- Hands-On Mathematics for Deep Learning
- Jay Dawani
- 130字
- 2021-06-18 18:55:19
Chain rule
Let's take an arbitrary function f that takes variables x and y as input, and there is some change in either variable so that . Using this, we can find the change in f using the following:

This leads us to the following equation:

Then, by taking the limit of the function as , we can derive the chain rule for partial derivatives.
We express this as follows:

We now divide this equation by an additional small quantity (t) on which x and y are dependent, to find the gradient along . The preceding equation then becomes this one:

The differentiation rules that we came across earlier still apply here and can be extended to the multivariable case.
推薦閱讀
- 同步:秩序如何從混沌中涌現(xiàn)
- 數(shù)據(jù)浪潮
- Test-Driven Development with Mockito
- Hadoop大數(shù)據(jù)實(shí)戰(zhàn)權(quán)威指南(第2版)
- Creating Dynamic UIs with Android Fragments(Second Edition)
- Python醫(yī)學(xué)數(shù)據(jù)分析入門
- Sybase數(shù)據(jù)庫在UNIX、Windows上的實(shí)施和管理
- WS-BPEL 2.0 Beginner's Guide
- 智能數(shù)據(jù)時(shí)代:企業(yè)大數(shù)據(jù)戰(zhàn)略與實(shí)戰(zhàn)
- 數(shù)據(jù)科學(xué)實(shí)戰(zhàn)指南
- Filecoin原理與實(shí)現(xiàn)
- Deep Learning with R for Beginners
- 企業(yè)大數(shù)據(jù)處理:Spark、Druid、Flume與Kafka應(yīng)用實(shí)踐
- AndEngine for Android Game Development Cookbook
- 數(shù)據(jù)應(yīng)用工程:方法論與實(shí)踐