论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot

Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot Collaboration

IEEE Transactions on Cybernetics ( Early Access ) 12 October 2021

task

人机交互中的意图识别

阅读记录

论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第1张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第2张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第3张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第4张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第5张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第6张图片
Softsign 是 Tanh 激活函数的另一个替代选择。就像 Tanh 一样,Softsign 是反对称、去中心、可微分,并返回-1 和 1 之间的值。其更平坦的曲线与更慢的下降导数表明它可以更高效地学习,比tanh更好的解决梯度消失的问题。另一方面,导数的计算比 Tanh 更麻烦;在实践中,可以深度用softsign替代tanh激活函数
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第7张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第8张图片
Hardsigmoid用于门函数(输入 遗忘 输出)
Softsign用于状态激活 和隐藏层
输入门决定值是否应该更新,遗忘门允许遗忘和丢弃信息,输出门和块输出选择输出信息。
存储在这个网络中的新状态ct是新的门控输入和之前的门控状态ct−1的总和
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第9张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第10张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第11张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第12张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第13张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第14张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第15张图片
出于测试仪器工作不稳定而使信号发生偶然跳动或因外界异常干扰的影响,往往会导致测量数据中包含一些很不合理的跳点。如虚线框内额外为多个连续的多个野值。
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第16张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第17张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第18张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第19张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第20张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第21张图片
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第22张图片

说明

以上内容均为作者本人平时阅读并且汇报使用,内容整理全凭个人理解,如有侵权,请联系我;内容如有错误,欢迎留言交流。转载请注明出处,并附有原文链接,谢谢!

此外,我还喜欢用ipad对论文写写画画(个人英文阅读的水平有限),做一些断句、重点勾画等,有兴趣大家可以按需下载:链接
论文阅读 Hybrid Recurrent Neural Network Architecture-Based Intention Recognition for Human-Robot_第23张图片
更多论文分享,请参考: 深度学习相关阅读论文汇总(持续更新

你可能感兴趣的:(深度学习论文阅读,计算机视觉,人工智能,神经网络,深度学习,cnn)