Effects of Dynamic Subspacing in Random Forest

Md Nasim Adnan; Md Zahidul Islam

doi:10.1007/978-3-319-69179-4_21

Effects of Dynamic Subspacing in Random Forest

Md Nasim Adnan, Md Zahidul Islam

Research output: Book chapter/Published conference paper › Conference paper › peer-review

5 Citations (Scopus)

Abstract

Due to its simplicity and good performance, Random Forest attains much interest from the research community. The splitting attribute at each node of a decision tree for Random Forest is determined from a predeﬁned number of randomly selected attributes (a subset of the entire attribute set). The size of an attribute subset (subspace) is one of the most important factors that stems multitude of inﬂuences over Random Forest. In this paper, we propose a new technique that dynamically determines the size of subspaces based on the relative size of the current data segment to the entire data set. In order to assess the eﬀects of the proposed technique, we conduct experiments involving ﬁve widely used data set from the UCI Machine Learning Repository. The experimental results indicate the capability of the proposed technique on improving the ensemble accuracy of Random Forest.

Original language	English
Title of host publication	Proceedings of the 13th International Conference of Advanced Data Mining and Applications (ADMA 2017)
Editors	Gao Cong, Wen-Chih Peng, Wei Emma Zhang, Chengliang Li, Aixin Sun
Publisher	Springer
Pages	303-312
Number of pages	10
ISBN (Electronic)	9783319691794
ISBN (Print)	9783319691787
DOIs	https://doi.org/10.1007/978-3-319-69179-4_21
Publication status	Published - 2017
Event	The 13th International Conference on Advanced Data Mining and Applications (ADMA): ADMA 2017 - Nanyang Technological University Alumni House, Singapore, Singapore Duration: 05 Nov 2017 → 06 Nov 2017 http://www.adma2017.net/#0 (Conference website)

Publication series

Name	Edit Lecture Notes in Artificial Intelligence
Publisher	Springer
Volume	10604
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	The 13th International Conference on Advanced Data Mining and Applications (ADMA)
Country/Territory	Singapore
City	Singapore
Period	05/11/17 → 06/11/17
Other	The year 2017 marks the 13th anniversary of the International Conference on Advanced Data Mining and Applications (ADMA 2017), which will be held in Singapore, 5—6 Nov 2016, co-located with ACM CIKM2017. It is our great pleasure to invite you to contribute papers and participate in this premier annual event on research and applications of data mining. A growing attention has been paid to the study, development, and application of data mining. As a result, there is an urgent need for sophisticated techniques and tools that can handle new subfields of data mining, e.g., smartphone and social network data mining, spatial data mining, streaming data mining, green computing data mining, biomedical data mining, the Internet of Things, and data mining for healthcare. Our expertise in data mining also has to be expanded to new applications.
Internet address	http://www.adma2017.net/#0 (Conference website)

Access to Document

10.1007/978-3-319-69179-4_21

Cite this

Adnan, M. N., & Islam, M. Z. (2017). Effects of Dynamic Subspacing in Random Forest. In G. Cong, W.-C. Peng, W. E. Zhang, C. Li, & A. Sun (Eds.), Proceedings of the 13th International Conference of Advanced Data Mining and Applications (ADMA 2017) (pp. 303-312). (Edit Lecture Notes in Artificial Intelligence; Vol. 10604). Springer. https://doi.org/10.1007/978-3-319-69179-4_21

@inproceedings{3e02ad2fb22f48c1bb7eebe584ff0ff7,

title = "Effects of Dynamic Subspacing in Random Forest",

abstract = "Due to its simplicity and good performance, Random Forest attains much interest from the research community. The splitting attribute at each node of a decision tree for Random Forest is determined from a predeﬁned number of randomly selected attributes (a subset of the entire attribute set). The size of an attribute subset (subspace) is one of the most important factors that stems multitude of inﬂuences over Random Forest. In this paper, we propose a new technique that dynamically determines the size of subspaces based on the relative size of the current data segment to the entire data set. In order to assess the eﬀects of the proposed technique, we conduct experiments involving ﬁve widely used data set from the UCI Machine Learning Repository. The experimental results indicate the capability of the proposed technique on improving the ensemble accuracy of Random Forest.",

