通过近似 Softmax 与 Squash 运算实现边缘端胶囊网络
机器学习
2022-06-22 v1
摘要
诸如胶囊网络(CapsNets)的复杂深度神经网络以计算密集型运算为代价展现出高学习能力。为使其能部署于边缘设备,我们提议利用近似计算来设计如 softmax 和 squash 等复杂运算的近似变体。在我们的实验中,我们评估了采用 ASIC 设计流程实现的设计在面积、功耗和关键路径延迟方面的权衡,以及量化 CapsNets 相比于精确函数的精度。
引用
@article{arxiv.2206.10200,
title = {Enabling Capsule Networks at the Edge through Approximate Softmax and Squash Operations},
author = {Alberto Marchisio and Beatrice Bussolino and Edoardo Salvati and Maurizio Martina and Guido Masera and Muhammad Shafique},
journal= {arXiv preprint arXiv:2206.10200},
year = {2022}
}
备注
To appear at the ACM/IEEE International Symposium on Low Power Electronics and Design (ISLPED), August 2022, Boston, MA, USA