以对象管理资源 当我们向系统申请某一笔资源后, 应该立刻在同一语句内用它来初始化某个管理对象. 如果有下面这个工厂函数用来动态分配一个资源, 并返回指向资源的指针: Resource* create_resource(); 大部分时候我们可以通过智能指针很好的管理资源, 例如下面的代码: std::shared_ptr<Resource> ptr(create_reso...
构造-析构-赋值
了解 C++ 默默编写并调用哪些函数 C++ 能够自动生成默认的构造函数, 析构函数, 复制构造函数和赋值运算符. 这些函数都被声明为 public 和 inline, 且仅当它们需要被调用时才会被生成. 示例代码如下: Empty e1; // 生成默认构造函数 Empty e2(e1); // 生成默认复制构造函数 Empty e3 = e2; // 生成默认复制构...
从 C 到 C++
enum hack 在 C++ 中, 大部分宏定义常量可以通过 const 关键字替换, 一个例外是在 class 编译期间需要使用一个 class 常量, 这种情况可以利用 enum 类型来实现替换. 其理论基础在于, 一个属于枚举类型的数值可以被视为 int 型来使用. 例如下面的类定义: class GamePlayer { private: enum { num_turn...
Python Mathematic Modules
numpy Import It is common to import numpy under a briefer name np and also import some class frequently used like array. Common import as follows: import numpy as np from np import array Arra...
Radial Basic Function Network
RBF Network Learning Firstly we consider the Gaussian SVM which map our data to a infinite-dimensional space [g_\text{SVM}(\boldsymbol x) = \text{sign}\left(\sum_\text{SV}\alpha_ny_n\exp(-\gamma\...
Neural Network
Neural Network is a classic algorithm originated from perceptron. Linearly combine various perceptrons such as the figure the final hypothesis $G$ is [G(\boldsymbol x)=\text{sign}\left(\sum_{t=...
Adaptive Boosting
AdaBoost is an algorithm that promotes weak learners to a strong learner. This algorithm first trains a base learner from the initial training set, and then adjusts the sample distribution accordin...
Support Vector Machine 2
Soft-Margin SVM In order to avoid the effect of noise in dataset, we allow the model to make a small number of mistakes, similar to PLA, we have a new optimization goal shaped like [\begin{split}...
Python基础
正则表达式 在Python中使用正则表达式 用 import re 导入正则表达式模块. 用 re.compile() 函数创建一个 Regex 对象 (使用原始字符串 r'') . 调用 Regex 对象的 search() 方法传入想要查找的字符串, 并返回一个 Match 对象. 调用 Match 对象的 group() 方法, 返回实际匹配文本的字符串. i...
Support Vector Machine 1
Linear SVM Large-Margin Problem Firstly, we consider 3 linear classifier on the same dataset as follows: both of them seem performing well but the rightmost one whose hyperplane is farthest fr...