M$^2$FedAQI: Multimodal Federated Learning for Air Quality Prediction on Heterogeneous Edge Devices

2026-05-19 · Source: cs.LG updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Internet of Things (IoT) & Connected Devices, Data Science & Analytics · Depth: Expert, quick

Summary

M^2FedAQI is a novel multimodal federated learning framework designed for decentralized Air Quality Index (AQI) prediction on heterogeneous edge devices. Submitted on May 10, 2026, this lightweight framework addresses the scalability, privacy, and communication overhead challenges of centralized and unimodal federated learning approaches in IoT environments. It integrates visual and tabular data modalities using a feature modulation-based fusion mechanism to enable efficient cross-modal interaction with low computational overhead. Evaluated on PM25Vision and TRAQID datasets for both classification and regression, M^2FedAQI significantly outperforms existing baselines, achieving up to 11.0% higher Accuracy, 3.53% higher AUC, 12.2% higher F1-score, and 18.0% higher R^2, while reducing MAE by up to 25.4% and RMSE by up to 20.4%. The framework also incorporates TLS-based authentication for secure client participation and efficient resource utilization on edge devices.

Key takeaway

For research scientists developing environmental monitoring systems, M^2FedAQI demonstrates a robust approach to air quality prediction. You should consider adopting multimodal federated learning architectures to improve prediction accuracy and ensure data privacy on distributed edge networks, especially when dealing with diverse data sources and heterogeneous devices.

Key insights

M^2FedAQI offers a multimodal federated learning solution for accurate, private, and efficient air quality prediction on edge devices.

Principles

• Multimodal data fusion enhances environmental pattern capture.
• Federated learning improves privacy and scalability.
• Edge deployment requires efficient resource utilization.

Method

M^2FedAQI integrates visual and tabular modalities via a feature modulation fusion mechanism, enabling decentralized AQI prediction on heterogeneous edge devices with TLS-based authentication.

In practice

• Deploy multimodal models on IoT edge devices.
• Use TLS for secure federated learning communication.
• Evaluate models with PM25Vision and TRAQID datasets.

Topics

• M$^2$FedAQI
• Multimodal Federated Learning
• Air Quality Prediction
• Heterogeneous Edge Devices
• Feature Modulation

Best for: Research Scientist, AI Scientist, Machine Learning Engineer, AI Hardware Engineer

Related on AIssential

• Federated orthogonal learning for detection of liver lesions from multi-phase contrast-enhanced CT images — FedOG enhances federated learning for medical image segmentation by using orthogonal gradient decomposition to counter data heterogeneity and incomplete phases.
• Quantization Impact on the Accuracy and Communication Efficiency Trade-off in Federated Learning for Aerospace Predictive Maintenance — INT4 quantization offers FP32-equivalent accuracy with $8\times$ communication savings in federated learning for aerospace predictive maintenance.
• FedMTFI: Feature Importance Based Optimized Multi Teacher Knowledge Distillation in Heterogeneous Federated Learning Environment — FedMTFI improves heterogeneous federated learning by combining multi-teacher knowledge distillation with SHAP-based feature importance for enhanced accuracy and interpretability.
• Boosting Multimodal Federated Learning via Chained Modality Optimization — FedMChain mitigates modality competition in MMFL via phased optimization and sign-guided aggregation for improved performance and efficiency.
• Statistical Limits and Efficient Algorithms for Differentially Private Federated Learning — New methods improve federated learning accuracy and communication efficiency while preserving differential privacy.
• VARS-FL: Validation-Aligned Client Selection for Non-IID Federated Learning in IoT Systems — Validation-aligned, history-aware client selection significantly improves federated learning convergence and stability in non-IID IoT environments.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.LG updates on arXiv.org.