Thomas Stützle

Thomas Stützle
Université Libre de Bruxelles | ULB · Artificial Intelligence Research Laboratory of the Université Libre de Bruxelles (IRIDIA)

Dr. rer. nat.

About

495
Publications
161,853
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
41,811
Citations
Additional affiliations
October 2005 - present
Université Libre de Bruxelles
Description
  • I'm a research associate of the Belgian F.R.S.-FNRS working at the IRIDIA lab of Université Libre de Bruxelles (ULB)
July 1995 - September 2005
Technische Universität Darmstadt

Publications

Publications (495)
Conference Paper
Metaheuristic algorithms are traditionally designed following a manual and iterative algorithm development process. The performance of these algorithms is, however, strongly dependent on their correct tuning, including their configuration and parametrization. This is labour-intensive, error-prone, difficult to reproduce and explores only a limited...
Article
We present a rigorous, component‐based analysis of six widespread metaphor‐based algorithms for tackling continuous optimization problems. In addition to deconstructing the six algorithms into their components and relating them with equivalent components proposed in well‐established techniques, such as particle swarm optimization and evolutionary a...
Article
Since the introduction of Simulated Annealing (SA), researchers have considered variants that keep the same temperature value throughout the whole search and tried to determine whether this strategy can be more effective than the original cooling scheme. Several studied have tried to answer this question without a conclusive answer and without prov...
Article
It has been more than 10 years since the first version of cuckoo search was proposed by Yang and Deb and published in the proceedings of the World Congress on Nature & Biologically Inspired Computing, in 2009. The two main articles on cuckoo search have now been cited almost 8 700 times (according to Google scholar), there are books and chapters pu...
Article
Full-text available
The particle swarm optimization (PSO) algorithm has been the object of many studies and modifications for more than twentyfive years. Ranging from small refinements to the incorporation of sophisticated novel ideas, the majority of modifications proposed to this algorithm have been the result of a manual process in which developers try new designs...
Article
Ant colony optimization (ACO) algorithms have originally been designed for static optimization problems, where the input data is known in advance and is not subject to changes over time. Later, the long term memory of ACO proved effective for reoptimization over environment changes when extended to deal with dynamic combinatorial optimization probl...
Chapter
Full-text available
Over the last decades, plenty of exact and non-exact methods have been proposed to tackle NP-hard optimisation problems.
Preprint
Since the introduction of Simulated Annealing (SA), researchers have considered variants that keep the same temperature value throughout the whole search and tried to determine whether this strategy can be more effective than the original cooling scheme. Several studied have tried to answer this question without a conclusive answer and without pro...
Chapter
Diving heuristics are methods that progressively enlarge a partial solution up to its possible completion, thus ˇjump˘into a solution with no way back. While this is common to all constructive heuristics, diving ones are usually characterized by working on the mathematical formulation of the problem to solve. Some contributions of this type showed...
Chapter
Very Large-Scale Neighborhood Search is not an algorithm or a class of algorithms, but rather a conceptual framework which can be used for solving combinatorial optimization problems. The approach “concentrates on neighborhood search algorithms where the size of the neighborhood is ‘very large’ with respect to the size of the input data.” Typically...
Chapter
Kernel search is a purely matheuristic method, which leverages MIP solvers to obtain heuristic, or possibly optimal, solutions of instances encoded as (mixed) integer linear programming problems. It was first presented as a method to solve mixed-integer linear problems defined on binary variables modeling items selection, together with other intege...
Chapter
The corridor method is a general method originally proposed as a way to gain efficiency in dynamic programming search, possibly losing optimality. Later, it has been extended beyond DP to other exact optimization methods. The basic idea is that of using the exact method over successive restricted portions of the solution space of the given problem....
Chapter
Decompositions are methods derived from the “divide et impera” principle, dictating to break up a difficult problem into smaller ones, and to solve each of the smaller ones separately, ultimately recomposing the individual solutions to get the overall one. Decompositions have longly been applied to solve optimization problems, and they come in many...
