Rule mining and classification in the presence of feature level and class label ambiguities

K. K R G K Hewawasam, Kamal Premaratne, Mei-Ling Shyu, S. P. Subasingha

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Numerous applications of topical interest call for knowledge discovery and classification from information that may be inaccurate and/or incomplete. For example, in an airport threat classification scenario, data from heterogeneous sensors are used to extract features for classifying potential threats. This requires a training set that utilizes non-traditional information sources (e.g., domain experts) to assign a threat level to each training set instance. Sensor reliability, accuracy, noise, etc., all contribute to feature level ambiguities; conflicting opinions of experts generate class label ambiguities that may however indicate important clues. To accommodate these, a belief theoretic approach is proposed. It utilizes a data structure that facilitates belief/plausibility queries regarding "ambiguous" itemsets. An efficient apriori-like algorithm is then developed to extract frequent such itemsets and to generate corresponding association rules. These are then used to classify an incoming "ambiguous" data instance into a class label (which may be "hard" or "soft"). To test its performance, the proposed algorithm is compared with C4.5 for several databases from the UCI repository and a threat assessment application scenario.

Original languageEnglish (US)
Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
EditorsK.L. Priddy
Pages98-107
Number of pages10
Volume5803
DOIs
StatePublished - 2005
EventIntelligent Computing: Theory and Applications III - Orlando, FL, United States
Duration: Mar 28 2005Mar 29 2005

Other

OtherIntelligent Computing: Theory and Applications III
CountryUnited States
CityOrlando, FL
Period3/28/053/29/05

Fingerprint

ambiguity
Labels
education
data mining
data structures
airports
performance tests
sensors
Association rules
Sensors
classifying
Airports
Data mining
Data structures

Keywords

  • Association rules
  • Classification
  • Data ambiguities
  • Data mining
  • Dempster-Shafer belief theory
  • Imperfect data
  • Missing data

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Condensed Matter Physics

Cite this

Hewawasam, K. K. R. G. K., Premaratne, K., Shyu, M-L., & Subasingha, S. P. (2005). Rule mining and classification in the presence of feature level and class label ambiguities. In K. L. Priddy (Ed.), Proceedings of SPIE - The International Society for Optical Engineering (Vol. 5803, pp. 98-107). [13] https://doi.org/10.1117/12.603993

Rule mining and classification in the presence of feature level and class label ambiguities. / Hewawasam, K. K R G K; Premaratne, Kamal; Shyu, Mei-Ling; Subasingha, S. P.

Proceedings of SPIE - The International Society for Optical Engineering. ed. / K.L. Priddy. Vol. 5803 2005. p. 98-107 13.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hewawasam, KKRGK, Premaratne, K, Shyu, M-L & Subasingha, SP 2005, Rule mining and classification in the presence of feature level and class label ambiguities. in KL Priddy (ed.), Proceedings of SPIE - The International Society for Optical Engineering. vol. 5803, 13, pp. 98-107, Intelligent Computing: Theory and Applications III, Orlando, FL, United States, 3/28/05. https://doi.org/10.1117/12.603993
Hewawasam KKRGK, Premaratne K, Shyu M-L, Subasingha SP. Rule mining and classification in the presence of feature level and class label ambiguities. In Priddy KL, editor, Proceedings of SPIE - The International Society for Optical Engineering. Vol. 5803. 2005. p. 98-107. 13 https://doi.org/10.1117/12.603993
Hewawasam, K. K R G K ; Premaratne, Kamal ; Shyu, Mei-Ling ; Subasingha, S. P. / Rule mining and classification in the presence of feature level and class label ambiguities. Proceedings of SPIE - The International Society for Optical Engineering. editor / K.L. Priddy. Vol. 5803 2005. pp. 98-107
@inproceedings{e2e801d30bcc4ae186107f4e8378eac3,
title = "Rule mining and classification in the presence of feature level and class label ambiguities",
abstract = "Numerous applications of topical interest call for knowledge discovery and classification from information that may be inaccurate and/or incomplete. For example, in an airport threat classification scenario, data from heterogeneous sensors are used to extract features for classifying potential threats. This requires a training set that utilizes non-traditional information sources (e.g., domain experts) to assign a threat level to each training set instance. Sensor reliability, accuracy, noise, etc., all contribute to feature level ambiguities; conflicting opinions of experts generate class label ambiguities that may however indicate important clues. To accommodate these, a belief theoretic approach is proposed. It utilizes a data structure that facilitates belief/plausibility queries regarding {"}ambiguous{"} itemsets. An efficient apriori-like algorithm is then developed to extract frequent such itemsets and to generate corresponding association rules. These are then used to classify an incoming {"}ambiguous{"} data instance into a class label (which may be {"}hard{"} or {"}soft{"}). To test its performance, the proposed algorithm is compared with C4.5 for several databases from the UCI repository and a threat assessment application scenario.",
keywords = "Association rules, Classification, Data ambiguities, Data mining, Dempster-Shafer belief theory, Imperfect data, Missing data",
author = "Hewawasam, {K. K R G K} and Kamal Premaratne and Mei-Ling Shyu and Subasingha, {S. P.}",
year = "2005",
doi = "10.1117/12.603993",
language = "English (US)",
volume = "5803",
pages = "98--107",
editor = "K.L. Priddy",
booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",

