ROAR-NET COST Action

⌛ Less than two days left to apply to participate in the ROAR-NET Problem Modelling Code Fest. Act now! 🔗 https://lnkd.in/gct4gF8s The event will be hosted online at the Departamento de Engenharia Informática - Universidade de Coimbra and physically at 19 hubs organised by the ROAR-NET COST Action community around the world: Florenc Skuka, at EPOKA University, Albania Marcus Gallagher, at The University of Queensland, Australia Roberto Maria Rosati, at WU Vienna University of Economics and Business, Austria Pablo Munhoz, at Federal University of Ouro Preto, Brazil Domagoj Jakobović, at University of Zagreb, Faculty of Electrical Engineering and Computing, Croatia David Sedlák, at Brno University of Technology, Faculty of Information Technology, Czechia Marco Chiarandini, at University of Southern Denmark, Denmark Mohamed ElWakil, at Lab of Industrial Engineering, Faculty of Engineering, Tanta University, Egypt Diederick Vermetten, at Sorbonne University, France Bernhard Berger, at Data Science Center Bremen, Germany Dr. Michael Heider, at University of Augsburg, Germany Luca Di Gaspero, at University of Udine, Italy Hernan E. Aguirre, at Shinshu University, Japan Wassila Mtalaa, at Luxembourg Institute of Science and Technology, Luxembourg Gonçalo Lopes, at University of Coimbra, Portugal Pedro Ferreira, at University of Lisbon, Portugal Tome Eftimov, at Jožef Stefan Institute, Slovenia Gara Miranda Valladares, at Universidad de La Laguna, Spain Işıl Öz, at Izmir Institute of Technology, Türkiye Whether or not you will join a hub, we look forward to receiving your team's application. #ROARNET #CodeFest #OpenSource #Optimization #OperationsResearch #COSTactions

🗓️ Call for Participation: ROAR-NET Problem Modelling Code Fest The ROAR-NET Problem Modelling Code Fest is just one month away. Gather your team now! 🌍 Participate online or join one of the many local hubs worldwide 📍 From 18:00 CEST on 9 September to 18:00 CEST on 11 September 2025 💡 Why join? - Learn how to use the ROAR-NET API to model combinatorial optimisation problems - Solve modelling challenges and contribute to open-source tools - Earn achievement badges 🏅 and receive a certificate of participation - Collaborate with peers in the ROAR-NET community 💻 Two tracks available: 🔹 Training Track – if you are new to ROAR-NET or even to optimisation 🔹 Development Track – for contributing new models, instances, or tools No hub in your region? Consider organising your own! 📬 Deadline for team applications: 2 September 2025 (AOE) 📬 Deadline for hub proposals: 26 August 2025 (AOE) 🔗 Info and registration: https://lnkd.in/giT8-vYh Get ready for 48 hours of creativity, code, and collaboration! #ROARNET #CodeFest #Optimization #OperationsResearch #OpenSource #COSTactions #Hackathon

📢 ROAR-NET COST Action is launching a Survey on Visualisation of Optimisation Results. 👉 We are looking for domain experts or consultants who work with optimisation problems and have experience in visualisation techniques. If this applies to you, please participate in this survey. The insights gathered will contribute to a better understanding of the current trends and user needs concerning visualisation in optimisation practice. 🗓️ Deadline: 15 September 2025 🔗 https://lnkd.in/eK7sPRxg The survey is conducted by Working Group 1 on Problem Modelling and User Experience, lead by Luca Di Gaspero and Teresa Montrone. This Working Group aims at making the deployment of Randomised Optimisation Algorithms easier and more accessible to end users in a wide variety of applications. #optimization #visualization #COSTActions #ROARNET #userneeds

