FitNets: Hints for Thin Deep Nets
其實應該先早點寫這篇文章的 這篇文章主要是將hinton的output distillation擴展到了feature distillation 該loss用來拉進student和teacher feature的距離 該loss就是與hard label、soft label做cross entroy 訓練過程需要注意： 先進行hints training，即選擇某一層feature對齊後，利用H
相關文章
- 1. FitNets: Hints for thin deep nets論文筆記
- 2. 知識蒸餾（Distillation）相關論文閱讀（3）—— FitNets : Hints for Thin Deep Nets
- 3. 系列論文閱讀之知識蒸餾（二）《FitNets : Hints for Thin Deep Nets》
- 4. A Taxonomy of Deep Convolutional Neural Nets for Computer Vision
- 5. A Fast Learning Algorithm for Deep Belief Nets
- 6. Deep Convolutional Nets for Semantic Image Segmentation with Deep Gaussian CRFs
- 7. CHAPTER 11-Training Deep Neural Nets-part3
- 8. Do Deep Nets Really Need to be Deep?
- 9. 「Deep Learning」Note on Deeply-Supervised Nets
- 10. Training Deep Nets with Sublinear Memory Cost
- 更多相關文章...
- • Swift for 循環 - Swift 教程
- • Scala for循環 - Scala教程
- • PHP開發工具
- • C# 中 foreach 遍歷的用法
相關標籤/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。
最新文章
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. FitNets: Hints for thin deep nets論文筆記
- 2. 知識蒸餾（Distillation）相關論文閱讀（3）—— FitNets : Hints for Thin Deep Nets
- 3. 系列論文閱讀之知識蒸餾（二）《FitNets : Hints for Thin Deep Nets》
- 4. A Taxonomy of Deep Convolutional Neural Nets for Computer Vision
- 5. A Fast Learning Algorithm for Deep Belief Nets
- 6. Deep Convolutional Nets for Semantic Image Segmentation with Deep Gaussian CRFs
- 7. CHAPTER 11-Training Deep Neural Nets-part3
- 8. Do Deep Nets Really Need to be Deep?
- 9. 「Deep Learning」Note on Deeply-Supervised Nets
- 10. Training Deep Nets with Sublinear Memory Cost