Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Denoising and Adaptive Online Vertical Federated Learning for Sequential Multi-Sensor Data in Industrial Internet of Things

Heqiang Wang, Xiaoxiong Zhong, Kang Liu, Fangming Liu, Weizhe Zhang

TL;DR

DAO-VFL presents an online vertical federated learning framework tailored for multi-sensor IIoT assembly lines. It combines denoising autoencoders to mitigate communication noise with a DRL-based adaptive local iteration strategy to balance learning performance and system latency. Theoretical regret analysis links denoising and adaptation to sublinear regret growth, while experiments on CIFAR-10 and C-MAPSS demonstrate robustness to noise and improved efficiency. This work offers a practical approach for privacy-preserving, real-time collaborative learning in noisy, heterogeneous industrial environments.

Abstract

With the continuous improvement in the computational capabilities of edge devices such as intelligent sensors in the Industrial Internet of Things, these sensors are no longer limited to mere data collection but are increasingly capable of performing complex computational tasks. This advancement provides both the motivation and the foundation for adopting distributed learning approaches. This study focuses on an industrial assembly line scenario where multiple sensors, distributed across various locations, sequentially collect real-time data characterized by distinct feature spaces. To leverage the computational potential of these sensors while addressing the challenges of communication overhead and privacy concerns inherent in centralized learning, we propose the Denoising and Adaptive Online Vertical Federated Learning (DAO-VFL) algorithm. Tailored to the industrial assembly line scenario, DAO-VFL effectively manages continuous data streams and adapts to shifting learning objectives. Furthermore, it can address critical challenges prevalent in industrial environment, such as communication noise and heterogeneity of sensor capabilities. To support the proposed algorithm, we provide a comprehensive theoretical analysis, highlighting the effects of noise reduction and adaptive local iteration decisions on the regret bound. Experimental results on two real-world datasets further demonstrate the superior performance of DAO-VFL compared to benchmarks algorithms.

Denoising and Adaptive Online Vertical Federated Learning for Sequential Multi-Sensor Data in Industrial Internet of Things

TL;DR

DAO-VFL presents an online vertical federated learning framework tailored for multi-sensor IIoT assembly lines. It combines denoising autoencoders to mitigate communication noise with a DRL-based adaptive local iteration strategy to balance learning performance and system latency. Theoretical regret analysis links denoising and adaptation to sublinear regret growth, while experiments on CIFAR-10 and C-MAPSS demonstrate robustness to noise and improved efficiency. This work offers a practical approach for privacy-preserving, real-time collaborative learning in noisy, heterogeneous industrial environments.

Abstract

With the continuous improvement in the computational capabilities of edge devices such as intelligent sensors in the Industrial Internet of Things, these sensors are no longer limited to mere data collection but are increasingly capable of performing complex computational tasks. This advancement provides both the motivation and the foundation for adopting distributed learning approaches. This study focuses on an industrial assembly line scenario where multiple sensors, distributed across various locations, sequentially collect real-time data characterized by distinct feature spaces. To leverage the computational potential of these sensors while addressing the challenges of communication overhead and privacy concerns inherent in centralized learning, we propose the Denoising and Adaptive Online Vertical Federated Learning (DAO-VFL) algorithm. Tailored to the industrial assembly line scenario, DAO-VFL effectively manages continuous data streams and adapts to shifting learning objectives. Furthermore, it can address critical challenges prevalent in industrial environment, such as communication noise and heterogeneity of sensor capabilities. To support the proposed algorithm, we provide a comprehensive theoretical analysis, highlighting the effects of noise reduction and adaptive local iteration decisions on the regret bound. Experimental results on two real-world datasets further demonstrate the superior performance of DAO-VFL compared to benchmarks algorithms.
Paper Structure (24 sections, 1 theorem, 21 equations, 10 figures, 2 algorithms)

This paper contains 24 sections, 1 theorem, 21 equations, 10 figures, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. FL for Industrial Internet of Things
  4. Vertical Federated Learning
  5. Online Federated Learning
  6. System Model
  7. Denoising and Adaptive Online Vertical Federated Learning
  8. Feature Embedding Extracting
  9. Feature Embedding Denoising
  10. Model Representation Distributing
  11. Feature and Head Model Updating
  12. Regret Analysis
  13. Adaptive Local Iteration Decisions
  14. Collection Latency
  15. Communication Latency
  16. ...and 9 more sections

Key Result

Theorem 1

Under Assumption 1-5, DAO-VFL with adaptive local iterations decisions $E_{t, k} \geq 1$, while considering the impact of denoising, achieves the following regret bound:

Figures (10)

  • Figure 1: DAO-VFL for IIoT-Based Assembly Line
  • Figure 2: Challenges and Solutions in DAO-VFL
  • Figure 3: The Architecture of DAO-VFL Noise Reduction
  • Figure 4: Learning Performance of DAO-VFL with Different Local Iteration Decisions Patterns $(E_{t, k})$ .
  • Figure 5: Timeline of Sensors in One Global Round
  • ...and 5 more figures

Theorems & Definitions (2)

  • Theorem 1
  • proof