In Optimization community, a broad spectrum of research directions have expanded the Learning-assisted Algorithm Design (LEAD) field as a vibrant scientific universe, including Learning to Optimize (L2O), Neural Combinatorial Optimization (NCO), Data-Driven Evolutionary Algorithm (DDEA), Meta-Black-Box Optimization (MetaBBO) and LLM-based Algorithm Design (LLM4AD). Given the novel methodologies and promising empirical/theoretical results from diverse LEAD directions, these novel techniques have shown powerful, often human-competitive performance on many realistic optimization scenarios, mitigating the tedious, human-in-the-loop development of algorithms. Consequently, LEAD represents a pivotal avenue for promoting the synergy between machine learning and optimization algorithms, contributing to broader progress in Artificial Intelligence.

Despite these successes, substantial challenges persist in each sub-field. More importantly, due to the fast development pace of LEAD works and the corresponding rapidly updated literature, there has seen a lack of a dedicated forum for researchers to debate, discuss and analyze up-to-date methods and results. Besides, research directions such as L2O, NCO, LLM4AD and MetaBBO often advance in parallel, sometimes developing distinct terminologies or overlooking synergistic opportunities. This fragmentation creates a substantial barrier to cohesive progress. To address this gap, we propose the first international workshop on Learning-assisted Algorithm Design (IJCAI-LEAD 2026). This event is designed to be a dedicated, interactive forum that moves beyond simple paper presentations to actively unify the community. The workshop will be organized in a hybrid (onsite and online) format (the workshop’s website will be made available shortly), guided by four concrete objectives:

• To identify and unify the shared foundational challenges and methodologies across diverse LEAD sub-fields.

• To highlight synergistic research opportunities between areas like surrogate modeling, reinforcement learning, and large language models.

• To serve as a platform to begin defining a coherent, community-driven research agenda for the LEAD field as a whole.

• To provide a supportive venue for sharing the latest discoveries and early-stage works, enabling timely discussion and dissemination of emerging ideas and results.