© 2026 LIN TIANYU. All rights reserved. Preview Version.

Powered bynextjs-blog

GitHub LinkedIn

ArticlePDF机器学习

PAC 学习：无限假设类（PAC Learning）

PublishedJuly 24, 2026Document7 pages

Contents

从有限类到无限类：为什么要换思路 (From Finite to Infinite)
轴平行矩形的 PAC 学习 (PAC-Learning Axis-Parallel Rectangles)
问题设定 (Problem Setup)
算法：最紧拟合矩形 (The Tightest-Fitting Rectangle)
坏事件与四条带 (Bad Events & the Four Strips)
样本复杂度的推导 (Deriving the Sample Complexity)
半空间的 PAC 学习 (PAC-Learning Half-Spaces)
半空间是什么 (What Is a Half-Space)
从样本到线性不等式系统 (From Examples to Linear Inequalities)
用线性规划求解 & 通向感知机 (Linear Programming & Toward Perceptron)

1

2

3

4

5

6

7

Thanks for reading.