.png){kind=icon}
返回
Aggressive Approximation of the SoftMax Function for Power-Efficient Hardware Implementations
.png){kind=icon}
PRE
AI
DOI：10.1109/TCSII.2021.3120495.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Neural Network models most often exploit the SoftMax function in the classification stage for computing probabilities through exponentiation and division operations. To reduce the complexity and the energy consumption of such stage, several hardware-friendly approximation strategies have been disclosed in the recent past. This brief evaluates the effects of an aggressive approximation of the SoftMax layer on both classification accuracy and hardware characteristics. Experimental results demonstrate that the proposed circuit, when implemented in a 28 nm FDSOI technology, saves similar to 65% of silicon area with respect to competitors, dissipating less than 1 pJ. FPGA implementation results confirm a massive energy dissipation reduction with respect to the conventional baseline architecture, without introducing penalties in the Top-1 accuracy.
Keyword:
Computer architecture
Table lookup
Hardware
Approximation algorithms
Fitting
Energy consumption
Task analysis
Deep neural networks
SoftMax
approximate computing
low-power hardware architectures
期刊
I
IF:
4.9
论文数:
8.8K
被引数:
2.5W
机构
引用论文
A Multi-Class Objects Detection Coprocessor With Dual Feature Space and Weighted Softmax具有双重特征空间和加权Softmax的多类目标检测协处理器
PREAI
分享收藏Fibroblast Growth Factor-4 Enhances Proliferation of Mouse Embryonic Stem Cells via Activation of c-Jun Signaling.png)
PLoS ONE
OAAI
PLoS ONEIF0
OAAI分享收藏A pseudo-softmax function for hardware-based high speed image classificationSCIENTIFIC REPORTS
OAAI
SCIENTIFIC REPORTSIF3.9
OAAI分享收藏Effects of Essential Amino Acid Deficiencies on Syngeneic Tumor Immunity and Carcinogenesis in MiceOAAI
OAAI分享收藏
查看更多内容