#GECCO2025 is over, and we are proud to have contributed to this year’s program with several papers acknowledging ROAR-NET COST Action, including a #BestPaperAward! 🏆 Special thanks to all the authors and attendees who contributed to the discussions and sessions. Below is a short recap of the four full papers: 🏆 Best Paper Award 🔹 On the Importance of Reward Design in Reinforcement Learning-Based Dynamic Algorithm Configuration: A Case Study on OneMax with (1+(λ,λ))-GA  Tai Nguyen, Phong Le, André Biedenkapp, Carola Doerr, Nguyen Dang The paper tackles Dynamic Algorithm Configuration with Reinforcement Learning (RL), emphasising the crucial role of reward design. Using the (1+(λ,λ))-GA on OneMax as a case study, the authors show that poor rewards can lead to learning failures. They propose a reward shaping strategy that improves exploration and scalability, demonstrating RL's potential for adaptive evolutionary algorithm control. 🔹 A Path-Relinking-Based Heuristic for the Multiobjective Subgraph Problem Daniela Scherer dos Santos, Kathrin Klamroth, Pedro Coimbra Martins, Luis Paquete The paper addresses a bi-objective graph problem with very nice mathematical properties that can be effectively exploited by a path-relinking approach. This is a follow-up of their work recently published in EJOR:  https://lnkd.in/dbwRrDCw. 🔹 Interpretable Non-linear Survival Analysis with Evolutionary Symbolic Regression Luigi Rovito, Marco Virgolin, PhD  The paper addresses survival regression, a bi-objective task where accuracy and interpretability must be balanced. The authors propose a symbolic regression approach tailored to survival analysis, aiming to discover non-linear but interpretable models. The resulting models provide high predictive performance while remaining mathematically understandable. 🔹 On the Pareto Set and Front of Multiobjective Spherical Functions with Convex Constraints Anne Auger, Dimo Brockhoff, Jordan Cork, Tea Tušar The paper studies a fundamental class of constrained multiobjective problems where objectives are spherical functions and constraints are convex. Using the projection theorem, the authors prove that the constrained Pareto set is the orthogonal projection of the unconstrained one onto the feasible region. They illustrate this geometric property with visualizations and evaluate NSGA-II’s ability to approximate these Pareto sets in different dimensions. The work provides valuable benchmarks for understanding algorithm behavior in constrained multiobjective optimization. 🔗 We’re proud to see ROAR-NET COST Action supporting high-quality, collaborative research in randomised optimisation and operations research. If you’re interested in these fields, consider joining ROAR-NET COST Action! #ROARNET #EvolutionaryAlgorithms #OperationsResearch

🚀ROAR-NET is at #GECCO2025! This year’s programme includes 4 companion papers (presented today!) and 4 full papers acknowledging ROAR-NET COST Action. Do not miss the next presentations! Brockhoff, D., & Villain, T. Benchmarking Powell’s Legacy: Performance of Five Derivative-Free Solvers in pdfo on the bbob Test Suite. https://lnkd.in/eEnpKAgm Workshop on Black Box Optimization Benchmarking 2025 Tuesday, July 15, 11:40–13:30, Azul Brockhoff, D., & Villain, T. How Robust is UOBYQA to Worsening, Frozen Noise? Investigations on the bbob Test Suite With Outliers. https://lnkd.in/eJsZBmMJ Workshop on Black Box Optimization Benchmarking 2025 Tuesday, July 15, 11:40–13:30, Azul Santos, F.J.J.B., De Lorenzo, A., Manzoni, L., & Pietropolli, G. Learning the Particle Swarm Optimization Velocity Update via Genetic Programming. https://lnkd.in/e7W3EhuZ 15th Workshop on Evolutionary Computation for the Automated Design of Algorithms Tuesday, July 15, 11:40–13:30, Alborán Sakallioglu, B., Nadizar, G., Manzoni, L., & Medvet, E. Evolving Typed Token Processing Networks. https://lnkd.in/e8ZeJn3b Graph-based Genetic Programming Tuesday, July 15, 11:40–13:30, Minotauro Scherer dos Santos, D., Klamroth, K., Martins, P., & Paquete, L. A Path-Relinking-based Heuristic for the Multiobjective Subgraph Problem. https://lnkd.in/eiXAH385 ECOM 1 Wednesday, July 16, 15:30–16:50 Rovito, L. & Virgolin, M. Interpretable Non-linear Survival Analysis with Evolutionary Symbolic Regression. https://lnkd.in/egEmge6b  EML 1 Wednesday, July 16, 12:00–13:20 Auger, A., Cork, J.N., Brockhoff, D. & Tušar, T. On the Pareto Set and Front of Multiobjective Spherical Functions with Convex Constraints. https://lnkd.in/eTnW2GCW EMO 3 Thursday, July 17, 12:00–13:20 Nguyen, T., Le, P., Biedenkapp, A., Doerr, C., & Dang, N. On the Importance of Reward Design in Reinforcement Learning-based Dynamic Algorithm Configuration: A Case Study on OneMax with (1+(𝜆,𝜆))-GA. https://lnkd.in/eghaf-hj Learning for Evolutionary Computation Thursday, July 17, 16:10–16:30 To find out more, look out for the ROAR-NET roll up in the conference registration area, pick up a flyer, or visit https://roar-net.eu. #EvolutionaryComputation #OperationsResearch #COSTActions #ROARNET

