

Meetings 20132014
May 2014, Friday 9 (02:30 pm): W. H. WOODALL (Virginia Tech, Blacksburg, VA 240610439, USA)
The Difficulty in Designing Control Charts with Estimated Parameters
The performance of control charts, such as the Shewhart X control chart, with estimated incontrol parameters has been widely discussed in the literature. Previous studies showed, for example, that at least 400/(n1) Phase I samples, where n > 1 is the sample size, are required so that the X chart performs on average as if the incontrol process parameter values were known. This recommendation was based on the incontrol expected average run length (ARL) performance. The reliance on the expected ARL metric, however, neglects the practitionertopractitioner variability. This variability occurs due to the different historical data sets practitioners use, which results in varying parameter estimates, control limits, and incontrol ARL values.
In this presentation, it is shown that taking this additional type of variability into consideration leads to much larger Phase I samples, far beyond what many previous researchers have recommended, in order to have low levels of variation of incontrol performance among practitioners. The standard deviation of the ARL (SDARL) metric is used to evaluate performance for various amounts of Phase I data. Surprisingly, we show that for a variety of charts no realistic Phase I sample size is sufficient to have confidence that the attained incontrol performance is close to that desired. These results have significant implications on the relationship between process monitoring theory and practice. An alternative approach is presented for designing control charts.
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège Room 025
May 2014, Tuesday 6 (11:00 am): Franco Mascia & Manuel LópezIbáńez (IRIDIA, ULB)
Automatic algorithm configuration methods and automatic design of metaheuristics
Automatic algorithm configuration methods have shown that automatically tuning the parameters of optimization algorithms may lead to much better results, while at the same time saving significant effort to the human designer. Moreover, the potential of these methods to handle a large number of numerical, categorical and conditional parameters opens the door to more powerful applications. One of these applications is the automatic design of metaheuristics. In particular, our recent work has shown that it is possible to use an automatic configuration tool, such as irace, to generate hybrid local search algorithms. The design space of the hybrid local search algorithms is given as a grammar, from which particular algorithms may be instantiated. We convert this grammar description to a parameter space, which can be tuned for a particular problem by means of an automatic configuration tool. The resulting system allows a human designer to automatically find the best hybrid local search algorithm for a particular problem among thousands of potential algorithm designs just by implementing a few problemspecific components.
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège Room 223
 April 2014, Thursday 3 (9:00 am): Pierre SCHAUS (Université Catholique de Louvain)
Constraint programming: an overview
In this talk I will introduce constraint programming (strengths and weaknesses).
I will explain some details about the internals of a CP Solver and the facilities offered to extend the solver.
I will also illustrate that many CP constraints embed wellknown ideas and algorithms from OR (flows, assignments, relaxations, etc) for solving optimization problems with CP.
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège Room 220
March 2014, Tuesday 25 (3:00 pm): Prof. Dr. Herbert Meyr (Universität Hohenheim, Stuttgart)
Supply Chain Planning: software support and promising fields of future research
The seminar will describe the history of computer support in supply
chain planning (SCP). On basis of this, the common structure of today's
commercial SCP software  socalled Advanced Planning systems (APS) as,
for example, the SAP Advanced Planner and Optimizer or Oracle Value
Chain Planning  will be discussed. Deficiencies of current APS give
reason to briefly sketch some promising fields of further scientific
research like the design of industryspecific hierarchical planning
systems, simultaneous lotsizing and scheduling in consumer goods
industries, the transfer of revenue management ideas to manufacturing
industries, using advanced demand information for forecasting and for
measuring decoupling points or robustness issues of strategic network
design.
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège Room 130
December 2013, Friday 6 (2:00 pm): Ashwin Itoo (HECUniversity of Liege)
Harnessing Big (Text) Data with Natural Language Processing
The ubiquity of ICT, and especially, the pervasiveness of Social Media networks have resulted in the generation of a massive amount of data. This phenomenon is commonly referred to as Big Data. It is considered to be of such great relevance that the Economist magazine devoted one issue to it, entitled the Data Deluge. According to estimates by IBM, the volume of data by 2020 is expected to be of the order of 3500 exabytes.
An increasing number of organizations have realized the value that can be extracted from Big Data, and subsequently exploited for corporate activities, such as marketing. In fact, the most successful organizations thrive and achieve their competitive edge based on their ability to leverage upon and monetize huge amounts of data.
In my presentation, I will start by introducing and defining Big Data, since we often hear the term “Big Data” without having a clear indication of what it stands for. I will also provide statistics from wellknown data sources (e.g. Facebook, Twitter) to illustrate the astounding rate of data creation.
A significant fraction of Big Data is in unstructured format, including text (e.g. Facebook messages, Tweets, customer opinions on blog) , video (e.g. on Youtube) and photos (e.g. Instangram, Flickr). I will focus on text data. I will delve into the fascinating topic of Natural Language Processes (known as Text Analytics in business), which is my area of expertise. I will give an overview of this topic, and illustrate a sentiment analysis application developed in collaboration with Philips Consumer Lifestyle. This application automatically classifies opinions of consumers from blogs, forums (e.g. Amazon reviews) as either positive or negative depending on their polarity.
Finally, I will discuss more general issues associated with Big Data and Text Analytics, including the emergence of Data Science as a field of study in its own right, and key research questions/issues that are expected to be crucial in research associated with Big Data in the future.
Big data and analytics have rocketed to the top of the corporate agenda. The most successful companies in the digital age have thrived and achieved their competitive edge based on their ability to leverage and to monetize data.
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège Room 1715
November 2013, Friday 8 (10:00 am): Joseph B. (Joe) MAZZOLA (Cleveland State University)
Business Analytics and Big Data: The emerging field of Data Science
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège Room 1715
October 2013, Thursday 16 (11:00 am): Tom Van Woensel (Eindhoven University of Technology)
Combining Passenger and Freight Transportation with Public Scheduled Line Services
The Pickup and Delivery Problem (PDP) with public scheduled line
services concerns scheduling a set of vehicles to serve two types of requests (passengers and freight).
Part of the freight journey can be carried on scheduled public transportation. We propose an arcbased
mixed integer programming formulation which is solved to optimality using CPLEX. Computational results
provide a clear understanding of the benefits of combining passenger and freight transportation in
current networks.
It is a joint work with Vaeceslav Slavic and Emrah Demir.
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège Room 015
September 2013, Friday 20 (10:00 am): Prof. Martine Labbé (Université Libre de Bruxelles)
A bilevel programming approach for network pricing optimization problems
Consider a general pricing model involving two levels of decisionmaking. The upper level (leader) imposes prices
on a specified set of goods or services while the lower level (follower) optimizes its own objective function, taking
into account the pricing scheme of the leader. This model belongs to the class of bilevel optimization problems
where both objective functions are bilinear.
In this talk, we review this class of hierarchical problems from both theoretical and algorithmic points of view.
We first briefly introduce a general taxation model. and present some of its properties. Then, we focus on the
problem of setting tolls on a specified subset of arcs of a multicommodity transportation network. In this
context the leader corresponds to the profitmaximizing owner of the network, and the follower to users traveling between
nodes of the network. The users are assigned to shortest paths with respect to a generalized cost equal to the sum of the
actual prices for using specific arcs plus routing costs.
Among others, we present complexity results, identify some polynomial cases and propose mixed integer linear
formulations for those pricing problem.
Where: HECManagement School of the University of Liege  14, rue Louvrex (N1) 4000 Liège

