One Search Fits All: Pareto-Optimal Eco-Friendly Model Selection
By: Filippo Betello , Antonio Purificato , Vittoria Vineis and more
Potential Business Impact:
Finds smart computer brains that use less power.
The environmental impact of Artificial Intelligence (AI) is emerging as a significant global concern, particularly regarding model training. In this paper, we introduce GREEN (Guided Recommendations of Energy-Efficient Networks), a novel, inference-time approach for recommending Pareto-optimal AI model configurations that optimize validation performance and energy consumption across diverse AI domains and tasks. Our approach directly addresses the limitations of current eco-efficient neural architecture search methods, which are often restricted to specific architectures or tasks. Central to this work is EcoTaskSet, a dataset comprising training dynamics from over 1767 experiments across computer vision, natural language processing, and recommendation systems using both widely used and cutting-edge architectures. Leveraging this dataset and a prediction model, our approach demonstrates effectiveness in selecting the best model configuration based on user preferences. Experimental results show that our method successfully identifies energy-efficient configurations while ensuring competitive performance.
Similar Papers
Choosing to Be Green: Advancing Green AI via Dynamic Model Selection
Computers and Society
AI uses less power, stays almost as smart.
Small is Sufficient: Reducing the World AI Energy Consumption Through Model Selection
Computers and Society
Saves energy by using smaller AI models.
Energy-Efficient Green AI Architectures for Circular Economies Through Multi-Layered Sustainable Resource Optimization Framework
Machine Learning (CS)
Saves energy and resources by recycling smarter.