Machine learning reading notes 1

1 Introduction

Hello there! This is the first time you use the welcome page displayed by the Markdown editor . If you want to learn how to use the Markdown editor, you can read this article carefully to learn about the basic syntax of Markdown.

1.1 Basic terminology

A data set is a collection X of a set of records, and each record is a description of an object.
For each sample xi, we describe its d attributes (features), so a d-dimensional sample space is generated, and each sample corresponds to a d-dimensional vector.
In addition to d attributes (features), each object also needs a label (yi), which is the classification category of the object. The set Y of yi is all possible categories, called the output space.
Y is a discrete value, corresponding to the "classification" task; Y is a continuous value, corresponding to the "regression" task.
The process of obtaining a model through data learning is called training, and the result is a mapping from the input set X to the output space Y.

Supervised learning: the training set has label information, such as classification and regression;
unsupervised learning: the training set has no label information, such as clustering.
The ability of the learned model to apply to new samples is called "generalization" ability.

1.2 Hypothetical space

The set of possible value combinations for each attribute of the sample has (num1+1) (num2+1) ……*(numd+1)+1 hypotheses.
The result of training is to find a "hypothesis set" (a subset of the hypothesis space) that is consistent with the training set.