484.txt

This complete matlab for neural network

发信人: yaomc (白头翁&山东大汉), 信区: DataMining
标  题: [合集]关于DM与AI关系的调查
发信站: 南京大学小百合站 (Fri Jan 18 13:16:46 2002), 站内信件

daniel (飞翔鸟) 于Sun Jan  6 21:55:15 2002提到：

   In general, there are two kinds of statements for the relationship 

between Data mining & Artificial Intelligence.

   The first is: DM is a branch of AI because it is initialized from AI 

(comment 1), and most techniques of DM came from another branch of AI, i.e. 

Machine learning. (comment 2 and 3) 

  (comment 1: this is almost the truth. at least, the first KDD workshop
was held with IJCAI-89)

  (comment 2: although many machine learning researchers now regard ML
as a seperate discipline parallel to AI, here we still adopt the recognition
of traditional AI researchers, that is, ML is a branch of AI.)

  (comment 3: although statistical techniques play increasingly important
roles in data mining, in fact statistics was introduced into DM by Machine
learning instead of statistics itself, and most techniques implemented in 
commercial DM softwares are from ML. this is the truth at least at the end
of 2001.)

  The second is: DM is a seperate discipline parallel to AI because although

DM encompasses many techniques from AI, and although DM is very related to

AI, there are significant differences between them so that they form different

disciplines. (comment 4)

  (comment 4: this statement almost requires little comments if not none, 
that is, at least subdomains such as parallel data mining involves other 
disciplines such as parallel data processing as well as AI.)

  From my point of view, both statements have their strength and weakness.

I have also discussed such a problem with some famous data mining researchers.

But I found their opinions are very diversified. 

  Now I hope to know the opinions of the BBSers. You can express your 

agreement on those two statements, also you can express your own opinion.

  I also hope our board manager help do this survey because I know he is very 

active on several other BBS, and I do not have enough time to initialize the 

survey and stat the results.

  Thank you!


yeli (旗手) 于Sun Jan  6 22:25:04 2002)
提到：

I support the first opinion.







explorer (void) 于Sun Jan  6 22:26:03 2002提到：

sp 2


ASSP (SAP) 于Sun Jan  6 22:58:05 2002提到：

Personally, I prefer that DM is a branch of AI rather than it is 
parallel to AI.



yaomc (白头翁&山东大汉) 于Mon Jan  7 10:31:06 2002提到：

First of all, we all appreciate daniel giving us such valuable topic 
to discuss.
Second, I do think DM neither is a branch of AI nor is parllel to AI.
The reason is that though some of the method and technology of DM are coming
from AI, the content of DM really pay more attention to its application, DM i
s only 
an tool to help us to discover the knowledge hidden the large amount of data 
whcih belongs to one specially field,and it used the opinions of database 
field, especially important is DM must connect the domain knowledge which 
you want to mine. Only has the tools can not help us to mine knowledge.
That is only my some of opinions.



yuns (大愚若智) 于Mon Jan  7 15:11:56 2002提到：

由于现在时间比较忙，偷懒用中文来发表意见。

首先说先入为主的观念，由于本人并非AI出身，所以当我第一次接触到DM的时候
就自然将其作为一种独立的学科分支看待，而象daniel这样从AI或者ML出身的研
究者，如果倾向于将DM视为AI的分支有起自然的心理来源。

某种程度上说，yaomc的观点是有其合理性，由于AI更多的是从理论上对很多问题
进行思考，而DM开始应该就是由实际应用需要推动的，所以这两件事情很难从一个
共同的层面上来考虑。或者说DM更象一个边缘学科，边缘的特质使其既不能简单的
归为哪个的branch，也不能简单说与谁parellel。

