为什么叫“机器学习”Machine Learning 而不是“计算机学习”——深度学习Note

有一门学科“机器学习”火了起来，它是计算机科学与数学结合的产物，它的目的是使计算机“聪明”起来，实现人工智能。可是，令人困惑的是它明明就是计算机学习，为什么不叫“计算机学习”而叫“机器学习”呢？这要从英语“计算机”computer这个单词说起。

熟悉英语构词法的人都知道，后缀-er 指人，那么最初computer 的意思是“计算者”，是做计算工作的人。当图灵第一次构想计算机的时候，他当然不会用computer 这个词，他用的是computing machines——一个很自然的表达。后人把他的构想称为图灵机Turing Machine。到后来人们真正造出计算机来以后就借用了computer 这个单词并广为使用，所以machine 才是计算机的“原名”。

但是，在偏理论的“计算理论”、“计算复杂度”等方向，machine这一词还是常用，它的意思就是计算机，而不是任意的机器如洗衣机、挖土机等等。

所以，叫Machine Learing（翻译为“机器学习”）并不影响计算机学习的本意，而且显得典雅，也应该是为了向图灵致敬。

现代机器学习主要包括统计学习和深度学习，其中深度学习大放异彩，正开启人工智能时代。

（说明，由于本博客是计算机英语写作，后续将把本文翻译成英语供交流。）

A discipline named machine learning has attracted particular attention. It’s a combination of computer science and mathematics, and aims to render computers intelligent and achieve artificial intelligence. However, what is confusing is why a discipline of computer learning is called machine learning. Let’s try to find it out.

Those who are familiar with the construction rules of English words know that the suffix -er represents a person. Originally, the word “computer”meant a person who did the job of computation. When Turning first conceived “computers”, of course he wouldn’t use the word “computer”. Turning used a rather natural expression “computing machines”. Turning's conception is well known as Turning Machine. Subsequently, as real “computers” were created, the word “computer” was adopted and accepted widely. So “machine” is the original name of computers.

Yet, in theoretical disciplines like theory of computation and computational complexity, the word “machine” is still often used. The name machine learning seems more classical than “computer learning”, doesn't it?  Also, It’s supposed to pay a tribute to Turning. 

Modern machine learning includes statistical learning and deep learning. The latter is just booming at present.