Chapter
A specialized thread of metaheuristic research, bordering and often overlapping with Artificial Intelligence, studied heuristics that evolved whole sets of candidate solutions, often named “populations” of solutions. Genetic algorithms were among the first results, and following their success it became common to get inspiration from some natural ph...
Chapter
Matheuristics have become widespread and effective methods for tackling the generalized assignment problem (GAP) and many other NP-hard problems. In fact, in this book we have many such methods, ranging from metaheuristics and mathematical programming techniques but mainly to real matheuristics. In these methods we will see no parameter settings, b...
Chapter
The generalized assignment problem (GAP) asks to assign nclients to mservers in such a way that the assignment cost is minimized, provided that all clients are assigned to a server and that the capacity of each server is not exceeded. It is a problem that appears, by itself or as a subproblem, in a very high number of practical applications and has...
Chapter
Fore-and-Back, previously also denoted as Forward&Backward or simply as F&B (though the algorithm presented in this chapter differs in some details from the previously published ones), is an extension of beam search that can improve its effectiveness and that, when run with no limits on computational resources, becomes an exact solution method. How...
Chapter
Metaheuristic approaches can be classified according to different criteria, one being the number of solutions that are evolved at each stage of the algorithm: one single solution or more than one. This chapter deals with metaheuristic algorithms that evolve one single solution; they are all enhancements of a basic local search procedure. Many diffe...
Article
Full-text available
Automatic design of stochastic local search algorithms has been shown to be very effective in generating algorithms for the permutation flowshop problem for the most studied objectives including makespan, flowtime and total tardiness. The automatic design system uses a configuration tool to combine algorithmic components following a set of rules de...
Book
This book is the first comprehensive tutorial on matheuristics. Matheuristics are based on mathematical extensions of previously known heuristics, mainly metaheuristics, and on original, area-specific approaches. This tutorial provides a detailed discussion of both contributions, presenting the pseudocodes of over 40 algorithms, abundant literature...
Article
Full-text available
Iterative improvement is an optimization technique that finds frequent application in heuristic optimization, but, to the best of our knowledge, has not yet been adopted in the automatic design of control software for robots. In this work, we investigate iterative improvement in the context of the automatic modular design of control software for ro...
Article
Grammar‐based automatic algorithm design has been shown to generate stochastic local search algorithms that compete with or outperform state‐of‐the‐art methods. In such systems, algorithms are divided in components and a grammar is used to describe how to properly combine the components to create a working algorithm. In our approach, the grammar is...
Chapter
In this paper, we carry out a review of the grey wolf, the firefly and the bat algorithms. We identify the concepts involved in these three metaphor-based algorithms and compare them to those proposed in the context of particle swarm optimization. We provide compelling evidence that the grey wolf, the firefly, and the bat algorithms are not novel,...
Chapter
Heuristic optimizers are an important tool in academia and industry, and their performance-optimizing configuration requires a significant amount of expertise. As the proper configuration of algorithms is a crucial aspect in the engineering of heuristic algorithms, a significant research effort has been dedicated over the last years towards moving...
Article
Full-text available
This special issue of the International Transactions in Operational Research focuses on Matheuristics and Metaheuristics and is the largest published to date, highlighting the importance of the field and the broad scope of these methods and the reach of their applications. Academicians and practitioners responded with enthusiasm to three parallel c...
Book
This book constitutes the proceedings of the 12th International Conference on Swarm Intelligence, ANTS 2020, held online -due to COVID-19- in Barcelona Spain, in October 2020. The 20 full papers presented , together with 8 short papers and 5 extended abstracts were carefully reviewed and selected from 50 submissions. ANTS 2020 contributions are dea...
Article
Full-text available
In this article, we rigorously analyze the intelligent water drops (IWD) algorithm, a metaphor-based approach for the approximate solution of discrete optimization problems proposed by Shah-Hosseini (in: Proceedings of the 2007 congress on evolutionary computation (CEC 2007), IEEE Press, Piscataway, NJ, pp 3226–3231, 2007). We demonstrate that all...
