adaboost的过程要点理解
2016-02-15 11:53
477 查看
1. 确定弱分类器的个数m;
2. 因设定共m个分类器,所以分m个阶段,每个阶段训练一个弱分类器;
3. 训练一个弱分类器的本质就是对相应的分类函数不断的调整其参数使得其对所有样本(如N个样本)进行分类的误差最小化;
4. 最后m个阶段的分类器被训练出来之后,按权重进行加和,合成一个强分类;
2. 因设定共m个分类器,所以分m个阶段,每个阶段训练一个弱分类器;
3. 训练一个弱分类器的本质就是对相应的分类函数不断的调整其参数使得其对所有样本(如N个样本)进行分类的误差最小化;
4. 最后m个阶段的分类器被训练出来之后,按权重进行加和,合成一个强分类;
相关文章推荐
- Union-find
- [LeetCode] Unique Binary Search Trees
- [LeetCode] Unique Binary Search Trees II
- [LeetCode] Sum Root to Leaf Numbers
- [LeetCode] Single Number
- [LeetCode] Single Number II
- [LeetCode] Maximum Subarray
- [LeetCode] Remove Element
- [LeetCode] Remove Duplicates from Sorted List
- [LeetCode] Remove Duplicates from Sorted List II
- [LeetCode] Remove Duplicates from Sorted Array II
- [LeetCode] Remove Duplicates from Sorted Array
- [LeetCode] Climbing Stairs
- [LeetCode] Pascal's Triangle
- [LeetCode] Pascal's Triangle II
- [LeetCode] Plus One
- [LeetCode] Linked List Cycle
- [LeetCode] Binary Tree Preorder Traversal
- [LeetCode] Binary Tree Postorder Traversal
- [LeetCode] Binary Tree Inorder Traversa