PlatMetaX: An Integrated MATLAB platform for Meta-Black-Box Optimization

PlatMetaX: An Integrated MATLAB platform for Meta-Black-Box Optimization Abstract--The landscape of optimization problems has become increasingly complex, necessitating the development of advanced optimization techniques. Meta-Black-Box Optimization (MetaBBO), which involves refining the optimization algorithms themselves via meta-learning, has emerged as a promising approach. Recognizing the limitations in existing platforms, we presents PlatMetaX, a novel MA TLAB platform for MetaBBO with reinforcement learning. The platform is designed to handle a wide range of optimization problems, from single-objective to multi-objective, and is equipped with a rich set of baseline algorithms and evaluation metrics. W e demonstrate the utility of Plat-MetaX through extensive experiments and provide insights into its design and implementation. PlatMetaX is available at: https://github.com/Y Index Terms --evolutionary optimization, meta-black-box optimization, reinforcement learning, automated algorithm design, PlatEMO, MetaBox I. Introduction Optimization problems are fundamental to numerous fields, including engineering, economics, and artificial intelligence. As the complexity of these problems grows, traditional optimization methods often fall short, particularly when faced with high-dimensional, non-linear, multi-objective, or dynamic optimization landscapes [1]- [3]. This has led to a surge in interest in meta-optimization, which involves optimizing the parameters of optimization algorithms to improve their performance across a variety of problems [4], [5]. With the developing of meta-optimization, Meta-Black-Box Optimization (MetaBBO) emerges as a novel and effective framework, which employs meta-learning to design black-box optimizers automatically [6]. It is a sophisticated framework that uses meta-learning to improve the effectiveness of traditional Black-box Optimization (BBO) methods, which inspired by the concept of "meta" to highlight its ability to adapt optimization strategies based on past experiences, much like how meta-learning algorithms adapt over time. It is done within a two-level optimization framework Figure 1, where the optimizer used to optimize the black-box optimizers in the meta-level is termed meta-optimizer, while the black-box optimizers used to solve the specific problem is termed base-optimizer.