Invited speakers

Kaisa Miettinen is Professor of Industrial Optimization at the University of Jyväskylä (JYU), Finland. She holds a PhD degree in mathematical information technology from JYU. Her research interests include theory, methods, applications and software of nonlinear multiobjective optimization. She heads the Research Group on Multiobjective Optimization and is the director of the thematic research area “Decision Analytics utilizing Causal Models and Multiobjective Optimization” (DEMO, jyu.fi/demo) at JYU. With her group, she develops an open source software framework DESDEO for interactive multiobjective optimization methods (desdeo.it.jyu.fi). It was preceded by WWW-NIMBUS and IND-NIMBUS systems. She has authored about 220 refereed journal, proceedings and collection papers, edited 20 proceedings, collections and special issues and written a monograph on Nonlinear Multiobjective Optimization. She is a member of the Finnish Academy of Science and Letters, Section of Science and has been the President of the International Society on Multiple Criteria Decision Making (MCDM). She belongs to the editorial boards of seven international journals and has received the Georg Cantor Award of the International Society on MCDM for developing innovative ideas. The Finnish Operations Research Society appointed her as the OR Person of the Year in 2023.

 

Some Views to Multiobjective Optimization with a Focus on Interactive Methods

Abstract: In various real decision problems, we must optimize several conflicting objective functions simultaneously. This means that we must solve multiobjective optimization problems. These problems have so-called Pareto optimal solutions representing different trade-offs and they that cannot be ordered mathematically without some additional information. Typically, we assume that a domain expert called a decision maker provides preference information to guide the solution process. By applying appropriate methods, we can find the best balance among the trade-offs.
In this talk, I classify multiobjective optimization methods based on the role of the decision maker and devote most attention to interactive methods, where the decision maker augments the problem formulation with domain expertise. The decision maker directs the iterative solution process with one’s preferences to find the most preferred solution. At the same time, the decision maker gains insight into the interdependencies and trade-offs among the conflicting objective functions and can get convinced of the quality of the most preferred solution. I demonstrate the advantages of applying interactive methods with some example problems. In addition, I give a brief overview of the modular, open-source software framework DESDEO containing different interactive methods

Stefan Røpke is Professor in Operations Research at the Technical University of Denmark. His main research interest is the development of advanced meta-heuristics and mathematical programming methods to solve hard optimization problems, primarily vehicle routing problems.  He is the co-inventor of the well known metaheuristic Adaptive Large Neighbourhood Search (ALNS). ALNS has proven to be a high-performance optimization algorithm, applied to a large number of hard optimization problems, including many scheduling problems. He has (co-)authored more than 40 journal articles about both heuristics and exact optimization approaches, applied for different applications mainly related to Vehicle Routing, but also regarding freight carriers, berth allocation, and technician scheduling.

 

Adaptive Large Neighborhood Search

Abstract: Adaptive Large Neighborhood Search (ALNS) is a metaheuristic that extends the Large Neighborhood Search heuristic (LNS) proposed by Paul Shaw. While traditional LNS employs a single method for destroying and repairing solutions iteratively, ALNS introduces multiple such methods. The algorithm keeps track of the performance of each method and attempts to utilize the best methods for the instance at hand. ALNS allows the user to incorporate domain-specific knowledge by adding tailored destroy and repair methods that can exploit the problem’s structure or even be targeted at a subset of the instances that need to be solved.

This talk briefly introduces the ALNS algorithm and explores applications to time-tabling problems. We discuss the relationship between ALNS and hyperheuristics and review efforts to parallelize ALNS. Additionally, we explore the integration of machine learning into ALNS, particularly focusing on enhancing the selection of destroy and repair methods.

Sigrid Knust is Professor of Combinatorial Optimization at the University of Osnabrück, Germany. Her main research interest is the development of efficient algorithms to solve complex combinatorial optimization problems, especially from the areas of scheduling, resource-constrained project scheduling, logistics, transportation, sports league scheduling, timetabling, shift scheduling. She has (co-)authored more than 60 journal articles, dealing with different theoretical and practical aspects of complex optimization problems.

 

Decomposition methods for sports scheduling problems

Abstract: Generating a sports league schedule is a challenging task due to the variety of different requirements which have to be addressed. The basic problem is to find a schedule for a single/double round robin tournament in which every team plays against each other team exactly once/twice, and every team plays one game per round. Additionally, several side constraints have to be respected, e.g., the avoidance of breaks (consecutive home/away games of a team), fairness issues (like opponent strengths, carry-over effects), the consideration of regions or wishes of teams and media.

