Table 2 Distinctive features of SFO compared with swarm-based algorithms.
From: Swift Flight Optimizer: a novel bio-inspired optimization algorithm based on swift bird behavior
Algorithm | Inspiration | Core mechanism | Exploration strategy | Exploitation strategy | Distinctive features |
|---|---|---|---|---|---|
PSO | Bird flocking behavior | Velocity–position update guided by personal and global bests | Random coefficients and inertia weight promote global search | Cognitive and social terms refine convergence | Simple structure, widely used but prone to premature convergence |
ACO | Ant foraging and pheromone trails | Probabilistic path construction based on pheromone intensity | Exploration via pheromone evaporation | Exploitation through reinforcement of best paths | Strong combinatorial optimization, weaker in continuous domains |
ABC | Foraging of honey bees | Employed, onlooker, and scout phases | Scouts introduce random search | Onlooker bees exploit best nectar sources | Efficient balance, but slower convergence in high dimensions |
GWO | Grey wolf hunting hierarchy | Encircling and attacking prey guided by alpha, beta, delta wolves | Diversification through coefficient vectors | Intensification near best wolves | Competitive performance with simplicity |
WOA | Bubble-net hunting of whales | Spiral updating and encircling mechanism | Shrinking encircling and random movement | Exploitation via spiral bubble-net attack | Strong intensification, weaker diversity maintenance |
GPSOM | Group-based swarm strategies | Division into three sub-swarms (exploration, exploitation, balance) | Sub-swarm 1 enhances diversity via dynamic operators | Sub-swarm 2 and 3 refine search with adaptive sine–cosine and positional adjustment | Proven effectiveness on CEC benchmarks and engineering tasks |
EMBGO | Multiplayer battle royale dynamics | Integrated movement–battle phases with greedy selection | Differential mutation and Lévy flight enhance diversity | Battle phase exploits stronger opponents and safe zone convergence | Robust across CEC2017–2022, also applied to adversarial neural architecture search |
QSHO | Quantum-enhanced spotted hyena hunting | Incorporates quantum computing into SHO model | Quantum states expand search space and prevent trapping | Encircling and attacking prey with enhanced convergence | Outperforms classical SHO and variants on CEC2013/2017 and real-world engineering tasks |
SFO (Proposed) | Swift birds’ adaptive flight | Multi-mode structure: glide, target, micro + stagnation reset | Glide mode ensures broad coverage | Target and micro modes refine local optima | First swift-inspired optimizer; integrates explicit stagnation-aware reinitialization |
