Skip to Content
从零开始学习深度学习
book

从零开始学习深度学习

by Seth Weidman
May 2025
Beginner to intermediate
252 pages
3h 16m
Chinese
O'Reilly Media, Inc.
Content preview from 从零开始学习深度学习

附录 A. 深度调查

在本节中,我们将深入探讨对完成工作非常重要但并非必不可少的几个技术领域。

矩阵链规则

首先, ,解释为什么我们可以用WT代替 ν u ( X ) 第 1 章的链式规则表达式中。

记住,L是字面意思:

σ ( X W 11 ) + σ ( X W 12 ) + σ ( X W 21 ) + σ ( X W 22 ) + σ ( X W 31 ) + σ ( X W 32 )

在这里,这是一个事实的简称:

σ ( X W 11 ) = σ ( x 11 × w 11 + x 12 × w 21 + x 13 × w 31 )
σ ( X W 12 ) = σ ( x 11 × w 12 + x 12 × w 22 + x 13 × w 32 )

等等。让我们放大这些表达式中的一个。比如说,如果我们对 σ 进行偏导数运算,结果会是怎样? σ ( X W 11 ) 关于 X (的所有六个分量(这也是我们最终要对 L )?

嗯,因为

σ ( X W 11 ) = σ ( x 11 × w 11 + x 12 × w 21 + x 13 × w 31 )

不难看出,与 x 1 的偏导数:

σ u ( X W 11 ) × w 11

由于XW11表达式中x11乘以的唯一值是w11,因此相对于其他值的偏导数为 0。

因此,计算σ(XW11)相对于X的所有元素的偏导数,可以得到以下总体表达式 σ(XW 11 ) X :

σ(XW 11 ) X = σ u ( X W 11 ) × w 11 σ u ( X W 11 ) × w 21 σ u ( X W 11 ) × w 31 0 0 0 0 0 0

同样,我们可以计算出σ(XW32) 相对于X 的每个元素的偏导数:

0 0 0 0 0 0 σ u ( X W 32 ) × w 12 σ u ( X W 32 ) × w 22 σ u ( X W 32 ) × w 32

现在我们有了实际计算的所有组件 Λ X ( S ) 直接计算。我们只需计算与前面矩阵形式相同的六个矩阵,然后将结果相加即可。

请注意,数学运算再次变得混乱,不过并不复杂。你可以跳过下面的计算,直接看结论,结论最终是一个简单的表达式。但是,通过计算,你会更深刻地体会到结论是多么出人意料的简单。除了欣赏,生活还有什么意义呢?

这里只有两个步骤。首先,我们要明确写出 Λ X ( S ) 是上述六个矩阵之和:

Λ X ( S ) = σ(XW 11 ) X + σ(XW 12 ) X + σ(XW 21 ) X + σ(XW 22 ) X + σ(XW 31 ) X + σ(XW 32 ) X = σ u ( X W 11 ) × w 11 σ u ( X W 11 ) × w 21 σ u ( X W 11 ) × w 31 0 0 0 0
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

AirBnbBlueOriginElectronic ArtsHomeDepotNasdaqRakutenTata Consultancy Services

QuotationMarkO’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.
Julian F.
Head of Cybersecurity
QuotationMarkI wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.
Addison B.
Field Engineer
QuotationMarkI’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.
Amir M.
Data Platform Tech Lead
QuotationMarkI'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.
Mark W.
Embedded Software Engineer

You might also like

深度学习入门4 :强化学习

深度学习入门4 :强化学习

斋藤康毅
Python算法交易实战

Python算法交易实战

Posts & Telecom Press, Sebastien Donadio
R深度学习权威指南

R深度学习权威指南

Posts & Telecom Press, Joshua F. Wiley
深度学习:核心原理与案例分析

深度学习:核心原理与案例分析

Posts & Telecom Press, Ahmed Menshawy

Publisher Resources

ISBN: 9798341657755