Federated Split Learning for Human Activity Recognition with Differential Privacy

TL;DR

A novel intelligent human activity recognition framework based on a new design of Federated Split Learning with Differential Privacy with Differential Privacy over edge networks is proposed, showing that the FSL framework outperforms FL models in both accuracy and loss metrics.

Abstract

This paper proposes a novel intelligent human activity recognition (HAR) framework based on a new design of Federated Split Learning (FSL) with Differential Privacy (DP) over edge networks. Our FSL-DP framework leverages both accelerometer and gyroscope data, achieving significant improvements in HAR accuracy. The evaluation includes a detailed comparison between traditional Federated Learning (FL) and our FSL framework, showing that the FSL framework outperforms FL models in both accuracy and loss metrics. Additionally, we examine the privacy-performance trade-off under different data settings in the DP mechanism, highlighting the balance between privacy guarantees and model accuracy. The results also indicate that our FSL framework achieves faster communication times per training round compared to traditional FL, further emphasizing its efficiency and effectiveness. This work provides valuable insight and a novel framework which was tested on a real-life dataset.

Figures (5)

• Figure 1: Proposed FSL framework for HAR over distributed egde networks. Each HAR model is divided into a device-side model and a server-side model. EDs collaborate to train the device-side model and share feature representations to the server that assists to complete the model training by executing the server-side model.
• Figure 2: Comparison between FSL schemes with and without DP.
• Figure 3: Comparison between FSL schemes under different data settings.
• Figure 4: Comparison between the proposed FSL and traditional FL.
• Figure 5: Comparison of communication time between the proposed FSL and traditional FL.