16.txt

This complete matlab for neural network

发信人: GzLi (笑梨), 信区: DataMining
标  题: 推荐一篇文章
发信站: 南京大学小百合站 (Sat Dec  7 11:01:32 2002), 站内信件


Breiman, L. (2001). Statistical Modeling: The two cultures. Statistical Scien
ce, 16, No 3, 199-231.
我没有找到电子版，愿有大牛能出之共享。网上的确没有，有的网站已经证实了not onl
ine。

这篇文章在清华有人推荐by
**********************************************************
发信人: cczhu (葱葱猪), 信区: AI       
标  题: 推荐一篇论文
发信站: BBS 水木清华站 (Mon Nov 25 20:21:59 2002)

虽然不是很新，而且都是讨论已有的方法和算法，但感觉写得很好。
不愧是牛人写的东西，思路清晰，说理充分，把几种机器学习的模型（NN,SVM等）和统计
模型融合在一起对比分析。刚熬夜看完，觉得值:)
唯一不足是好像没找到电子版。

Statistical Modeling: The Two Cultures
Leo Breiman(Berkeley)
Statistical Science 2001 Vol.16 No.3 199-231.
*******************************************************************
我在网上搜了一下，不少这方面的workshop。很象很牛阿。

Quantitative Studies Group (QSG) 
Lecture Series  
Spring Semester '02 Meetings:
Every Thursday, 3:30-4:45 pm in Haggar 128. 
Discussion of (Breiman (2001) Statistical Modeling: The Two Cultures 

There are two cultures in the use of statistical modeling to reach conclusion
s from data. One assumes tht the data are generated by a given stochastic dat
a model. The other uses algorithmic models and treats the data mechanism as u
nknown. The 
statistical community has led to irrelevant theory, questionable conclusions,
 and has kept statisticians from working on a large range of interesting curr
ent problems. Algorithmic modeling, both in theory and practice, has develope
d rapidly in fields 
outside statistics. It can be used both on large complex data sets and as a m
ore accurate and informative alternative to data modeling on smaller data set
s. If our goal as a field is to use data to solve problems, then we need to m
ove 
away from exclusive dependence on data models and adopt a more diverse set of
 tools. 

Pattern Recognition and Prediction
Machine Learning and Data Mining
ONE DAY WORKSHOP
SUNDAY, JULY 22, 2001
Keynote Speaker
LEO BREIMAN
Emeritus Professor of Statistics
University of California at Berkeley
P.R. Krishnaiah Visiting Scholar: 2001
Leo Breiman received his Ph.D. in mathematics in 1950 from the University of 
California, Berkeley. He is a member of the National Academy of Sciences and 
Fellow of the American Statistical Association and the Institute of Mathemati
cal Statistics. He 
received Youden Prize, Technometrics, in 1992. He is the author of the textbo
oks, Probability and Stochastic Processes with a View Toward Applications, St
atistics with a View Toward Applications, Probability, and co-author of Class
ification and 
Regression Trees.
Breiman is one of the leading statisticians in the world today. He made funda
mental contributions to stochastic processes, information theory and mathemat
ical statistics. A distinguished career as a statistical consultant, in traff
ic flow analysis, air 
pollution analysis, and computerized speech recognition led him to his develo
pment of the widely used algorithm for classification and prediction CART wit
h Friedman, Olsen and Stone. More recently he made another original discovery
, "bagging", a way of 
improving "supervised learning" by combining noisy predictions by resampling 
the original data. His current interests are in computationally intensive mul
tivariate analysis including the use of nonlinear methods for pattern recogni
tion and prediction 
in high dimensional spaces.

Abstracts of talks by Leo Breiman
STATISTICAL MODELING – THE TWO CULTURES
Suppose that the data consists of dependent variables y and a number of predi
ctor variables x. The dominant approach to modeling in statistics is data mod
eling – the assumption is made that the data are generated from a known para
metric family 
containing a stochastic element. That is, that the y’s are generated as a sp
ecified function of the x’s, parameters and noise variables. A small minorit
y in statistics and many in fields outside of statistics use algorithmic mode
ling (more loosely 
called data mining). This approach makes no assumptions about how the data ar
e generated. Instead algorithms (f(x)) are constructed so as the predict the 
y’s as accurately as possible from the x’s. In my talk, I will discuss the 
advantages and 
disadvantages of these two approaches.

  --               ***  端庄厚重 谦卑含容 事有归着 心存济物  *** 数据挖掘  ht
tp://[e-mail]/ 

--
※ 修改:．GzLi 于 Dec  7 11:08:17 修改本文．[FROM: 211.80.38.17]
※ 来源:．南京大学小百合站 bbs.nju.edu.cn．[FROM: 211.80.38.17]