Subspace Clustering for all Seasons

Sergio Peignier 1, 2 Christophe Rigotti 2, 3, 1 Guillaume Beslon 2, 1
2 BEAGLE - Artificial Evolution and Computational Biology
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information, Inria Grenoble - Rhône-Alpes, LBBE - Laboratoire de Biométrie et Biologie Evolutive - UMR 5558
3 DM2L - Data Mining and Machine Learning
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
Abstract : Subspace clustering is recognized as a more general and difficult task than standard clustering since it requires to identify not only objects sharing similar feature values but also the various subspaces where these similarities appear. Many approaches have been investigated for subspace clustering in the literature using various clustering paradigms. In this paper, we present Chameleoclust, an evolutionary subspace clustering algorithm that incorporates a genome having an evolvable structure. The genome is a coarse grained genome defined as a list of tuples (the "genes"), each tuple containing numbers. These tuples are mapped at the phenotype level to denote core point locations in different dimensions, which are then used to collectively build the subspace clusters, by grouping the data around the core points. The algorithm has been assessed using a reference framework for subspace clustering evaluation, and compared to state-of-the-art algorithms on both real and synthetic datasets. The results obtained with the Chameleoclust algorithm show that evolution of evolution, through an evolvable genome structure, is usefull to solve a difficult problem such as subspace clustering.
Document type :
Conference papers
Complete list of metadatas

Cited literature [8 references]  Display  Hide  Download

https://hal.inria.fr/hal-01252793
Contributor : Guillaume Beslon <>
Submitted on : Friday, January 8, 2016 - 5:00:41 PM
Last modification on : Wednesday, November 20, 2019 - 2:44:13 AM
Long-term archiving on: Thursday, November 10, 2016 - 9:44:16 PM

File

EvoEvo2015_submission_2.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01252793, version 1

Citation

Sergio Peignier, Christophe Rigotti, Guillaume Beslon. Subspace Clustering for all Seasons. EvoEvo Workshop (satellite workshop of ECAL 2015), Jul 2015, york, United Kingdom. pp.1-3. ⟨hal-01252793⟩

Share

Metrics

Record views

600

Files downloads

225