Encoded Spatial Attribute in Multi-Tier Federated Learning

TL;DR

This work addresses geospatial prediction under privacy constraints by proposing Encoded Spatial Multi-Tier Federated Learning, which encodes spatial information at the client tier and aggregates through multiple tiers to capture patterns at varying spatial granularities. The method combines spatial encoding with an N-tier FL architecture, where local updates follow gradient-based optimization on encoded data and higher-tier aggregations use FedAvg to fuse regional and global insights. Empirical results on Canadian greenhouse gas emissions and New Brunswick electric vehicle charging datasets demonstrate global accuracies of 75.62% and 89.52% without tier-specific training, highlighting the potential for real-time, localized geospatial forecasting while preserving data privacy. The approach suggests promising avenues for improving predictive accuracy and generalization via refined architectures, additional features, and alternative modeling strategies in privacy-preserving, multi-scale spatial learning.

Abstract

This research presents an Encoded Spatial Multi-Tier Federated Learning approach for a comprehensive evaluation of aggregated models for geospatial data. In the client tier, encoding spatial information is introduced to better predict the target outcome. The research aims to assess the performance of these models across diverse datasets and spatial attributes, highlighting variations in predictive accuracy. Using evaluation metrics such as accuracy, our research reveals insights into the complexities of spatial granularity and the challenges of capturing underlying patterns in the data. We extended the scope of federated learning (FL) by having multi-tier along with the functionality of encoding spatial attributes. Our N-tier FL approach used encoded spatial data to aggregate in different tiers. We obtained multiple models that predicted the different granularities of spatial data. Our findings underscore the need for further research to improve predictive accuracy and model generalization, with potential avenues including incorporating additional features, refining model architectures, and exploring alternative modeling approaches. Our experiments have several tiers representing different levels of spatial aspects. We obtained accuracy of 75.62% and 89.52% for the global model without having to train the model using the data constituted with the designated tier. The research also highlights the importance of the proposed approach in real-time applications.

Figures (1)

• Figure 1: N-tier FL architecture with spatial encoding.