Entity

Time filter

Source Type

Villeurbanne, France

Alalga A.,Annaba University | Benabdeslem K.,University of Lyon1 | Taleb N.,Annaba University
Knowledge and Information Systems | Year: 2015

Feature selection, semi-supervised learning and multi-label classification are different challenges for machine learning and data mining communities. While other works have addressed each of these problems separately, in this paper we show how they can be addressed together. We propose a unified framework for semi-supervised multi-label feature selection, based on Laplacian score. In particular, we show how to constrain the function of this score, when data are partially labeled and each instance is associated with a set of labels. We transform the labeled part of data into soft constraints and show how to integrate them in a measure of feature relevance, according to the available labels. Experiments on benchmark data sets are provided for validating the proposed approach and comparing it with some other state-of-the-art feature selection methods in a multi-label context. © 2015 Springer-Verlag London Source


Benabdeslem K.,University of Lyon1 | Hindawi M.,Zirve University | Makkhongkaew R.,University of Lyon1
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2015

Semi-supervised feature selection has become more important as the number of features has increased in partially labeled data sets. In this paper we present a feature weighting-based model to address this problem. Our proposal is based on a semi-supervised clustering paradigm that can rank features according to their relevance from high-dimensional data. We propose an adaptation of the constrained KMeans algorithm to semi-supervised feature selection by an embedded approach. Experiments are provided on several known data sets for validating our proposal. The results are promising and competitive with several representative methods. © Springer International Publishing Switzerland 2015. Source


Benabdeslem K.,University of Lyon1 | Elghazel H.,University of Lyon1 | Hindawi M.,Zirve University
Knowledge and Information Systems | Year: 2015

In this paper, we propose an efficient and robust approach for semi-supervised feature selection, based on the constrained Laplacian score. The main drawback of this method is the choice of the scant supervision information, represented by pairwise constraints. In fact, constraints are proven to have some noise which may deteriorate learning performance. In this work, we try to override any negative effects of constraint set by the variation of their sources. This is achieved by an ensemble technique using both a resampling of data (bagging) and a random subspace strategy. Experiments on high-dimensional datasets are provided for validating the proposed approach and comparing it with other representative feature selection methods. © 2015 Springer-Verlag London Source


Ayadi H.,CNRS Research on Catalysis and Environment in Lyon | Fang W.,CNRS Research on Catalysis and Environment in Lyon | Fang W.,East China University of Science and Technology | Mishra S.,CNRS Research on Catalysis and Environment in Lyon | And 4 more authors.
RSC Advances | Year: 2015

Bottom-up synthesis of a series of GdF3 nanocrystals (NCs) co-doped with 20 mol% Yb3+, 2 mol% Tm3+ and varying amount of Na+ ions is reported using novel molecular precursors [Ln2(TFA)6(diglyme)2] [Ln = Gd (1), Tm (2), Yb (3)] and [Na4(TFA)4(diglyme)]∞ (4) [TFA = CF3COO-; diglyme = MeO(CH2CH2O)2Me]. The single crystal X-ray structures of the new complexes 1-4, which act as excellent precursors because of the absence of water molecules and the ability of the diglyme ligand to behave as a capping reagent during decomposition to render the nanoparticles monodisperse in organic solvents, show a versatile bonding mode of the TFA (dangling η1 or bridging μ,η1,η1; μ4-η1:η1:η1:η1- and μ4-η1:η1:η1:η2 (O, F)) and diglyme (η3 or μ-η3:η1) ligands. The influence of Na+ ion doping on the phase change and upconversion emissions of these GdF3: Yb3+, Tm3+ NCs was studied and compared with the upconversion (UC) intensity of the well-known upconverting NaGdF4: Yb3+, Tm3+ NCs. Among the several UC samples studied, the GdF3: Yb3+, Tm3+ NCs with 30 mol% Na+ seemed to be the most promising as a red emitting UC phosphor, emitting more photons than the NaGdF4: Yb3+, Tm3+ UC NCs in the near IR region. © The Royal Society of Chemistry. Source


Allab K.,University of Lyon1 | Benabdeslem K.,University of Lyon1
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2011

In this paper, we propose to adapt the batch version of self-organizing map (SOM) to background information in clustering task. It deals with constrained clustering with SOM in a deterministic paradigm. In this context we adapt the appropriate topological clustering to pairwise instance level constraints with the study of their informativeness and coherence properties for measuring their utility for the semi-supervised learning process. These measures will provide guidance in selecting the most useful constraint sets for the proposed algorithm. Experiments will be given over several databases for validating our approach in comparison with another constrained clustering ones. © 2011 Springer-Verlag. Source

Discover hidden collaborations