QuantOM

About QuantOM
Members
Publications
Some projects
Seminars&conferences
What is OR ?
Podcasts
Links
Contact

 

Meetings 2018-2019

  • March 2019, Monday 25 (3 pm): by Hande Yaman Paternotte (KU Leuven)

    Team as a Service: Team Formation on Social Networks

    Team as a Service (TaaS) is a new outsourcing service that enables on-demand creation and management of distributed teams for fast growing companies. The companies that use the TaaS model form a team according to the needs of a given project and provide managerial service throughout. Motivated by this new application, we study the Team Formation Problem (TFP) in which the quality and cost of communication is taken into account using the proximity of the potential members in a social network. Given a set of required skills, TFP aims to construct a capable team that can communicate and collaborate e ffectively. We study TFP with two measures for communication e ffectiveness: sum of distances and diameter. We compare the quality of solutions using various performance measures. Our experiments indicate the following: First, considering only the sum of distances or the diameter as the communication cost may yield solutions that perform poorly in other performance measures. Second, the heuristics in the literature fi nd solutions that are signifi cantly far from optimal whereas a general purpose solver is capable of solving small and medium size instances to optimality. To fi nd solutions that perform well on several performance measures, we propose the diameter constrained TFP with sum of distances objective. To solve larger sizes, we propose a novel branch and bound approach: we formulate the problem as a quadratic set covering problem, use the reformulation-linearization technique and relaxation to be able to decompose it into a series of linear set covering problems and impose the relaxed constraints through branching. Our computational experiments show that the algorithm is capable of solving large sizes that are intractable for the solver. Finally, we study a two stage stochastic variant of the problem with uncertain communication costs and report the results of preliminary experiments.

    Where: HEC-University of Liege - 14, rue Louvrex (N1)- 4000 Liège room 1711
  • March 2019, Monday 18 (3:30 pm): by Le Minh Nguyen (Advanced Institute of Science and Technology in Japan)

    Deep Learning

    Deep Learning methods have revolutionized the field of AI, and have shown much promise in applications such as computer vision and natural language processing. In this talk, we will focus on the legal domain. Specifically, we will present our current research (algorithms and results) on deep learning methods applied for analyzing legal texts. These latter texts are particularly problematic to classical NLP algorithms as they are syntactically complex and long, which makes it hard to create computational models of the underlying language. We will describe how deep learning helps in overcoming some of these challenges.

    Where: HEC-University of Liege - 14, rue Louvrex (N1)- 4000 Liège room 1711
  • December 2018, Monday 10 (12:30 pm): by Michael Schyns and Quentin Valembois (HEC-Liège)

    Et si (presque) tout était possible: La réalité virtuelle au service des entreprises

    Nous vivons dans un monde digital où la technologie permet de transformer en profondeur de nombreux processus de gestion. Machine et deep learning, blockchain, machines intelligentes et véhicules autonomes, digital twins, IoT, e-textiles, immersion via réalité virtuelle, augmentée et mixte sont quelques-unes des technologies les plus prometteuses pour les entreprises. Dans cet exposé, nous nous concentrerons essentiellement sur les technologies immersives et les présenterons brièvement. Avec la réalité virtuelle, (presque) tout est possible ! En coiffant un casque de réalité virtuelle, l'utilisateur se déconnecte momentanément du monde réel pour être immergé dans un environnement graphique et sonore qui devient sa nouvelle réalité. Dans ce nouvel univers entièrement contrôlable, il peut expérimenter différentes tâches, parfois délicates, en toute sécurité et à moindre coût. Grâce à sa cousine, la réalité augmentée, une personne peut enrichir son environnement (réel) avec des couches d'information et ainsi être aidée dans ses prises de décision et dans le pilotage de processus.
    Au HEC Digital Lab, dans le cadre de nos missions d'enseignement, de recherche et de service à la communauté, régulièrement pour répondre à des demandes d'entreprises partenaires, nous concevons de tels environnements et testons comment les rendre les plus immersifs et efficaces. Nous sommes également un membre fondateur du groupe Teaching With VR qui, au niveau de l'Université, prépare de nouvelles formes d'enseignement basée sur ces technologies.
    Lors de cette séance, nous utiliserons des casques Oculus, Vive et Hololens pour présenter quelques-uns de nos environnements: apprentissage d'une langue étrangère, validation d'un projet de construction (HEC), gestion d'un entrepôt logistique, sensibilisation à la sécurité sur chantier et aux procédures d'urgence, formation au pilotage d'un Boeing 737, guidage pour la conception de palettes, tourisme et marketing, visite de sites...
    http://www.teachingwithvr.uliege.be

    Where: HEC-University of Liege - 14, rue Louvrex (N1)- 4000 Liège room 025
  • November 2018, Tuesday 13 (12:30 pm): by Morteza Davari (KU Leuven)

    Minimizing makespan on a single-machine with release dates and inventory constraints

    We consider a single-machine scheduling problem with release dates and inventory constraints. Each job has a deterministic processing time and has an impact (either positive or negative) on the central inventory level. We aim to find a sequence of jobs such that the makespan is minimized while all release dates and inventory constraints are met. We show that the problem is strongly NP-hard even when the capacity of the inventory is infinite. To solve the problem, we introduce two mixed integer programming (MIP) formulations, a dynamic programming (DP) approach and also a branch-and-bound (B&B) algorithm.
    This problem has some interesting applications in scheduling load/unload operations of warehouses. In a loading/unloading terminal, trucks arrive in a loading/unloading dock either to deliver (unload) or to pick up (load) a certain number of final products. The terminal include a storage space that is used to store these final products in inventory.

    Where: HEC-University of Liege - 14, rue Louvrex (N1)- 4000 Liège room 119
  • November 2018, Thursday 8 (01:30 pm): by Samedi Heng (HEC-Liège)

    Using Goal-Oriented Modeling to Analyze Requirements in Agile Methods

    User Stories (US) are the most commonly used requirements artifacts within agile methods such as eXtreme Programming (XP) and Scrum. They are written in natural language using prose or following a specific template. In practice, many templates have been proposed with no semantic for each syntax used in these templates. Within this research, we first propose the unified model for US templates and provide semantic of each syntax. Then, we adapt the goal-oriented modeling language for analyzing US sets; it is called Rationale Tree (RT). It allows identifying dependencies between US and provides a global view of the systems which allows improving the planning activities, and therefore the development life cycle of agile methods.
    This research is part of the PhD thesis of Dr. Samedi Heng under the supervision of Prof. Manuel Kolp (UClouvain) and Prof. Yves Wautelet (KULeuven).

    Where: HEC-University of Liege - 14, rue Louvrex (N1)- 4000 Liège room 126

 
 
  HEC-Management School - University of Liège