Optimizing Evolutionary Algorithms with a Green Design Framework

Green artificial intelligence represents a new paradigm closely aligned with the green transition in the economy. The main aim is to minimize the computational processes’ carbon footprint while indirectly reducing these methods’ time complexity. This study explores approaches to reducing the complexity of evolutionary algorithms, mainly when deployed on resource-constrained hardware. Green evolution strategies are introduced and evaluated against compact differential evolution using the IEEE CEC’14 (Congress on Evolutionary Computation 2014) benchmark suite in this context. The results indicate a substantial reduction in solution quality; however, this trade-off is offset by a significant decrease in time complexity (up to 25 % less time complexity). These findings highlight the potential of green evolutionary algorithms and motivate further research.