Communication efficient construction of decision trees over heterogeneously distributed data

被引：14

作者：

Giannella, C ^{[1
]}

Liu, K ^{[1
]}

Olsen, T ^{[1
]}

Kargupta, H ^{[1
]}

机构：

[1] Univ Maryland Baltimore Cty, Dept Comp Sci & Elect Engn, Baltimore, MD 21250 USA

来源：

FOURTH IEEE INTERNATIONAL CONFERENCE ON DATA MINING, PROCEEDINGS | 2004年

关键词：

decision trees; distributed data mining; random projection;

D O I：

10.1109/ICDM.2004.10114

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

We present an algorithm designed to efficiently construct a decision tree over heterogeneously distributed data without centralizing. We compare our algorithm against a standard centralized decision tree implementation in terms of accuracy as well as the communication complexity. Our experimental results show that by using only 20% of the communication cost necessary to centralize the data we can achieve trees with accuracy at least 80% of the trees produced by the centralized version.

引用

页码：67 / 74

页数：8

共 12 条

[1]

ARRIAGA RI, 1999, P 40 FDN COMP SCI NE

[2]

CARAGEA D, 2003, P C INT SYST DES APP

[3] An experimental comparison of three methods for constructing ensembles of decision trees: Bagging, boosting, and randomization [J].

Dietterich, TG .

MACHINE LEARNING, 2000, 40 (02) :139-157

[4]

Du W, 2002, PROC IEEE INT C PRIV, V14, P1

[5]

HALL LO, 2000, LECT NOTES COMPUTER, V1759

[6]

Hecht-Nielsen R., 1994, Computational Intelligence: Imitating Life, P43

[7]

KARGUPTA H, 2004, 2004002 LLC, P1450

[8]

KARGUPTA H, 2004, DATA MINING NEXT GEN

[9]

Maclin R., 1999, J ARTIF INTELL RES, V11, P169, DOI [DOI 10.1613/JAIR.614, 10.1613/jair.614]

[10]

PARK B, 2002, APPL INTELLIGEN 0116

← 1 2 →