Green-NAS: A Global-Scale Multi-Objective Neural Architecture Search for Robust and Efficient Edge-Native Weather Forecasting

TL;DR

Green-NAS presents a multi-objective neural architecture search framework optimized for low-resource environments, achieving high accuracy weather forecasting with minimal energy consumption and model size, emphasizing sustainability.

Contribution

The paper introduces Green-NAS, a novel multi-objective NAS method that balances accuracy and efficiency for weather forecasting, incorporating transfer learning for improved performance in data-scarce scenarios.

Findings

Green-NAS models achieve RMSE of 0.0988 with 153k parameters.

Green-NAS models are 239 times smaller than existing models like GraphCast.

Transfer learning improves forecasting accuracy by approximately 5.2%.

Abstract

We introduce Green-NAS, a multi-objective NAS (neural architecture search) framework designed for low-resource environments using weather forecasting as a case study. By adhering to 'Green AI' principles, the framework explicitly minimizes computational energy costs and carbon footprints, prioritizing sustainable deployment over raw computational scale. The Green-NAS architecture search method is optimized for both model accuracy and efficiency to find lightweight models with high accuracy and very few model parameters; this is accomplished through an optimization process that simultaneously optimizes multiple objectives. Our best-performing model, Green-NAS-A, achieved an RMSE of 0.0988 (i.e., within 1.4% of our manually tuned baseline) using only 153k model parameters, which is 239 times fewer than other globally applied weather forecasting models, such as GraphCast. In addition, we also describe how the use of transfer learning will improve the weather forecasting accuracy by approximately 5.2%, in comparison to a naive approach of training a new model for each city, when there is limited historical weather data available for that city.