QFedAgent: Quantum-Enhanced Personalized Federated Learning for Multi-Agent Activity Recognition

2026-07-02

AI Summary

This work addresses the performance degradation of conventional federated learning in multi-agent systems caused by heterogeneous, non-independent and identically distributed (non-IID) multimodal sensor data, as well as the high communication overhead and large parameter count of classical fusion modules. To overcome these challenges, the authors propose QFedAgent, a personalized federated learning framework that uniquely integrates quantum and classical computation by introducing variational quantum circuits into the multimodal fusion stage. Leveraging quantum state encoding and entanglement mechanisms, QFedAgent efficiently models accelerometer and gyroscope data. Evaluated on the OPPORTUNITY dataset under user-level non-IID partitioning, the method achieves an average test accuracy of 97.7% while enabling approximately 10× model compression, substantially reducing both communication and computational costs.

Abstract

Federated learning (FL) enables collaborative model training across distributed devices without sharing raw data, making it suitable for privacy-sensitive robotic sensing applications. However, multi-agent systems generate heterogeneous and non-independent and identically distributed (non-IID) multimodal sensor streams that degrade conventional FL algorithms, while classical fusion modules introduce substantial parameter overhead and communication cost. This paper proposes QFedAgent, a hybrid quantum-classical personalized FL framework for multi-agent activity recognition. The approach integrates a variational quantum circuit fusion module that models accelerometer--gyroscope interactions through quantum state encoding and entanglement, requiring only 72 quantum rotation parameters versus 33K in classical multi-layer perceptron-based fusion, achieving approximately 10x total parameter reduction. Experiments on the OPPORTUNITY dataset under subject-based non-IID partitions demonstrate 97.7% mean test accuracy, confirming that parameter-efficient quantum fusion remains competitive with conventional federated baselines.

Problem

• Federated Learning
• Non-IID Data
• Multi-Agent Systems
• Activity Recognition
• Multimodal Sensor Fusion

Innovation

• quantum-enhanced fusion
• personalized federated learning
• variational quantum circuit
• multi-agent activity recognition
• parameter-efficient learning