keywords = "Decision tree, Decision forest, Random forest",

author = "Adnan, {Md Nasim} and Islam, {Md Zahidul}",

year = "2017",

doi = "10.1007/978-3-319-69179-4_21",

language = "English",

isbn = "9783319691787",

series = "Edit Lecture Notes in Artificial Intelligence",

publisher = "Springer",

pages = "303--312",

editor = "Gao Cong and Wen-Chih Peng and Zhang, {Wei Emma} and Chengliang Li and Aixin Sun",

booktitle = "Proceedings of the 13th International Conference of Advanced Data Mining and Applications (ADMA 2017)",

address = "United States",

note = "The 13th International Conference on Advanced Data Mining and Applications (ADMA) : ADMA 2017 ; Conference date: 05-11-2017 Through 06-11-2017",

url = "http://www.adma2017.net/#0",

}

Adnan, MN & Islam, MZ 2017, Effects of Dynamic Subspacing in Random Forest. in G Cong, W-C Peng, WE Zhang, C Li & A Sun (eds), Proceedings of the 13th International Conference of Advanced Data Mining and Applications (ADMA 2017). Edit Lecture Notes in Artificial Intelligence, vol. 10604, Springer, pp. 303-312, The 13th International Conference on Advanced Data Mining and Applications (ADMA), Singapore, Singapore, 05/11/17. https://doi.org/10.1007/978-3-319-69179-4_21

Effects of Dynamic Subspacing in Random Forest. / Adnan, Md Nasim; Islam, Md Zahidul.
Proceedings of the 13th International Conference of Advanced Data Mining and Applications (ADMA 2017). ed. / Gao Cong; Wen-Chih Peng; Wei Emma Zhang; Chengliang Li; Aixin Sun. Springer, 2017. p. 303-312 (Edit Lecture Notes in Artificial Intelligence; Vol. 10604).

Research output: Book chapter/Published conference paper › Conference paper › peer-review

TY - GEN

T1 - Effects of Dynamic Subspacing in Random Forest

AU - Adnan, Md Nasim

AU - Islam, Md Zahidul

PY - 2017

Y1 - 2017

N2 - Due to its simplicity and good performance, Random Forest attains much interest from the research community. The splitting attribute at each node of a decision tree for Random Forest is determined from a predeﬁned number of randomly selected attributes (a subset of the entire attribute set). The size of an attribute subset (subspace) is one of the most important factors that stems multitude of inﬂuences over Random Forest. In this paper, we propose a new technique that dynamically determines the size of subspaces based on the relative size of the current data segment to the entire data set. In order to assess the eﬀects of the proposed technique, we conduct experiments involving ﬁve widely used data set from the UCI Machine Learning Repository. The experimental results indicate the capability of the proposed technique on improving the ensemble accuracy of Random Forest.

AB - Due to its simplicity and good performance, Random Forest attains much interest from the research community. The splitting attribute at each node of a decision tree for Random Forest is determined from a predeﬁned number of randomly selected attributes (a subset of the entire attribute set). The size of an attribute subset (subspace) is one of the most important factors that stems multitude of inﬂuences over Random Forest. In this paper, we propose a new technique that dynamically determines the size of subspaces based on the relative size of the current data segment to the entire data set. In order to assess the eﬀects of the proposed technique, we conduct experiments involving ﬁve widely used data set from the UCI Machine Learning Repository. The experimental results indicate the capability of the proposed technique on improving the ensemble accuracy of Random Forest.

KW - Decision tree

KW - Decision forest

KW - Random forest

UR - https://link.springer.com/content/pdf/bfm%3A978-3-319-69179-4%2F1.pdf

U2 - 10.1007/978-3-319-69179-4_21

DO - 10.1007/978-3-319-69179-4_21

M3 - Conference paper

SN - 9783319691787

T3 - Edit Lecture Notes in Artificial Intelligence

SP - 303

EP - 312

BT - Proceedings of the 13th International Conference of Advanced Data Mining and Applications (ADMA 2017)

A2 - Cong, Gao

A2 - Peng, Wen-Chih

A2 - Zhang, Wei Emma

A2 - Li, Chengliang

A2 - Sun, Aixin

PB - Springer

T2 - The 13th International Conference on Advanced Data Mining and Applications (ADMA)

Y2 - 5 November 2017 through 6 November 2017

ER -

Effects of Dynamic Subspacing in Random Forest

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this