Learning-assisted algorithm design is becoming an important direction in evolutionary computation and optimization. Recent advances in machine learning, reinforcement learning, neural optimization, meta-learning, foundation models, and large language models are creating new opportunities to automate and improve the design of optimization algorithms. This workshop will focus on learning-assisted and data-driven methods for designing, configuring, and enhancing optimization algorithms. Topics include learning to optimize, neural combinatorial optimization, meta black-box optimization, large-language-model-based algorithm design, automated algorithm selection and configuration, data-driven evolutionary algorithms, surrogate-assisted optimization, and reinforcement-learning-assisted evolutionary computation. The workshop aims to bring together researchers from evolutionary computation, machine learning, operations research, and real-world optimization applications to discuss recent progress, open challenges, and future directions in learning-assisted algorithm design. The objectives and scope of the workshop are as follows: 

• Advance learning-driven algorithm design for optimization: The workshop will cover machine learning and reinforcement learning methods for algorithm configuration, operator selection, algorithm selection, parameter control, and adaptive search strategy design.

•  Promote automated algorithm design across emerging paradigms: Relevant topics include meta black-box optimization, large-language-model-assisted automated algorithm design, autoregressive learning for algorithm generation, hyper-heuristics, and pre-trained or foundational optimization models.

• Support data-driven optimization modeling: The workshop welcomes research on surrogate modeling in online and offline evolutionary systems, data-driven optimization modeling, large-language-model-assisted optimization modeling, and learning-based representations of optimization problems.

• Encourage cross-disciplinary integration: Topics include evolutionary reinforcement learning, large language models for foundational optimization systems, self-evolved agentic systems, and other hybrid approaches that connect evolutionary computation, machine learning, automated algorithm design, and intelligent agents.

• Build foundations for reliable evaluation and deployment: The workshop will include exploratory landscape analysis, fitness landscape analysis, instance space analysis, benchmark construction, tools, platforms, evaluation metrics, and performance assessment for learning-assisted optimization methods.