The integration of Large Language Models (LLMs) into the field of Evolutionary Computing (EC) is an emerging area of research that promises to enhance optimization processes, creativity, and problem-solving strategies in various domains. LLMs, such as GPT-4, have shown remarkable capabilities in understanding and generating human-like text, making them valuable tools for tasks that involve natural language processing, knowledge representation, and even creative ideation.

The EvoLLMs Special Session aims to explore the synergies between LLMs and EC, inviting novel contributions that apply evolutionary techniques to optimize, enhance or understand LLMs, as well as works that leverage LLMs to advance the field of EC. This session will provide a platform for researchers and practitioners to present their latest findings, share insights and discuss future directions at the intersection of these two rapidly evolving fields.

Topics of Interest

The combination of LLMs and Evolutionary Computing opens up new avenues for research that
can push the boundaries of both fields. Potential topics of interest include, but are not limited to:

• Evolutionary Prompt Engineering: Applying evolutionary algorithm to develop effective
  prompts that maximize the utility of LLMs in various applications, such as text generation,
  question answering and summarization.
• LLM-Guided Evolutionary Algorithms: Incorporating LLMs into evolutionary algorithms
  as components that guide the search process, provide domain knowledge or generate
  candidate solutions.
• Co-evolution of LLMs and EC Techniques: Exploring the co-evolution of LLMs and EC techniques, where both evolve in tandem to solve complex, multi-modal or multi-objective problems.
• Benchmarking and Comparative Studies: Evaluating the performance of LLM-
  integrated evolutionary approaches against traditional EC methods across different optimization problems and domains.
• LLMs for Automated Code Generation in EC: Utilizing LLMs to automatically generate
  or refine code for evolutionary algorithms, potentially reducing the development time and
  improving the adaptability of EC methods.
• Optimization of LLM Architectures: Using evolutionary algorithms to optimize the
  hyperparameters, architecture and training processes of LLMs to enhance their
  performance on specific tasks.
• Applications in Real-World Problems: Demonstrating the application of LLM-EC hybrid
  approaches in real-world scenarios, such as optimization in engineering, healthcare,
  finance and creative industries.

Organisers

• Niki van Stein
  Leiden University, Netherlands
  n.vanstein(at)leidenuniv.nl
• Thomas Bäck
  Leiden University, Netherlands
  t.h.w.baeck(at)liacs.leidenuniv.nl
• Anna V. Kononova
  Leiden University, Netherlands
  a.kononova(at)liacs.leidenuniv.nl