Machine learning (zhouzhihua) Chapter 1 Introduction notes learning experience

The first 1 Chapter The introduction The learning

Because I have learned from teacher Li Hang before 《 Statistical learning method （ The first 2 edition ）》, So there is nothing in this concept that I don't understand , It won't be as difficult as the tutorial says , After all, I have learned some . and , This book is indeed more extensive than teacher Li Hang , It can be seen from the introduction .

Among them 1.4 Generalize preferences It can be said that it has expanded my knowledge , I just thought , There's a hypothesis （ Model ）, With the loss function , To optimize the minimum loss is enough , But in fact, I have been ignoring a more important , Something higher , Suppose the final model learned in the space , In fact, he has his own “ Preference ” Of , It's just a model that happens to have some preferences , We can get better training and standardization results . especially , It's mentioned here NFL（No Free Lunch） Theorem , It's very vivid , That is, no model is smarter , He's just smarter about a specific need , From another measurement dimension , There must be something stupid about him , We must look at specific practical problems , This kind of cleverness and dullness makes sense .

Besides , It is also helpful to annotate some contents of the book in an annotated way .