplet Loss for Knowledge Distillation
- 6. knowledge distillation论文阅读之：Learning from a Lightweight Teacher for Efficient Knowledge Distillation
- 7. Knowledge Distillation via Route Constrained Optimization
- 8. Knowledge Distillation 笔记
- 9. Tutorial: Knowledge Distillation
- 10. Similarity-Preserving Knowledge Distillation
- 更多相关文章...
- • Swift for 循环 - Swift 教程
- • Scala for循环 - Scala教程
- • PHP开发工具
- • C# 中 foreach 遍历的用法
相关标签/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
欢迎关注本站公众号,获取更多信息
相关文章
- 1. Awesome Knowledge-Distillation
- 2. Knowledge Distillation
- 3. Structured Knowledge Distillation for Semantic Segmentation
- 4. Knowledge Distillation for Segmentation 笔记
- 5. knowledge distillation 论文阅读之：Triplet Loss for Knowledge Distillation
- 6. knowledge distillation论文阅读之：Learning from a Lightweight Teacher for Efficient Knowledge Distillation
- 7. Knowledge Distillation via Route Constrained Optimization
- 8. Knowledge Distillation 笔记
- 9. Tutorial: Knowledge Distillation
- 10. Similarity-Preserving Knowledge Distillation