欧美日韩第一区,中日韩高清电影网,亚洲一区二区高清

stander random forest and ensemble of Extremely randomized trees

bigrabbit — Sun, 28 Feb 2016 13:01:00 GMT

stander random forest: random K features, enum all values as split, find best split.

LINKS:https://en.wikipedia.org/wiki/Random_forest

Extremely randomized trees: random K features, random a split value, find best split.

ensemble Extremely randomized trees: use all data.

LINKS:http://docs.opencv.org/2.4/modules/ml/doc/ertrees.html

Extremely randomized trees don’t apply the bagging procedure to construct a set of the training samples for each tree. The same input training set is used to train all trees.
Extremely randomized trees pick a node split very extremely (both a variable index and variable splitting value are chosen randomly), whereas Random Forest finds the best split (optimal one by variable index and variable splitting value) among random subset of variables.

Extremely randomized trees用了所有的��h��作�ؓ训练集；Extremely randomized trees随机选一个特征和一个��g��为分割标准；

LINKS:http://scikit-learn.org/stable/modules/generated/sklearn.tree.ExtraTreeRegressor.html#sklearn.tree.ExtraTreeRegressor

This class implements a meta estimator that fits a number of randomized decision trees (a.k.a. extra-trees) on various sub-samples of the dataset and use averaging to improve the predictive accuracy and control over-fitting.

Extra-trees differ from classic decision trees in the way they are built. When looking for the best split to separate the samples of a node into two groups, random splits are drawn for each of the max_features randomly selected features and the best split among those is chosen. When max_features is set 1, this amounts to building a totally random decision tree.

extra-trees 的ensemble用了bagging�Q�然后选取多个特征�Q�每个特征随机选一个��g��为分割标准徏树�?br />
一�U�实现方法：
��h��bagging, random n features & random k values �Q�求最优，建树�?nbsp;

bigrabbit 2016-02-28 21:01 发表评论

cocos2d-iphone基础(1)

bigrabbit — Thu, 15 May 2014 13:14:00 GMT

主要�c�：
CCNode

   CCDirector
   CCScene
   CCLayer

定时更新�Q?br />

[[[CCDirector sharedDirector] scheduler] scheduleUpdateForTarget:self priority:0 paused:NO];

//[[[CCDirector sharedDirector] scheduler] unscheduleUpdateForTarget:self];

接收输入�Q?br /> v0.99

CCStandardTouchDelegate

CCTargetedTouchDelegate

v2.10

CCTouchOneByOneDelegate

CCTouchAllAtOnceDelegate

[[[CCDirector sharedDirector] touchDispatcher] addTargetedDelegate:self priority:0 swallowsTouches:YES];

//[[[CCDirector sharedDirector] touchDispatcher] removeDelegate:self];

坐标�pȝ��Q?br /> position是设�|�相对于父亲节点的坐�?br /> self.anchorPoint和self.position重合

多层�Q?br /> [cclayer.addchild cclayer];
一层叠一�?/p>

bigrabbit 2014-05-15 21:14 发表评论

bigrabbit — Wed, 24 Oct 2012 14:47:00 GMT

摘要: 阅读全文

bigrabbit 2012-10-24 22:47 发表评论

bigrabbit — Thu, 02 Aug 2012 07:36:00 GMT

摘要: 阅读全文

bigrabbit 2012-08-02 15:36 发表评论

backup之关于图的双联通，割点�Q�桥�Q�有向图��通等

bigrabbit — Tue, 31 Jul 2012 14:36:00 GMT

摘要: 阅读全文

bigrabbit 2012-07-31 22:36 发表评论

bigrabbit — Thu, 26 Jul 2012 04:14:00 GMT

摘要: 阅读全文

bigrabbit 2012-07-26 12:14 发表评论

UVA 10801 Lift Hopping

bigrabbit — Sun, 22 Jul 2012 15:43:00 GMT

摘要: UVA 10801 Lift Hopping 阅读全文

bigrabbit 2012-07-22 23:43 发表评论

uva10557 囄��杂题

bigrabbit — Fri, 13 Jul 2012 01:02:00 GMT

摘要: 阅读全文

bigrabbit 2012-07-13 09:02 发表评论

武大校赛�ȝ��

bigrabbit — Mon, 30 Apr 2012 08:30:00 GMT

摘要: 阅读全文

bigrabbit 2012-04-30 16:30 发表评论

bigrabbit — Sun, 22 Apr 2012 09:23:00 GMT

今天做了�ơCF�Q�两个小时比赛时间。。用一��时水了两题之后�Q�又用一个小时的龟速想了一个不知道什么玩意的玩意�Q�比赛没A掉，比赛�l�束�Q�掉了。�ؓ什么要想那么久呢。。。。。。水题也要想那么久。。。。。小�l�节处理不好。。。。�?br />不过话说CF的题目不错，不像�q�大中文OJ的无脑题
http://codeforces.com/problemset/problem/180/E
不脓(chu��ng)代码了�?br />

bigrabbit 2012-04-22 17:23 发表评论

欧美日韩第一区,中日韩高清电影网,亚洲一区二区高清

stander random forest and ensemble of Extremely randomized trees

cocos2d-iphone基础(1)

backup之关于图的双联通，割点�Q�桥�Q�有向图�����通等

UVA 10801 Lift Hopping

uva10557 囄���杂题

武大校赛�ȝ��

backup之关于图的双联通，割点�Q�桥�Q�有向图��通等

uva10557 囄��杂题