GenAI and Heuristic Optimization

• Dr. Stefan Menzel, Honda Research Institute Europe, Germany. stefan.menzel@honda-ri.de
• Prof. Yew Soon Ong, Nanyang Technological University, Singapore. asysong@ntu.edu.sg
• Prof. Yaochu Jin, Westlake University, China. jinyaochu@westlake.edu.cn
• Prof. Michael T. M. Emmerich, University of Jyväskylä, Finland. michael.t.m.emmerich@jyu.fi
• Dr. Niki van Stein, Leiden University, the Netherlands. n.van.stein@liacs.leidenuniv.nl
• Dr. Anna Konovova, Leiden University, the Netherlands. a.kononova@liacs.leidenuniv.nl
• Prof. Thomas Bäck, Leiden University, the Netherlands. t.h.w.baeck@liacs.leidenuniv.nl
• Prof. Bernhard Sendhoff, Honda Research Institute Europe, Germany. bernhard.sendhoff@honda-ri.de

Generative AI and Large Language Models are groundbreaking technological innovations with large impact on contemporary science and engineering by generating context-sensitive text, knowledge-based answers, software code, images, music, and 3D assets from text prompts and image inputs. Trained on large datasets, these models conserve knowledge, identify hidden patterns, and reason across diverse data modalities. In evolutionary computation, generative models enable new ways of (a) mapping input types, such as text and images, to various output classes, such as text, numerical values or 2D/3D objects, and (b) developing evolutionary optimization methods from a fundamental new perspective. This allows for optimization tasks like prompt engineering, image preference optimization and design optimization, but furthermore the exploration of advanced evolutionary optimization methods, incorporating e.g. novel selection and mutation mechanisms for single, multi-objective and multi-task optimization reasoning on the knowledge of past optimization steps or additional data sources. They also facilitate data synthesis and the generation of missing data samples with evolutionary optimization guiding the search within computational budget constraints. This synergy has far-reaching implications for various disciplines, enabling innovative solutions and applications.

By combining the strengths of generative models and evolutionary computation, researchers can tackle complex problems and uncover new insights, leading to breakthroughs in fields like engineering, biology, finance, mobility and logistics. By interpreting and communicating optimization process and results through natural language they also improve the decision-making process. Potential applications of this synergy are vast, ranging from design optimization and decision support systems to predictive modeling and anomaly detection. As research in this area continues to advance, we expect to see significant improvements in the performance and synergy of generative models and evolutionary computation, leading to new opportunities for innovation and discovery. Moreover, the use of generative AI and Large Language Models in evolutionary computation can also lead to the development of more robust and adaptive systems, which are capable of handling uncertain and dynamic environments. The potential for interdisciplinary collaboration and knowledge sharing is huge, with opportunities for researchers and practitioners from diverse fields to come together and explore new applications and solutions. Overall, the integration of generative AI and Large Language Models with evolutionary computation offers new and innovative ways we approach complex problem solving and optimize systems, leading to significant advancements and innovations in the field of evolutionary computation and its applications in industries.

The topics of this special session include but are not limited to:

• Evolutionary optimization using LLMs
• LLM-assisted guided evolutionary search
• Evolutionary prompt engineering
• LLM-assisted evolutionary algorithm design
• LLM-assisted interactive evolutionary optimization
• Representations and generative models for evolutionary optimization
• Generative AI for real world applications
• Data synthesis for evolutionary computing
• Analyzing evolutionary optimization data with LLMs
• Benchmarks and quality metrics for LLMs in evolutionary optimization

Important Dates

31 January: Paper submission deadline (23:59, anywhere on Earth, i.e. UTC-12), WCCI 2026 Submission (external link)

Past Session

WCCI/CEC 2024 Special Session Link

Acknowledgements

This special session is supported by IEEE CIS Task Force on Transfer Learning & Transfer Optimization from ISATC.

This special session is partially supported by the European Union under Horizon Europe (GenAIDE, Grant Agreement No. 101226927). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Executive Agency (REA). Neither the European Union nor the granting authority can be held responsible for them.