Instance-based classifiers dealing with ambiguous attributes and class labels

Hans Holland, Miroslav Kubat, Jan Žižka

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

1 Citation (Scopus)

Abstract

Machine learning usually assumes that attribute values, as well as class labels, are either known precisely or not known at all. However, in our attempt to automate evaluation of intrusion detection systems, we have encountered ambiguous examples such that, for instance, an attribute's value in a given example is known to be a or b but definitely not c or d. Previous research usually either "disambiguated" the value by giving preference to a or b, or just replaced it with a "don't-know" symbol. Disliking both of these two approaches, we decided to explore the behavior of other ways to address the situation. To keep the work focused, we limited ourselves to nearest-neighbor classifiers. The paper describes a few techniques and reports relevant experiments. We also discuss certain ambiguity-related issues that deserve closer attention.

Original languageEnglish
Title of host publicationProceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007
Pages598-603
Number of pages6
StatePublished - Dec 28 2007
Event20th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007 - Key West, FL, United States
Duration: May 7 2007May 9 2007

Other

Other20th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007
CountryUnited States
CityKey West, FL
Period5/7/075/9/07

Fingerprint

Intrusion detection
Learning systems
Labels
Classifiers
Experiments

ASJC Scopus subject areas

  • Artificial Intelligence
  • Software

Cite this

Holland, H., Kubat, M., & Žižka, J. (2007). Instance-based classifiers dealing with ambiguous attributes and class labels. In Proceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007 (pp. 598-603)

Instance-based classifiers dealing with ambiguous attributes and class labels. / Holland, Hans; Kubat, Miroslav; Žižka, Jan.

Proceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007. 2007. p. 598-603.

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

Holland, H, Kubat, M & Žižka, J 2007, Instance-based classifiers dealing with ambiguous attributes and class labels. in Proceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007. pp. 598-603, 20th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007, Key West, FL, United States, 5/7/07.
Holland H, Kubat M, Žižka J. Instance-based classifiers dealing with ambiguous attributes and class labels. In Proceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007. 2007. p. 598-603
Holland, Hans ; Kubat, Miroslav ; Žižka, Jan. / Instance-based classifiers dealing with ambiguous attributes and class labels. Proceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007. 2007. pp. 598-603
@inproceedings{c51600a1673c47f29756720c86b62b4d,
title = "Instance-based classifiers dealing with ambiguous attributes and class labels",
abstract = "Machine learning usually assumes that attribute values, as well as class labels, are either known precisely or not known at all. However, in our attempt to automate evaluation of intrusion detection systems, we have encountered ambiguous examples such that, for instance, an attribute's value in a given example is known to be a or b but definitely not c or d. Previous research usually either {"}disambiguated{"} the value by giving preference to a or b, or just replaced it with a {"}don't-know{"} symbol. Disliking both of these two approaches, we decided to explore the behavior of other ways to address the situation. To keep the work focused, we limited ourselves to nearest-neighbor classifiers. The paper describes a few techniques and reports relevant experiments. We also discuss certain ambiguity-related issues that deserve closer attention.",
author = "Hans Holland and Miroslav Kubat and Jan Žižka",
year = "2007",
month = "12",
day = "28",
language = "English",
isbn = "1577353196",
pages = "598--603",
booktitle = "Proceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007",

}

TY - GEN

T1 - Instance-based classifiers dealing with ambiguous attributes and class labels

AU - Holland, Hans

AU - Kubat, Miroslav

AU - Žižka, Jan

PY - 2007/12/28

Y1 - 2007/12/28

N2 - Machine learning usually assumes that attribute values, as well as class labels, are either known precisely or not known at all. However, in our attempt to automate evaluation of intrusion detection systems, we have encountered ambiguous examples such that, for instance, an attribute's value in a given example is known to be a or b but definitely not c or d. Previous research usually either "disambiguated" the value by giving preference to a or b, or just replaced it with a "don't-know" symbol. Disliking both of these two approaches, we decided to explore the behavior of other ways to address the situation. To keep the work focused, we limited ourselves to nearest-neighbor classifiers. The paper describes a few techniques and reports relevant experiments. We also discuss certain ambiguity-related issues that deserve closer attention.

AB - Machine learning usually assumes that attribute values, as well as class labels, are either known precisely or not known at all. However, in our attempt to automate evaluation of intrusion detection systems, we have encountered ambiguous examples such that, for instance, an attribute's value in a given example is known to be a or b but definitely not c or d. Previous research usually either "disambiguated" the value by giving preference to a or b, or just replaced it with a "don't-know" symbol. Disliking both of these two approaches, we decided to explore the behavior of other ways to address the situation. To keep the work focused, we limited ourselves to nearest-neighbor classifiers. The paper describes a few techniques and reports relevant experiments. We also discuss certain ambiguity-related issues that deserve closer attention.

UR - http://www.scopus.com/inward/record.url?scp=37349023594&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=37349023594&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:37349023594

SN - 1577353196

SN - 9781577353195

SP - 598

EP - 603

BT - Proceedings of the Twentieth International Florida Artificial Intelligence Research Society Conference, FLAIRS 2007

ER -