This variety of specific problem settings has led to a multitude of alternative approaches (cf., e.g. [1] [2], [3], [4]). Due to its complexity, the problem is often solved by decomposition techniques, i.e., it is divided into different subproblems which are solved consecutively. In this talk, the following three approaches are discussed:

• In a “first-schedule, then-break” approach, in the first stage it is decided which teams play against each other in which round. Afterwards, in the second stage home-away patterns (with a minimum number of breaks) cor- responding to the pairings from the first stage are determined.
• In a “first-break, then-schedule” approach, at first home-away patterns are generated for the teams. Then, the subproblem of the second stage consists in finding a corresponding feasible schedule.
• In a “first assign modes, then schedule” approach, at first for each game a home team is fixed. In the second stage, all games are scheduled in these fixed modes taking into account additional constraints.

[1] Kendall, G., Knust, S., Ribeiro, C.C., Urrutia, S.: Scheduling in sports: An annotated bibliography. Computers & Operations Research 37, 1–19 (2010).
[2] Rasmussen,R.V.,Trick,M.A.:Round robin scheduling –a survey. EuropeanJournal of Operational Research 188, 617–636 (2008).
[3] Ribeiro, C.C.: Sports scheduling: problems and applications. International Transactions in Operational Research 19, 201–226 (2012).
[4] Van Bulck, D., Goossens, D., Schönberger, J., Guajardo, M.: RobinX: A three-field classification and unified data format for round-robin sports timetabling. European Journal of Operational Research 280, 568–580 (2020).

Nysret Musliu is Associate Professor at the Technical University of Vienna, Austria. His main research interest is in AI techniques, Combinatorial Optimization and Constraint Satisfaction Techniques, often applied to Scheduling problems. He has (co-)authored more than 20 articles, dealing with both different theoretical and practical problems when applying combinatorial optimization to scheduling problems.

 

AI Techniques for Timetabling and Scheduling Problems

Abstract: In this talk, we will first provide an overview of various AI-based methods proposed by our lab for solving problems in application domains such as employee timetabling and project scheduling. The topics covered will include solver-independent modelling, constraint programming, and hybrid techniques. In the second part of the talk, we will discuss methods that utilize machine learning techniques for automatic algorithm selection and heuristic algorithm design. We will also briefly present innovative decision support systems that incorporate our solution methods and an approach for preference explanation to guide decision-makers toward solutions that align with their expectations. The talk will conclude with a discussion of future challenges in the domain of scheduling and timetabling.

Kate Smith-Miles is a Melbourne Laureate Professor of Applied Mathematics, as well as Pro Vice-Chancellor (Research Capability), at the University of Melbourne, Australia. She is also Director of a doctoral training centre for Optimisation Technologies, Integrated Methodologies and Applications (OPTIMA, see optima.org.au). Kate obtained a B.Sc(Hons) in Mathematics and a Ph.D. in Electrical Engineering, both from The University of Melbourne. She has held Professorships in three disciplines (mathematics, engineering, information technology), and is involved in many interdisciplinary collaborations and industry partnerships. She has published 2 books on neural networks and data mining, and over 300 refereed publications in the areas of neural networks, optimisation, machine learning, and various applied mathematics topics. Her awards include the Australian Mathematical Society Medal in 2010 for distinguished research; the EO Tuck Medal from ANZIAM in 2017 for outstanding research and distinguished service; and the Ren Potts Medal for outstanding research in the theory and practice of operations research from the Australian Society for Operations Research in 2019. She is a Fellow of the Australian Academy of Science, a Fellow of the Australian Mathematical Society, and a past-President of the Australian Mathematical Society. She is frequently invited as keynote speaker at leading international conferences, including IFORS, GECCO, and CPAIOR, to discuss her Instance Space Analysis methodology.

 

Stress-testing algorithms via Instance Space Analysis

Abstract: Instance Space Analysis (ISA) is a recently developed methodology to support objective testing of algorithms. Rather than reporting algorithm performance on average across a chosen set of test problems, as is standard practice, ISA offers a more nuanced understanding via visualisation of the unique strengths and weaknesses of algorithms across different regions of the instance space that may otherwise be hidden on average. It also facilitates objective assessment of any bias in the chosen test instances, and provides guidance about the adequacy of benchmark test suites and the generation of more diverse and comprehensive test instances to span the instance space. This tutorial provides an overview of the ISA methodology, and the online software tools (seematilda.unimelb.edu.au) that are enabling its worldwide adoption in many disciplines. Several case studies from classical operations research problems will be presented to illustrate the methodology and tools, including  timetabling, travelling salesman problem, 0-1 knapsack; and applications to machine learning will also be highlighted.