Article
Industrial production scheduling problems are challenges that researchers have been trying to solve for decades. Many practical scheduling problems such as the hybrid flowshop are NP-hard. As a result, researchers resort to metaheuristics to obtain effective and efficient solutions. The traditional design process of metaheuristics is mainly manual,...
Article
A recent comparison of well-established multi-objective evolutionary algorithms (MOEAs) has helped better identify the current state-of-the-art by considering (i) parameter tuning through automatic configuration, (ii) a wide range of different setups, and (iii) various performance metrics. Here, we automatically devise MOEAs with verified state-of-...
Article
Full-text available
Designing collective behaviors for robot swarms is a difficult endeavor due to their fully distributed, highly redundant, and ever-changing nature. To overcome the challenge, a few approaches have been proposed, which can be classified as manual, semi-automatic, or automatic design. This paper is intended to be the manifesto of the automatic off-li...
Conference Paper
Early works on external solution archiving have pointed out the benefits of unbounded archivers and there have been great advances, theoretical and algorithmic, in bounded archiving methods. Moreover, recent work has shown that the populations of most multi- and many-objective evolutionary algorithms (MOEAs) lack the properties that one would desir...
Article
Stochastic local search methods are at the core of many effective heuristics for tackling different permutation flowshop problems (PFSPs). Usually, such algorithms require a careful, manual algorithm engineering effort to reach high performance. An alternative to the manual algorithm engineering is the automated design of effective SLS algorithms t...
Chapter
One way to speed up the algorithm configuration task is to use short runs instead of long runs as much as possible, but without discarding the configurations that eventually do well on the long runs. We consider the problem of selecting the top performing configurations of Conditional Markov Chain Search (CMCS), a general algorithm schema that incl...
Article
Full-text available
We study the impact of altering the sampling space of parameters in automatic algorithm configurators. We show that a proper transformation can strongly improve the convergence towards better configurations; at the same time, biases about good parameter values, possibly based on misleading prior knowledge, may lead to wrong choices in the transform...
Article
Full-text available
We study the empirical scaling of the running time required by state-of-the-art exact and inexact TSP algorithms for finding optimal solutions to Euclidean TSP instances as a function of instance size. In particular, we use a recently introduced statistical approach to obtain scaling models from observed performance data and to assess the accuracy...
Article
Simulated Annealing (SA) is one of the oldest metaheuristics and has been adapted to solve many combinatorial optimization problems. Over the years, many authors have proposed both general and problem-specific improvements and variants of SA. We propose to accumulate this knowledge into automatically configurable, algorithmic frameworks so that for...
Article
Full-text available
Research on multi-objective evolutionary algorithms (MOEAs) has produced over the past decades a large number of algorithms and a rich literature on performance assessment tools to evaluate and compare them. Yet, newly proposed MOEAs are typically compared against very few, often a decade older MOEAs. One reason for this apparent contradiction is t...
Conference Paper
The \({\textsf {irace}} \) package is a widely used for automatic algorithm configuration and implements various iterated racing procedures. The original \({\textsf {irace}} \) was designed for the optimisation of the solution quality reached within a given running time, a situation frequently arising when configuring algorithms such as stochastic...
Conference Paper
Past research has shown that the performance of algorithms for solving the Quadratic Assignment Problem (QAP) depends on the structure and the size of the instances. In this paper, we study the bi-objective QAP, which is a multi-objective extension of the single-objective QAP to two objectives. The algorithm we propose extends a high-performing Sim...
Article
Full-text available
Many algorithms for minimizing the weighted tardiness in the permutation flowshop problem rely on local search procedures. An increase in the efficiency of evaluating the objective function for neighboring candidate solutions directly also improves the performance of such algorithms. In this paper, we introduce a speed up of the evaluation of the w...
