|
|
Lundi 9 Mars
Heure: |
14:00 - 15:00 |
Lieu: |
Salle B107, bâtiment B, Université de Villetaneuse |
Résumé: |
Sequence Classification Based on Delta-Free Sequential Patterns |
Description: |
Pierre Holat Sequential pattern mining is one of the most studied and challenging tasks in data mining. However, the extension of well-known methods from many other classical patterns to sequences is not a trivial task. This talk presents a study of the notion of delta-freeness for sequences. While this notion has extensively been discussed for itemsets, the work described in this talk is the first to extend it to sequences. In this work, we define an efficient algorithm devoted to the extraction of delta-free sequential patterns. Furthermore, we show the advantage of the delta-free sequences and highlight their importance when building sequence classifiers, and we show how they can be used to address the feature selection problem in statistical classifiers, as well as to build symbolic classifiers which optimizes both accuracy and earliness of predictions. |
Mardi 10 Mars
Heure: |
12:30 - 13:30 |
Lieu: |
Salle B107, bâtiment B, Université de Villetaneuse |
Résumé: |
Multiband Robust Optimization: theory and applications |
Description: |
Fabio D'Andreagiovanni Over the last years, Robust Optimization (RO) has emerged as an effective and efficient methodology to tackle data uncertainty in real-world optimization problems. RO takes into account data uncertainty in the shape of hard constraints that restrict the feasible set and maintain only robust solutions, i.e. solutions that remain feasible even when the values of the input data change. In this talk, we provide an overview of our research about theory and applications of RO. Specifically, we present Multiband Robustness, a new model for RO that we recently proposed to generalize and refine the classical Gamma-robustness model by Bertsimas and Sim. The main aim of our new model is to provide a refined representation of arbitrary non- symmetric distributions of the uncertainty, that are commonly present in real-world applications. Such refined representation grants a reduction in conservatism of robust solutions, while maintaining the accessibility and computational tractability that have been a key factor of success of Gamma-robustness. We also provide an overview of applications of the Multiband model to real-world problems that we have considered in past and ongoing research and industrial projects. |
Heure: |
14:00 - 17:00 |
Lieu: |
Salle B107, bâtiment B, Université de Villetaneuse |
Résumé: |
Une généralisation des mots de Christoffel en dimension d. |
Description: |
Sébastien Labbé In this work, we extend the definition of Christoffel words todirected subgraphs of the hypercubic lattice in arbitrary dimensionthat wecall Christoffel graphs. Christoffel graphs when $d=2$ correspond towell-known Christoffel words.We show that Christoffel graphs have similar properties to those ofChristoffel words: symmetry of their central part and conjugation withtheirreversal. Our main result extends Pirillo's theorem (characterization ofChristoffel words which asserts that a word $amb$ is a Christoffelword if andonly if it is conjugate to $bma$) in arbitrary dimension.In the generalization, the map $ambmapsto bma$ is seen as a flipoperation ongraphs embedded in $mathbb{Z}^d$ and the conjugation is a translation.We show that a fully periodic subgraph of the hypercubic lattice is atranslate of its flip if and only if it is a Christoffel graph.This is joint work with Christophe Reutenauer.Preprint is available at http://arxiv.org/abs/1404.4021. |
Jeudi 12 Mars
Heure: |
12:15 - 13:30 |
Lieu: |
Salle B107, bâtiment B, Université de Villetaneuse |
Résumé: |
Extraction de motifs graduels pour le résumé de données numériques |
Description: |
Amal Oudni Les résumés linguistiques constituent une représentation linguistique décrivant un ensemble de données numériques en phrases simples, compréhensibles et facilement interprétables. Cet exposé se concentre sur les résumés linguistiques sous une forme particulière, appelée motifs graduels, exprimant des corrélations de co-variations des valeurs des attributs : on peut les illustrer pa r un exemple du type « plus l'âge du père est avancé à la naissance des enfants, plus le risque d'autisme chez les enfants augmente ».
Lors de mon exposé, je présenterai trois types de contextualisation de ces motifs graduels. Dans un premier temps, je décrirai le problème de motifs graduels contradictoires et une nouvelle approche permettant déviter toute ambiguïté entre motifs validés. Je présenterai ensuite la caractérisation des motifs graduels qui permet de faciliter l'interprétation des motifs générés en grand nombre en introduisant une nouvelle clause linguistiquement introduite par l'expression « surtout si ». Je terminerai par la présentation dune contextualisation qualifiant le mode de dépendances graduelles, en introduisant la notion d'accélération par rapport aux autres attributs.
Pour chacune des formes de motifs enrichis extraits, je présenterai une formalisation de la sémantique et l'interprétation souhaitées, des mesures de qualité pour évaluer et quantifier la validité des motifs proposés, ainsi que des algorithmes efficaces d'extraction automatique de ces motifs maximisant les critères de qualité définis. |
Heure: |
16:00 - 17:00 |
Lieu: |
Salle B107, bâtiment B, Université de Villetaneuse |
Résumé: |
Building Bridges Between Sets of Partial Orders |
Description: |
Hernán Ponce de León Partial orders are a fundamental mathematical structure capable of representing concurrency and causality on a set of atomic events. In many applications it is essential to consider multiple partial orders, each representing a particular behavioral scenario or an operating mode of a modeled system.
In this talk I will present two mathematical formalisms capable of the compressed representation of sets of partial orders: Labeled Event Structures (LESs) and Conditional Partial Order Graphs (CPOGs). I will demonstrate their advantages and disadvantages and propose efficient algorithms for transforming of a set of partial orders from a given compressed representation in one formalism into an equivalent representation in another formalism without the explicit enumeration of each scenario. These algorithms make use of an intermediate mathematical formalism, which we call Conditional Labeled Event Structures (CLESs), that combines the advantages of LESs and CPOGs.
This is joint work with Andrey Mokhov (Newcastle University) |
Vendredi 13 Mars
Heure: |
11:00 - 12:30 |
Lieu: |
Salle B107, bâtiment B, Université de Villetaneuse |
Résumé: |
Functors are Type Refinement Systems |
Description: |
Noam Zeilberger The standard reading of type theory through the lens of category theory is based on the idea of viewing a type system as a category of well-typed terms. In this joint work with Paul-André Melliès we propose a basic revision of this reading: rather than interpreting type systems as categories, we describe them as functors from a category of typing derivations to a category of underlying terms. Then, turning this around, we explain how in fact *any* functor gives rise to a generalized type system, with an abstract notion of typing judgment, typing derivations and typing rules. This leads to a purely categorical reformulation of various natural classes of type systems as natural classes of functors.
In the talk I want to motivate and introduce this general framework (which can also be seen as providing a categorical analysis of _refinement types_), and as a larger example give a sketch of how the framework can be used to formalize an elegant proof of a coherence theorem by John Reynolds. If time permits, I will also describe some of the natural questions raised by this perspective that are the subject of ongoing research. |
|
|