我认为学科的分类（甚至其他很多东西，比如民族的区分等等），就象一个B树的分
裂。一开始是一个单一的学科，比如计算机，但是当它的研究内容逐渐增多，这个
节点就可以分裂成多个分支，比如现在的AI应该就是一个独立分支，而ML的研究由于
也比较多，可以比较倾向于认为ML应该分裂成一个独立分支，而DM由于是一个比较新
兴地研究方向，所以我比较倾向于认为它尚没有完全分裂出来，如果这个研究方向
可以继续得到普遍的关注，那时候它就可以成为一个完全独立的研究方向了。或许就
是由于DM目前处于这种即将可以分裂成独立节点附近的临界状态，才会产生这样DM到
底是AI的branch还是parellel的争论。我想十年前这种争论是不存在的，或许十年后
同样也不需要这种争论了。



yaomc (白头翁&山东大汉) 于Mon Jan  7 15:28:17 2002提到：

有道理。
作为一个迅速发展的多学科交叉的边缘学科来说，说他完全属于哪一类确实是比较困难
的，现在有人说DM更多的讲的是应用（当然对于我们这些与IT相差比较远的来说），我
认为他的理论方面的东西相对还不是很成熟。就象是现在关于DM的最基本的概念来说
都是存在一个争议的问题，不同的研究者从不同的角度给与了不同的定义，因此也正因为
这种不确定性，才使得DM能够迅速的应用各个领域中去，或许很多领域的学者都认为
DM是属于本领域的，由此也造成了DM的欣欣向荣。
说句比较中庸的话：DM是属于我们大家的！



armen (理性疯狂◎伢子阿蒙◎准备考试) 于Mon Jan  7 16:50:20 2002提到：

版主开投票吧



yuns (大愚若智) 于Mon Jan  7 17:04:52 2002提到：

为什么要投票呢？
学术讨论并不是一个少数服从多数的问题呀


yaomc (白头翁&山东大汉) 于Mon Jan  7 17:08:07 2002提到：

我觉得投票的范围总是比较小，大家先尽情的发挥自己的想法，这样的结果可能
更具有普遍意义。也能够从不同的侧面来讨论DM的现在以及将来。



yaomc (白头翁&山东大汉) 于Mon Jan  7 17:08:58 2002提到：

那就请大家多多发表自己的见解。OK？



armen (理性疯狂◎伢子阿蒙◎准备考试) 于Mon Jan  7 17:19:56 2002提到：

我觉得投票不是要谁服从谁，是了解一下情况嘛

关键是版主要讲投票做的好一点才行



yaomc (白头翁&山东大汉) 于Mon Jan  7 19:21:38 2002提到：

好一点啊？
现在好像还没有看到更多更好的选择啊？
况且大家只是选择其中之一，可是每个人的理解好像也会是不同的，即使选择的是同一个
选项。请多提点更多的想法？



GzLi (笑梨) 于Tue Jan  8 00:06:00 2002提到：

I consider DM as a mixture of Database and AI
In DM of HanJiawei, he writes his book in a view from database.And we can fi
nd that some mining tools are from machine learning, so I have the opnion th
at DM is a mixture.
My major is DM, I do some research on classiers such as SVM and ANN.




adson (青年抗日同盟会) 于Tue Jan  8 22:46:23 2002)
提到：

I agree with you.








waterflower (拥抱阳光) 于Wed Jan  9 00:47:24 2002提到：

I agree to the opinion that DM is a branch of AI, first, the method in 
DM almost stemmed from that of ML like what the comments told. second,
but not the least, their ultimate goal both focus on automation of process.

I think their relation is similiar to the difference between 
statistics and maths. To solve some practical problems, we used
some methods of maths, and at last classified these methods 
into statistics which has identical characters to other methods.
And I am sure none will object that statistics is a branch of 
maths. Trying this angle, I agree with me.

P.S:it seems that we need some knowledge of philosophy.


chouqin (chouqin) 于Wed Jan  9 09:07:30 2002提到：

I prefer opinion 2.



aniky (童话中的红房子) 于Wed Jan  9 13:22:46 2002)
提到：

搞了张图看看。







yuns (大愚若智) 于Wed Jan  9 14:27:23 2002提到：

其实严格的说统计学和数学并不是一个学科，
在国外通常这也是两个完全独立的专业，
我虽然对数学并不太懂，但是由于我今年住在一个数学系的宿舍里面，
正好整天都在听我下铺的那外唠叨统计学不是数学，
所以我并不同意统计学是数学的一个分支的说法。

lakecloud (天高云淡) 于Thu Jan 10 17:17:08 2002提到：

从目前的DM方法来看, 显然是机器学习, 数据库, 统计学
的结合. DM的叫法是从机器学习扩展而来的, 但现已包括
很多统计学习方法.
然而, 在此之前,统计学也一直在做类似的东西, 只不过没
叫这个名字而已. 现在是双方互相"发现"的过程.