Marine Predators Algorithms for Global Optimization and Feature Selection

The Marine Predators Algorithm (MPA) is a nature-inspired metaheuristic that emulates the foraging strategies of marine predators to tackle complex optimisation tasks. It employs dynamic phases of exploration and exploitation, modelled on Lévy flight and Brownian motion movements, to navigate high-dimensional search spaces efficiently. In its standard form, MPA generates a population of candidate solutions that iteratively adapt through simulated predator–prey interactions, balancing global search with local refinement. Recent adaptations extend MPA to feature selection by mapping continuous position updates onto binary inclusion or exclusion decisions. This hybridised approach reduces dimensionality and enhances classification or regression performance by focusing on the most informative variables. Applications span engineering design, communication network planning, power allocation and data mining, where MPA variants have demonstrated competitiveness against established optimisers. By integrating novel operators—such as local escaping mechanisms, nonlinear transition functions and chaotic maps—researchers have further improved convergence speed, diversity maintenance and robustness against local optima. The global significance of these developments lies in providing versatile, general-purpose tools for real-world problems characterised by non-convexity, multimodality and high dimensionality.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

An enhanced MPA incorporating a Local Escaping Operator has been proposed to address premature convergence. By selectively replacing the weakest solutions with locally generated candidates, this variant improves interaction among population members and prevents stagnation, yielding superior performance on standard benchmark suites and engineering case studies.

A Nonlinear Marine Predator Algorithm introduces adjustable transition functions to refine the shift from global exploration to local exploitation. This cost-effective adaptation has proven resilient in power allocation scenarios for next-generation communication systems, delivering fair user resource distribution while maintaining rapid convergence.

For feature selection tasks, a Chaos-Embed MPA employs chaotic maps to guide binary solution updates, optimising both the number of features and classification accuracy. Statistical tests on benchmark datasets confirm its ability to identify minimal yet informative feature subsets, outperforming several established selection techniques.

Marine Predators Algorithms for Global Optimization and Feature Selection publication trend

The graph below shows the total number of articles in marine predators algorithms for global optimization and feature selection across all publications each year (not limited to Nature Index journals).

Technical terms

Marine Predators Algorithm (MPA): A population-based metaheuristic inspired by the hunting patterns of marine predators, combining exploration and exploitation via stochastic movements.

Exploration: Phase in which candidate solutions broadly survey the search space to discover promising regions.

Exploitation: Phase in which candidate solutions intensively search around known good regions to refine optima.

Lévy flight: Random walk characterised by occasional large jumps, enhancing exploration by escaping local traps.

Brownian motion: Random walk with small, incremental steps, supporting fine-grained local search.

Feature selection: Process of choosing a subset of relevant variables from data to improve model performance and interpretability.

Metaheuristic algorithm: High-level strategy that guides subordinate heuristics to efficiently explore large and complex solution spaces.

References

1. Enhanced Marine Predators Algorithm with Local Escaping Operator for Global Optimization. Knowledge-Based Systems (2021).
2. Nonlinear marine predator algorithm: A cost-effective optimizer for fair power allocation in NOMA-VLC-B5G networks. Expert Systems with Applications (2022).
3. Chaos Embed Marine Predator (CMPA) Algorithm for Feature Selection. Mathematics (2022).