- Deep Learning Essentials
- Wei Di Anurag Bhardwaj Jianing Wei
- 51字
- 2021-06-30 19:17:52
Tanh or hyperbolic tangent function
The mathematical formula of tanh is as follows:
Its output is centered at zero with a range of -1 to 1. Therefore, optimization is easier and thus in practice, it is preferred over a sigmoid activation function. However, it still suffers from the vanishing gradient problem.
推薦閱讀
- 基于C語言的程序設(shè)計(jì)
- 集成架構(gòu)中型系統(tǒng)
- Internet接入·網(wǎng)絡(luò)安全
- 樂高機(jī)器人EV3設(shè)計(jì)指南:創(chuàng)造者的搭建邏輯
- TestStand工業(yè)自動(dòng)化測(cè)試管理(典藏版)
- UTM(統(tǒng)一威脅管理)技術(shù)概論
- VMware Performance and Capacity Management(Second Edition)
- 3D Printing for Architects with MakerBot
- INSTANT Autodesk Revit 2013 Customization with .NET How-to
- 愛犯錯(cuò)的智能體
- 水下無線傳感器網(wǎng)絡(luò)的通信與決策技術(shù)
- 空間站多臂機(jī)器人運(yùn)動(dòng)控制研究
- 從零開始學(xué)PHP
- 單片機(jī)技能與實(shí)訓(xùn)
- AVR單片機(jī)工程師是怎樣煉成的