Chapter
Iterated greedy is a search method that iterates through applications of construction heuristics using the repeated execution of two main phases, the partial destruction of a complete candidate solution and a subsequent reconstruction of a complete candidate solution. Iterated greedy is based on a simple principle, and methods based on this princip...
Conference Paper
Automatic algorithm configurators can greatly improve the performance of algorithms by effectively searching the parameter space. As algorithm configuration tasks can have large parameter spaces and the execution of candidate algorithm configurations is often very costly in terms of computation time, further improvements in the search techniques us...
Conference Paper
Full-text available
ABC-X is a generalized, automatically con€gurable framework for the Arti€cial Bee Colony (ABC) metaheuristic. ABC-X has recently been proposed and it has initially been tested on different benchmark functions sets, showing very good results when compared to known ABC algorithms. However, it has never been used in an industrial application. In this...
Conference Paper
Over the recent years, several tools for the automated configuration of parameterized algorithms have been developed. These tools, also called configurators, have themselves parameters that influence their search behavior and make them malleable to different kinds of configuration tasks. The default values of these parameters are set manually based...
Article
Methods for automatic algorithm configuration integrate some search mechanism for generating candidate algorithm configurations and mechanisms for handling the stochasticity of the algorithm configuration problem. One popular algorithm configurator is ParamILS, which searches the configuration space using an iterated local search algorithm. In our...
Conference Paper
Job-shop scheduling problems have received a considerable attention in the literature. While the most tackled objective in this area is makespan, job-shop scheduling problems with other objectives such as the minimization of the weighted or unweighted tardiness, the number of late jobs, or the sum of the jobs’ completion times have been considered....
Article
Full-text available
The artificial bee colony (ABC) algorithm is a popular metaheuristic that was originally conceived for tackling continuous function optimization tasks. Over the last decade, a large number of variants of ABC have been proposed, making it by now a well-studied swarm intelligence algorithm. Typically, in a paper on algorithmic variants of ABC algorit...
Conference Paper
The inverted generational distance (IGD) is a metric for assessing the quality of approximations to the Pareto front obtained by multi-objective optimization algorithms. The IGD has become the most commonly used metric in the context of many-objective problems, i.e., those with more than three objectives. The averaged Hausdorff distance and \(\text...
Chapter
Iterated local search is a metaheuristic that embeds an improvement heuristic within an iterative process generating a chain of solutions. Often, the improvement method is some kind of local search algorithm and, hence, the name of the metaheuristic. The iterative process in iterated local search consists in a perturbation of the current solution,...
Article
Permutation flowshop scheduling problems (PFSPs) and, in particular, the variant with the objective of minimizing makespan have received an enormous attention in scheduling research and combinatorial optimization. As a result, the algorithmic approaches to this PFSP variant have reached extremely high performance. Currently, one of the most effecti...
Article
Matheuristics are methods that exploit mathematical programming techniques in heuristic and metaheuristic frameworks, granting to mathematical programming approaches the problem robust-ness and time effectiveness that characterize heuristics, or exploiting the mathematical programming model formulation in the customization of a heuristic for specif...
Conference Paper
Effective traffic light control algorithms are of central importance for reducing congestion. While the currently most effective algorithms rely on expensive infrastructure to obtain knowledge of the traffic state, within the COLOMBO project, low-cost adaptive traffic light controllers have been examined that rely on swarm intelligence principles a...
Conference Paper
In this paper we argue that flexible algorithm frameworks can be useful to capture the wide variety of algorithmic components for heuristic algorithms and serve as basic experimental frameworks. One of the utilities is that they can implement the wide variety of different algorithm components and their alternative choices for single stochastic loca...
Conference Paper
Most optimization algorithms, including evolutionary algorithms and metaheuristics, and general-purpose solvers for integer or constraint programming, have often many parameters that need to be properly configured (i.e., tuned) for obtaining the best results on a particular problem. Automatic (offline) algorithm configuration methods help algorithm...