Trading Carbon for Physics: On the Resource Efficiency of Machine Learning for Spatio-Temporal Forecasting

2025-09-29

AI Summary

Contemporary deep learning research overemphasizes predictive accuracy, leading to oversized models, escalating computational costs, and substantial carbon emissions—particularly detrimental for physics-constrained spatiotemporal forecasting. Method: We propose a principled modeling paradigm that systematically incorporates physical inductive biases—such as conservation laws and symmetries—into both model architecture and training objectives. Our approach integrates physics-informed neural networks with flow matching to construct differentiable, interpretable, and physically consistent predictors. Contribution/Results: Evaluated on multiple real-world spatiotemporal forecasting benchmarks, our method matches or exceeds state-of-the-art accuracy while drastically reducing parameter count, inference latency, and energy consumption—achieving an average carbon footprint reduction of over 40%. This work establishes generalizable design principles and empirical evidence for energy-efficient, environmentally sustainable scientific machine learning.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Development of modern deep learning methods has been driven primarily by the push for improving model efficacy (accuracy metrics). This sole focus on efficacy has steered development of large-scale models that require massive resources, and results in considerable carbon footprint across the model life-cycle. In this work, we explore how physics inductive biases can offer useful trade-offs between model efficacy and model efficiency (compute, energy, and carbon). We study a variety of models for spatio-temporal forecasting, a task governed by physical laws and well-suited for exploring different levels of physics inductive bias. We show that embedding physics inductive biases into the model design can yield substantial efficiency gains while retaining or even improving efficacy for the tasks under consideration. In addition to using standard physics-informed spatio-temporal models, we demonstrate the usefulness of more recent models like flow matching as a general purpose method for spatio-temporal forecasting. Our experiments show that incorporating physics inductive biases offer a principled way to improve the efficiency and reduce the carbon footprint of machine learning models. We argue that model efficiency, along with model efficacy, should become a core consideration driving machine learning model development and deployment.

Problem

• Addressing high carbon footprint in large-scale spatio-temporal forecasting models
• Exploring physics inductive biases for efficiency-accuracy trade-offs
• Reducing computational resource requirements while maintaining model performance

Innovation

• Physics inductive biases improve model efficiency
• Flow matching enables spatio-temporal forecasting
• Balancing efficacy and efficiency reduces carbon footprint