There are two common unsupervised feature learning settings, depending on what type of unlabeled data you have. The more general and powerful setting is the self-taught learning setting, which does not assume that your unlabeled data xuxu has to be drawn from the same distribution as your labeled data xlxl. The more restrictive setting where the unlabeled data comes from exactly the same distribution as the labeled data is sometimes called the semi-supervised learning setting. This distinctions is best explained with an example, which we now give.
self taught learning和semi-supervised learning 最大的区别在于其特征分布是否假设相同。self taught learning相比于semi-supervised learning 更强大一些，因为可以假设unlabeled data和labeled data的分布不相同。而semi-supervised learning则需要相同。
然后作者举了一个例子就是对于car 和 motorcycle的例子。假设想要去区分汽车和摩托车，如果还有一对其他图片，比如说阿猫阿狗，杂七杂八的混在一起，最好使用self taught learning，但是如果unlabeled data只有汽车和摩托车，那么就可以使用unsupervised learning.