}

TY - GEN

T1 - Rule mining and classification in the presence of feature level and class label ambiguities

AU - Hewawasam, K. K R G K

AU - Premaratne, Kamal

AU - Shyu, Mei-Ling

AU - Subasingha, S. P.

PY - 2005

Y1 - 2005

N2 - Numerous applications of topical interest call for knowledge discovery and classification from information that may be inaccurate and/or incomplete. For example, in an airport threat classification scenario, data from heterogeneous sensors are used to extract features for classifying potential threats. This requires a training set that utilizes non-traditional information sources (e.g., domain experts) to assign a threat level to each training set instance. Sensor reliability, accuracy, noise, etc., all contribute to feature level ambiguities; conflicting opinions of experts generate class label ambiguities that may however indicate important clues. To accommodate these, a belief theoretic approach is proposed. It utilizes a data structure that facilitates belief/plausibility queries regarding "ambiguous" itemsets. An efficient apriori-like algorithm is then developed to extract frequent such itemsets and to generate corresponding association rules. These are then used to classify an incoming "ambiguous" data instance into a class label (which may be "hard" or "soft"). To test its performance, the proposed algorithm is compared with C4.5 for several databases from the UCI repository and a threat assessment application scenario.

AB - Numerous applications of topical interest call for knowledge discovery and classification from information that may be inaccurate and/or incomplete. For example, in an airport threat classification scenario, data from heterogeneous sensors are used to extract features for classifying potential threats. This requires a training set that utilizes non-traditional information sources (e.g., domain experts) to assign a threat level to each training set instance. Sensor reliability, accuracy, noise, etc., all contribute to feature level ambiguities; conflicting opinions of experts generate class label ambiguities that may however indicate important clues. To accommodate these, a belief theoretic approach is proposed. It utilizes a data structure that facilitates belief/plausibility queries regarding "ambiguous" itemsets. An efficient apriori-like algorithm is then developed to extract frequent such itemsets and to generate corresponding association rules. These are then used to classify an incoming "ambiguous" data instance into a class label (which may be "hard" or "soft"). To test its performance, the proposed algorithm is compared with C4.5 for several databases from the UCI repository and a threat assessment application scenario.

KW - Association rules

KW - Classification

KW - Data ambiguities

KW - Data mining

KW - Dempster-Shafer belief theory

KW - Imperfect data

KW - Missing data

UR - http://www.scopus.com/inward/record.url?scp=27544507840&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=27544507840&partnerID=8YFLogxK

U2 - 10.1117/12.603993

DO - 10.1117/12.603993

M3 - Conference contribution

VL - 5803

SP - 98

EP - 107

BT - Proceedings of SPIE - The International Society for Optical Engineering

A2 - Priddy, K.L.

ER -