📢 Two presentations at EURO 2025 supported by ROAR-NET We are pleased to announce that two of our members presented their work at #EURO2025Leeds with the support of Young Researcher and Innovator Conference Grants awarded by ROAR-NET COST Action. Congratulations! 🔹 Özlem Karsu' s study on “Using Equitable Optimization for the Hazmat Transport Network Design Problem” frames the problem of designing a Hazardous Material Transport Network as a multi-objective equitable optimization problem, aiming to minimize risk exposure across different neighborhoods. The resulting multi-objective mixed-integer linear programming model is solved to offer policymakers a range of equitably nondominated solutions, each balancing efficiency (total risk) and fairness (distribution of risk). 🔹 Çağrı Koç's research on “A Metaheuristic for the Location-Routing Problem with Simultaneous Pickup and Delivery” tackles the design of a distribution network that serves customers with both pickup and delivery needs. The problem is formulated as a mixed-integer linear program and solved using a commercial solver. To handle larger instances effectively, a metaheuristic based on Adaptive Large Neighborhood Search is developed. Extensive computational experiments on generated instances help characterize the problem and assess solution performance. For more information on ROAR-NET funding opportunities, check out https://roar-net.eu/calls/. #Optimization #OperationsResearch #ROARNET #COSTactions

🎉 Congratulations to Ahmad Othman and Marco Chiarandini on winning the best open-source software prize 🏆 and making it to second place overall in the Integrated Healthcare Timetabling Competition 2024 (https://lnkd.in/eBQKrPCt)! The award was presented to them at EURO 2025. The tackled optimisation problem integrates three critical aspects of healthcare operations: - Patient admission scheduling - Nurse-to-room assignment - Surgical case planning Their submission was a local-search-based meta-heuristic algorithm implemented in Python and C++ following the ROAR-NET API specification (https://lnkd.in/eWAGAEqD). It performed remarkably well within the imposed time limits, and is but one indication of the competitiveness of randomised optimisation algorithms in practice! 👏 Well done, Ahmad and Marco! #EURO2025Leeds #ROARNET #OperationsResearch #Optimization #COSTactions

🎁 First ROAR-NET Training School Wrap Up The Training School has ended, but it will be remembered for years to come by Trainees and Trainers alike! Special thanks to COST Association - European Cooperation in Science and Technology for the financial support, Wassila Mtalaa and the Luxembourg Institute of Science and Technology (LIST) for hosting the event, our Software Development Coordinator Alexandre Jesus and all Trainers for their commitment, and the Trainees for making it all worthwhile. We couldn't have asked for more! Thursday and Friday Highlights 🎓 Lectures Distance-Based Search Operators – Alberto Moraglio Multiobjective Optimisation – Andreia Guerreiro (online) A Share-a-Ride Problem – Wassila Mtalaa Problem Structuring – Pieter Smet 🛠️ Group work In addition to the modelling projects, each team of 6 Trainees was challenged by Pieter Smet to formulate an optimisation problem from a verbal description of one of the following application cases: - District heating and cooling design - Event organisation - Hospital management 🎳 Team building On Thursday evening, there was still the opportunity to celebrate striking successes and endure spectacular failures at the bowling alley before heading for dinner together at a local restaurant. Everyone had a great time! 🗨️ Project presentations by the Trainees The Training School ended with the presentation of the Trainees' modelling projects and problem-structuring work. It was impressive to see how much they were able to achieve in such a short amount of time. We hope to see everyone again in the upcoming ROAR-NET Problem Modelling Code Fest (https://lnkd.in/gZnum88c)! #ROARNET #TrainingSchool #OperationsResearch #Optimization #COSTactions

📅 Programme Highlights for the First ROAR-NET Training School As we move through the week at the Luxembourg Institute of Science and Technology (LIST), Trainees are exploring the modelling of combinatorial optimisation problems with the ROAR-NET API in Python (https://lnkd.in/giSctUVa) through lectures and projects. Tuesday and Wednesday Highlights 🎓 Lectures Constructive Search – Marco Chiarandini Local Search – Francesca Da Ros Software Development for Reuse – Sandra Greiner 🪧 Poster presentations by the Trainees 🛠️ Group projects Each team of 6 Trainees is actively working on a different optimisation problem under the guidance of a dedicated mentor. Find the problem statements on https://lnkd.in/g-7Dh-Ti: - Salt Spreading - Home Healthcare Routing and Scheduling - Auto-Carrier Loading - Level Scheduling - Seating Arrangements 🏭 Guided tour On Wednesday evening, all participants had the chance to learn about the cultural and industrial history of Belval with a visit to the iconic Blast Furnace Belval, a symbol of Luxembourg’s steelmaking legacy and now part of a vibrant research and innovation hub. Stand by for more updates, including photos and achievements. #ROARNET #TrainingSchool #OperationsResearch #Optimization #COSTactions