CMDO: A Cognitive Memory-Driven Optimization Algorithm for Adaptive Population-Based Search

Problem

The paper addresses a gap in existing population-based optimization methods regarding the retention of contextual information related to search behavior success or failure. This limitation hinders the ability of these algorithms to adaptively refine their search strategies based on past experiences. The work is presented as a preprint and has not undergone peer review.

Method

The authors propose the Cognitive Memory-Driven Optimization (CMDO) algorithm, which integrates three types of memory: working memory, episodic memory, and consolidated memory. These memory types facilitate the representation of experiences by capturing the relationship between search context, search behavior, and observed outcomes. CMDO employs a combination of exploratory, directed, and local search behaviors, allowing for an adaptive search geometry that can adjust based on the accumulated knowledge from previous search iterations. The algorithm is evaluated using the Blackbox Optimization Benchmarking test suite on COCO (BBOB/COCO) and CEC2017 problems. It is compared against established optimization algorithms, including Differential Evolution (DE), Covariance Matrix Adaptation Evolution Strategy (CMA-ES), Success-History Based Adaptive Differential Evolution (SHADE), Grey Wolf Optimization (GWO), Harris Hawks Optimization (HHO), and Optimized Randomized Cooperative Algorithm (ORCA).

Results

CMDO demonstrates competitive optimization performance, achieving the lowest median error on the CEC2017 F10 benchmark when compared to DE, CMA-ES, SHADE, GWO, HHO, and ORCA. The performance is noted to be problem-dependent, indicating that CMDO may excel in certain scenarios while remaining competitive in others. However, the available text does not report quantitative results beyond the median error comparison.

Limitations

The authors do not report any limitations in their work. However, the lack of extensive empirical validation across a broader range of benchmarks and problem types could be considered a potential limitation, as it may affect the generalizability of the findings.

Why it matters

The introduction of CMDO has significant implications for the field of optimization, particularly in enhancing the adaptability of population-based search algorithms. By incorporating cognitive memory mechanisms, CMDO could lead to more efficient search strategies that better leverage past experiences, potentially improving performance in complex optimization tasks. This work opens avenues for further research into memory-driven approaches in optimization, encouraging the exploration of cognitive-inspired mechanisms in algorithm design.