Cambricon-X: An Accelerator for Sparse Neural Networks
摘要 神經網絡（NNs）已被證明在廣泛的應用中很有用，例如圖像識別，自動翻譯和廣告推薦。由於不斷增加的深層結構，即具有大量神經元和連接（即突觸）的多層結構，因此，最新的NN既需要大量計算又需要大量內存。稀疏神經網絡已成爲減少所需計算量和內存量的有效解決方案。儘管現有的NN加速器能夠有效處理密集且規則的網絡，但它們無法從突觸權重的降低中受益。在本文中，我們提出了一種新型的加速器Cambricon-X
相關文章
- 1. 論文筆記：An Efficient Hardware Accelerator for Structured Sparse Convolutional Neural Networks on FPGAs
- 2. 論文筆記：An Efficient Hardware Accelerator for Sparse Convolutional Neural Networks on FPGAs
- 3. ReLU——Deep Sparse Rectifier Neural Networks
- 4. EffNet: An Efficient Structure for Convolutional Neural Networks
- 5. [ICLR19] THE LOTTERY TICKET HYPOTHESIS： FINDING SPARSE, TRAINABLE NEURAL NETWORKS
- 6. Multiresolution Recurrent Neural Networks: An Application to...
- 7. Data-Driven Sparse Structure Selection for Deep Neural Networks 論文翻譯
- 8. Global Sparse Momentum SGD for Pruning Very Deep Neural Networks 論文閱讀筆記
- 9. 【論文閱讀筆記】Data-Driven Sparse Structure Selection for Deep Neural Networks
- 10. Understanding Convolutional Neural Networks for NLP
- 更多相關文章...
- • Swift for 循環 - Swift 教程
- • Scala for循環 - Scala教程
- • RxJava操作符（一）Creating Observables
- • PHP開發工具
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. 論文筆記：An Efficient Hardware Accelerator for Structured Sparse Convolutional Neural Networks on FPGAs
- 2. 論文筆記：An Efficient Hardware Accelerator for Sparse Convolutional Neural Networks on FPGAs
- 3. ReLU——Deep Sparse Rectifier Neural Networks
- 4. EffNet: An Efficient Structure for Convolutional Neural Networks
- 5. [ICLR19] THE LOTTERY TICKET HYPOTHESIS： FINDING SPARSE, TRAINABLE NEURAL NETWORKS
- 6. Multiresolution Recurrent Neural Networks: An Application to...
- 7. Data-Driven Sparse Structure Selection for Deep Neural Networks 論文翻譯
- 8. Global Sparse Momentum SGD for Pruning Very Deep Neural Networks 論文閱讀筆記
- 9. 【論文閱讀筆記】Data-Driven Sparse Structure Selection for Deep Neural Networks
- 10. Understanding Convolutional Neural Networks for NLP