LiBOG: Lifelong Learning for Black-Box Optimizer Generation
This work addresses the need for adaptive optimizer generation in real-world scenarios where problem distributions shift over time, representing an incremental advancement in meta-optimization.
The paper tackles the problem of existing Meta-Black-Box Optimization methods requiring a stationary problem distribution by introducing LiBOG, a lifelong learning approach that continuously learns from sequentially encountered problems to generate high-performance optimizers, effectively addressing catastrophic forgetting and maintaining plasticity.
Meta-Black-Box Optimization (MetaBBO) garners attention due to its success in automating the configuration and generation of black-box optimizers, significantly reducing the human effort required for optimizer design and discovering optimizers with higher performance than classic human-designed optimizers. However, existing MetaBBO methods conduct one-off training under the assumption that a stationary problem distribution with extensive and representative training problem samples is pre-available. This assumption is often impractical in real-world scenarios, where diverse problems following shifting distribution continually arise. Consequently, there is a pressing need for methods that can continuously learn from new problems encountered on-the-fly and progressively enhance their capabilities. In this work, we explore a novel paradigm of lifelong learning in MetaBBO and introduce LiBOG, a novel approach designed to learn from sequentially encountered problems and generate high-performance optimizers for Black-Box Optimization (BBO). LiBOG consolidates knowledge both across tasks and within tasks to mitigate catastrophic forgetting. Extensive experiments demonstrate LiBOG's effectiveness in learning to generate high-performance optimizers in a lifelong learning manner, addressing catastrophic forgetting while maintaining plasticity to learn new tasks.