Dans cet article, on s'intéresse à la détermination des ordres médians des tournois valués. On propose d'une part des améliorations d'une méthode arborescente permettant de limiter le nombre de nœuds et donc d'accélérer l'énumération des ordres médians. D'autre part, pour les tournois difficiles qui restent incalculables, on propose de réduire le tournoi en éliminant certains candidats.
En classification conceptuelle d'un ensemble d'objets décrits dans un espace de représentation, on cherche à construire une partition des objets en classes disjointes et simultanément une caractérisation de chaque classe dans les termes de l'espace de représentation. Dans le cas, très courant, où cet espace est engendré par des données binaires nous présentons deux algorithmes, dérivés des méthodes ascendantes et descendantes en classification qui maximisent localement un indice de cohésion des...
We first describe four recent methods to cluster vertices of an undirected non weighted connected graph. They are all based on very different principles. The fifth is a combination of classical ideas in optimization applied to graph partitioning. We compare these methods according to their ability to recover classes initially introduced in random graphs with more edges within the classes than between them.
We first describe four recent methods to cluster vertices of an undirected non weighted connected graph. They are all based on very different principles. The fifth is a combination of classical ideas in optimization applied to graph partitioning. We compare these methods according to their ability to recover classes initially introduced in random graphs with more edges within the classes than between them.
En classification par arbre, on cherche à ajuster une dissimilarité donnée par une distance d'arbre. Mais bien souvent, surtout par comparaison de séquences biologiques, les valeurs obtenues sont peu fiables, voire indéterminées. On a alors une distance partielle qui n'est pas définie pour toute paire. Dans ce cas, on peut soit développer une méthode spécifique qui n'utilise que les valeurs disponibles, soit estimer les valeurs manquantes et utiliser une méthode classique pour reconstruire l'arbre....
A method to infer -trees (valued trees having as set of leaves) from incomplete distance arrays (where some entries are uncertain or unknown) is described. It allows us to build an unrooted tree using only 2-3 distance values between the elements of , if they fulfill some explicit conditions. This construction is based on the mapping between -tree and a weighted generalized 2-tree spanning .
Given a simple undirected weighted or unweighted graph, we try to cluster the vertex set into communities and also to quantify the robustness of these clusters. For that task, we propose a new method, called which consists in (i) defining a new clustering algorithm for graphs, (ii) building a set of graphs similar to the initial one, (iii) applying the clustering method to each of them, making a (set) of partitions, (iv) computing a consensus partition for this profile, which is the final graph...
A method to infer -trees (valued trees having as set of leaves) from incomplete distance arrays (where some entries are uncertain or unknown) is described. It allows us to build an unrooted tree using only 2-3 distance values between the elements of , if they fulfill some explicit conditions. This construction is based on the mapping between -tree and a weighted generalized 2-tree spanning .
Given a simple undirected weighted or unweighted graph, we try to cluster the vertex set into communities and also to quantify the robustness of these clusters. For that task, we propose a new method, called which consists in (i) defining a new clustering algorithm for graphs, (ii) building a set of graphs similar to the initial one, (iii) applying the clustering method to each of them, making a (set) of partitions, (iv) computing a consensus partition for this profile, which is the final graph...
Page 1