Tutorial: Knowledge Distillation
概述 Knowledge Distillation(KD)通常指利用一個大的teacher網絡做爲監督，幫助一個小的student網絡進行學習，主要用於模型壓縮。 其方法主要分爲兩大類web Output Distillation Feature Distillation Output Distillation Motivation 主要拉近teacher和student最終輸出的距離，參考論文：
相關文章
- 1. Tutorial: Knowledge Distillation
- 2. Knowledge Distillation
- 3. Knowledge Distillation 筆記
- 4. Awesome Knowledge-Distillation
- 5. Similarity-Preserving Knowledge Distillation
- 6. 論文Relational Knowledge Distillation
- 7. Correlation Congruence for Knowledge Distillation
- 8. knowledge distillation 論文閱讀之：ResKD: Residual-Guided Knowledge Distillation
- 9. knowledge distillation 論文閱讀之：Triplet Loss for Knowledge Distillation
- 10. Knowledge Distillation論文閱讀（2）：Learning Efficient Object Detection Models with Knowledge Distillation
- 更多相關文章...
- • C# 結構體（Struct） - C#教程
- • Lua 數組 - Lua 教程
- • Git五分鐘教程
- • Flink 數據傳輸及反壓詳解
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。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. Tutorial: Knowledge Distillation
- 2. Knowledge Distillation
- 3. Knowledge Distillation 筆記
- 4. Awesome Knowledge-Distillation
- 5. Similarity-Preserving Knowledge Distillation
- 6. 論文Relational Knowledge Distillation
- 7. Correlation Congruence for Knowledge Distillation
- 8. knowledge distillation 論文閱讀之：ResKD: Residual-Guided Knowledge Distillation
- 9. knowledge distillation 論文閱讀之：Triplet Loss for Knowledge Distillation
- 10. Knowledge Distillation論文閱讀（2）：Learning Efficient Object Detection Models with Knowledge Distillation