Kernel Rules Classifiers Revisited (Extended Abstract)
Identifieur interne : 00C626 ( Main/Merge ); précédent : 00C625; suivant : 00C627Kernel Rules Classifiers Revisited (Extended Abstract)
Auteurs : Nicolas Lachiche ; Pierre MarquisSource :
English descriptors
- KwdEn :
Abstract
Rymon shows that kernel rules play an important role in many induction techniques. In this paper, kernel rules are revisited along both the semantic and the computational dimensions. First, the inductive principle on which kernel rules classifiers rely is pointed out. Then, connections between rule-based classifiers, kernel rules and decision trees are mapped out. Particularly, kernel rules are shown at the logical core of decision trees. We also show how kernel rules classification can be improved w.r.t. a computational point of view\, : we present a polynomial time scope classification agorithm allowing us to classify any object as kernel rules classifiers would do, but without computing the kernel rules. This algorithm relies on a topological interpretation of the kernel rules inductive principle\, ; interestingly, the explanatory power of kernel rules classifiers is mainly preserved since kernel rules can be generated, on a as needed-basis, by this scope algorithm.
Links toward previous steps (curation, corpus...)
- to stream Crin, to step Corpus: 001C02
- to stream Crin, to step Curation: 001C02
- to stream Crin, to step Checkpoint: 002867
Links to Exploration step
CRIN:lachiche96aLe document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="fr" wicri:score="-102">Kernel Rules Classifiers Revisited (Extended Abstract)</title>
</titleStmt>
<publicationStmt><idno type="RBID">CRIN:lachiche96a</idno>
<date when="1996" year="1996">1996</date>
<idno type="wicri:Area/Crin/Corpus">001C02</idno>
<idno type="wicri:Area/Crin/Curation">001C02</idno>
<idno type="wicri:explorRef" wicri:stream="Crin" wicri:step="Curation">001C02</idno>
<idno type="wicri:Area/Crin/Checkpoint">002867</idno>
<idno type="wicri:explorRef" wicri:stream="Crin" wicri:step="Checkpoint">002867</idno>
<idno type="wicri:Area/Main/Merge">00C626</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title xml:lang="fr">Kernel Rules Classifiers Revisited (Extended Abstract)</title>
<author><name sortKey="Lachiche, Nicolas" sort="Lachiche, Nicolas" uniqKey="Lachiche N" first="Nicolas" last="Lachiche">Nicolas Lachiche</name>
</author>
<author><name sortKey="Marquis, Pierre" sort="Marquis, Pierre" uniqKey="Marquis P" first="Pierre" last="Marquis">Pierre Marquis</name>
</author>
</analytic>
</biblStruct>
</sourceDesc>
</fileDesc>
<profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Kernel rule</term>
<term>classification</term>
<term>decision tree</term>
</keywords>
</textClass>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en" wicri:score="2690">Rymon shows that kernel rules play an important role in many induction techniques. In this paper, kernel rules are revisited along both the semantic and the computational dimensions. First, the inductive principle on which kernel rules classifiers rely is pointed out. Then, connections between rule-based classifiers, kernel rules and decision trees are mapped out. Particularly, kernel rules are shown at the logical core of decision trees. We also show how kernel rules classification can be improved w.r.t. a computational point of view\, : we present a polynomial time scope classification agorithm allowing us to classify any object as kernel rules classifiers would do, but without computing the kernel rules. This algorithm relies on a topological interpretation of the kernel rules inductive principle\, ; interestingly, the explanatory power of kernel rules classifiers is mainly preserved since kernel rules can be generated, on a as needed-basis, by this scope algorithm.</div>
</front>
</TEI>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Main/Merge
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 00C626 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Merge/biblio.hfd -nk 00C626 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Wicri/Lorraine |area= InforLorV4 |flux= Main |étape= Merge |type= RBID |clé= CRIN:lachiche96a |texte= Kernel Rules Classifiers Revisited (Extended Abstract) }}
![]() | This area was generated with Dilib version V0.6.33. | ![]() |