Abstract
The rapid rise of Large Language Models (LLMs) has profoundly reshaped artificial intelligence research and practice, extending beyond natural language processing to influence reasoning, decision-making, and optimization. In parallel, Evolutionary Computation (EC) continues to advance as a powerful global optimization framework characterized by adaptability, scalability, and robustness. The convergence of these two paradigms has given birth to a new frontier in computational intelligence, where human-like reasoning meets large-scale evolutionary search.
This Special Session aims to provide a dedicated forum for exploring the bidirectional synergy between LLMs and EC, spanning theoretical foundations, algorithmic innovation, and real-world applications. From one direction, LLM-enhanced EC leverages the rich knowledge, generative capability, and reasoning skills of LLMs to drive new forms of intelligent evolutionary operators, algorithm generation, and explainable optimization. From the other direction, EC-enhanced LLM employs evolutionary search to optimize prompts, architectures, and model behaviors in closed-box or multi-objective settings, thereby improving efficiency, interpretability, and adaptivity. Together, this interplay opens up promising pathways for developing next-generation AI systems that integrate language understanding, optimization, and self-improvement.
Topics of Interest
- LLM-guided evolutionary optimization and algorithm generation
- Evolutionary prompt engineering and architecture search for LLMs
- Integration of EC and LLMs in AutoML and General-purpose AI, such as neural architecture search, algorithm selection, algorithm design, and reinforcement learning
- Evolutionary reasoning and decision-making with foundation models
- Benchmarking, explainability, and evaluation of hybrid EC–LLM systems
- Applications in scientific discovery, robotics, software engineering, and multimodal intelligence
This topic aligns closely with the core themes of IEEE CEC 2026 on Evolutionary Computation for Emerging AI Paradigms. The session aims to bring together researchers from evolutionary computation, machine learning, and large-model research communities to foster interdisciplinary exchange. By facilitating open discussions and showcasing state-of-the-art advances, this session seeks to define the roadmap for computational intelligence in the era of foundation models, inspiring the next wave of hybrid and autonomous intelligent systems.
Organizers
Postdoctoral Fellow, Department of Data Science and Artificial Intelligence, The Hong Kong Polytechnic University, Hong Kong SAR, China
Homepage: https://wuxingyu-ai.github.io/
Email: xingy.wu@polyu.edu.hk
Professor, School of Computer Science and Engineering, South China University of Technology (SCUT), China
Homepage: https://yuejiaogong.github.io/
Email: gongyuejiao@gmail.com
Professor, College of Computer Science, Chongqing University, China
Homepage: https://cs.cqu.edu.cn/info/1491/5633.htm
Email: liangf@cqu.edu.cn
Chair Professor and Founding Head, Department of Data Science and Artificial Intelligence, The Hong Kong Polytechnic University, Hong Kong SAR, China
Homepage: https://www.polyu.edu.hk/dsai/people/academic-staff/tankaychen/
Email: kctan@polyu.edu.hk
Primary Contact
Dr. Xingyu Wu
Email: xingy.wu@polyu.edu.hk
Website: https://wuxingyu-ai.github.io/
Session Webpage: https://wuxingyu-ai.github.io/LLM4EC/
Additional Activities
-
Mini Panel: “Will Evolution Outthink Language Models?” (30 min)
A lively, moderated discussion among invited experts and organizers on whether evolution-inspired optimization can drive LLMs toward self-evolving intelligence.
Format: 5-minute position statements + 20-minute roundtable debate + 5-minute audience poll (via live QR voting).
Expected Outcome: Identify key challenges and future directions for co-evolution of EC and LLMs. -
Open Co-Creation Roundtable: “Designing the EC–LLM Synergy Roadmap” (30 min)
Participants collaboratively contribute ideas toward a community roadmap.
Format: Audience members propose one “grand challenge” or “research question” via digital board; organizers synthesize and summarize themes in real time.
Expected Outcome: A post-session summary published on the Special Session website as the “WCCI 2026 EC–LLM Roadmap Snapshot.”
Together, these activities will transform the session from a traditional presentation block into an interactive exploration hub, inspiring new collaborations and positioning the EC+LLM synergy as a key theme within the computational intelligence community.
Related Resources
- Contact organiser: https://wuxingyu-ai.github.io/
- TEVC 2025 survey: https://ieeexplore.ieee.org/document/10767756
- Research paper and code repository: https://github.com/wuxingyu-ai/LLM4EC
- IEEE CIS Task Force: IEEE CIS Task Force on LLMs and Computational Intelligence for GPAIS