Browsing
     by title


0-9 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

or enter first few letters:   
OK
See detailEtude différentielle du développement opératoire
Beckers, Jacqueline ULg; Thirion, Anne-Marie ULg

in Recherches sur les handicaps socioculturels de 0 à 7-8 ans (1973)

Detailed reference viewed: 6 (0 ULg)
Full Text
See detailEtude du classement par forêts aléatoires d'échantillons perturbés à forte structure d'interaction
Brostaux, Yves ULg

Doctoral thesis (2005)

Amongst classification methods, forests of decision trees (Random Forests, BREIMAN, 2001) are highly versatile concerning descriptive attributes' or target variable's nature and shape of the concept to ... [more ▼]

Amongst classification methods, forests of decision trees (Random Forests, BREIMAN, 2001) are highly versatile concerning descriptive attributes' or target variable's nature and shape of the concept to estimate. Their diffusion in agronomical sciences is slowed by a lack of information about their ability to learn models with high interaction structures using learning samples with few examples and affected by random noise and irrelevant attributes. This research aim to fill this gap by a systematic exploration of those factors' effects and of the parameters of the Random Forests method, which is done by computer simulations, taking as a reference the classification trees generated by Breiman's CART method (1984). Results show that generating random forests with a partially deterministic attributes selection and a forest size of at least 100 or 500 trees give the best prediction accuracy. Those random forests show a significant increase in prediction accuracy on CART trees, even for low learning sample size (50 examples). This advantage reduce with the global perturbation level (noise and irrelevant attributes) but increase with the learning sample size, as random forests aren't affected by the asymptotic limitation of the learning curve showed by CART method. [less ▲]

Detailed reference viewed: 129 (6 ULg)
Peer Reviewed
See detailEtude du colmatage dans un bioréacteur membranaire à recirculation externe utilisé en nitrification partielle
Kouakou, Edouard; Pirard, Jean-Paul ULg; Marchot, Pierre ULg et al

in Proceedings of MEMPRO 3 - Intégration des membranes dans les procédés (2006)

Detailed reference viewed: 27 (2 ULg)
Full Text
See detailEtude du comportement aéroélastique des structures soumises à des écoulements décrochés
Andrianne, Thomas ULg; Dimitriadis, Grigorios ULg

Conference (2010, March 02)

L'objectif du travail de recherche est le développement d'un modèle numérique simulant les phénomènes aéroélastiques affectant les corps non-profilés, caractérisés par des effets de décrochage/raccrochage ... [more ▼]

L'objectif du travail de recherche est le développement d'un modèle numérique simulant les phénomènes aéroélastiques affectant les corps non-profilés, caractérisés par des effets de décrochage/raccrochage du fluide.L'application de ce modèle aux tabliers de point est prévue. [less ▲]

Detailed reference viewed: 57 (22 ULg)
See detailEtude du comportement au feu des structures mixtes acier-béton
Franssen, Jean-Marc ULg

Doctoral thesis (1987)

This Ph. D. thesis contains the basic therories at the base of the computer software CEFICOSS, the predecessor of the computer software SAFIR.

Detailed reference viewed: 133 (11 ULg)
See detailEtude du comportement d'un laser continu à colorant
Carapelle, Alain ULg

Master's dissertation (1996)

Detailed reference viewed: 15 (11 ULg)
Full Text
See detailEtude du comportement d'une toiture de grandes dimensions soumise à un vent turbulent
Blaise, Nicolas ULg

Master's dissertation (2010)

The aim of this project is the development of a model allowing the calculation of the wind turbulence, using Matlab software. The problems associated wind tunnel pressure measurements (e.g. frequency ... [more ▼]

The aim of this project is the development of a model allowing the calculation of the wind turbulence, using Matlab software. The problems associated wind tunnel pressure measurements (e.g. frequency noise, noise) are taken into account. In order to solve the equation of motion, several methods are applied. First, modal superposition solving decoupled equations by Newmark (NK) and by Fourier transform (FT). Secondly, a stochastic analysis (SRSS and CQC) is performed, using white noise approximation (BB). As an ultimate goal of designing the structure, the calculation of extreme values is realized using a simplified method as well as the statistic method of extreme values. Results show that Newmark method, with fixed time step, produces an elongation of the period and is therefore not recommend. The modal correlation being low, the SRSS analysis is sufficient to determine the dynamic contribution. Results provided by the FT method are closer to the results obtain with CQC-BB method, thus these methods appear to be equivalent. Regarding the calculation of extreme values, the simplified method can calculate positive and negative peak factors, whereas the statistic method of extreme values, applied considering a Gaussian process, can't. In conclusion, the stochastic methods is recommanded because she had the advantage of treating appropriately the noise frequencies for the dynamic contribution. [less ▲]

Detailed reference viewed: 120 (20 ULg)
Full Text
Peer Reviewed
See detailÉtude du comportement de Baillonella toxisperma Pierre (moabi) dans les trouées d’abattage enrichies
Kouadio, Y. L.; Doucet, Jean-Louis ULg

in Biotechnologie, Agronomie, Société et Environnement = Biotechnology, Agronomy, Society and Environment [=BASE] (2009), 13(2), 317-324

The moabi (Baillonella toxisperma Pierre) is a commercial tree in central Africa. Although forest gaps are generally considered as favorable to the regeneration of this species, in this study, a ... [more ▼]

The moabi (Baillonella toxisperma Pierre) is a commercial tree in central Africa. Although forest gaps are generally considered as favorable to the regeneration of this species, in this study, a preliminary inventory conducted within a selective exploitation located in the Dja district, southeastern Cameroon, found a low density of 12.7 seedlings.ha-1 within logging gaps. This suggests that the species was in fact poorly represented in logging gaps. In order to further study the behavior of the moabi in logging gaps, 795 seeds were planted in 15 logging gaps and 410 saplings from a tree nursery were introduced in 15 other gaps. A biannual monitoring spanning a period of 30 months showed a 75.9% survival rate for seedlings and 95.5% for saplings. Despite our observation of a significant difference in growth rates between both methods during the first semester, after 30 months of monitoring, seedlings and saplings had reached similar growth rates (167.5 cm and 160.6 cm for seedlings and saplings, respectively). On average, 6.1% of B. toxisperma saplings emerged naturally from the competitive vegetation and had a higher growth rate than other plants species. Growth differences between plants were mainly explained by the geographical orientation of the logging gap (N-S unfavorable) and the density of Macaranga spp. whose abundance inhibits the development of B. toxisperma. In conclusion, both the high survival rate of B. toxisperma in logging gap and its 10 fold higher growth rate in the latter environment when compared to canopy cover suggest the implementation of forestry methods using logging gap during reforesting operations. [less ▲]

Detailed reference viewed: 57 (17 ULg)
See detailEtude du comportement du film de surfactine à l'interface air-eau
Razafindralambo, Hary ULg; Paquot, Michel ULg; Hbid, Choukri et al

Poster (1993, May 13)

Detailed reference viewed: 11 (0 ULg)