A Bubble-Cluster Federated Learning Framework for Privacy-Preserving Demand Forecasting on Heterogeneous Retail Data

2025-03-15

AI Summary

To address the dual challenges of data heterogeneity and privacy preservation in retail forecasting, this paper proposes Bubble-Cluster, a novel federated learning framework. First, it introduces a risk-aware dynamic clustering mechanism that leverages differential privacy and feature importance distributions to adaptively partition heterogeneous retailers into multiple “bubbles.” Second, within each bubble, lightweight Transformer models are independently deployed for localized time-series forecasting. Third, robust aggregation and malicious/low-quality client detection enhance system security. Bubble-Cluster pioneers noise-adaptive bubble partitioning and a collaborative defense mechanism. Under strict privacy guarantees, it achieves a 5.4% improvement in R², a 69% reduction in RMSE, and a 45% decrease in MAE over FedAvg and local training—demonstrating superior trade-offs among prediction accuracy, privacy compliance, and robustness.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Natural Language Processing: Safety and Robustness
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Federated learning (FL) enables retailers to share model parameters for demand forecasting while maintaining privacy. However, heterogeneous data across diverse regions, driven by factors such as varying consumer behavior, poses challenges to the effectiveness of federated learning. To tackle this challenge, we propose Bubble-Cluster Federated Learning (BFL), a novel clustering-based federated learning framework tailored for sales prediction. By leveraging differential privacy and feature importance distribution, BFL groups retailers into distinct"bubbles", each forming its own federated learning (FL) system to effectively isolate data heterogeneity. Within each bubble, Transformer models are designed to predict local sales for each client. Our experiments demonstrate that BFL significantly surpasses FedAvg and outperforms local learning in demand forecasting performance across all participating clients. Compared to local learning, BFL can achieve a 5.4% improvement in R extsuperscript{2}, a 69% reduction in RMSE, and a 45% decrease in MAE. Our study highlights BFL's adaptability in enabling effective federated learning through dynamic adjustments to noise levels and the range of clients participating in each bubble. This approach strategically groups participants into distinct"bubbles"while proactively identifying and filtering out risky clients that could compromise the FL system. The findings demonstrate BFL's ability to enhance collaborative learning in regression tasks on heterogeneous data, achieving a balance between forecasting accuracy and privacy preservation in retail applications. Additionally, BFL's capability to detect and neutralize poisoned data from clients enhances the system's robustness and reliability, ensuring more secure and effective federated learning.

Problem

• Addresses privacy-preserving demand forecasting in retail using federated learning.
• Overcomes data heterogeneity challenges across diverse regions for accurate sales prediction.
• Enhances collaborative learning by grouping retailers into isolated 'bubbles' for improved accuracy and privacy.

Innovation

• Bubble-Cluster Federated Learning for heterogeneous data
• Transformer models for local sales prediction
• Differential privacy and feature importance distribution