Related Experiment Video
Updated: Jan 18, 2026
03:14
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
1.0K
Enhancing Data Privacy in Human Factors Studies with Federated Learning.
Bingyi Su1, Liwei Qing1, Lu Lu1
1North Carolina State University, USA.
Human Factors
|June 6, 2025
Summary
Federated learning offers a privacy-preserving alternative for machine learning in human factors research. This approach achieves accuracy comparable to centralized methods while protecting sensitive human data.
Area of Science:
- Human Factors and Ergonomics
- Machine Learning
- Data Privacy
Background:
- Machine learning is transforming human factors research but faces privacy challenges with sensitive data.
- Centralized machine learning models raise significant data privacy concerns.
- Federated learning (FL) is proposed to address these privacy issues.
Purpose of the Study:
- Develop and evaluate a privacy-preserving federated learning framework.
- Assess FL efficacy in classifying mental stress during human-robot collaboration.
- Assess FL efficacy in recognizing human activities during manual material handling.
Main Methods:
- Constructed classifiers using both centralized and federated learning approaches.
- Employed support vector machines for mental stress classification.
- Utilized a deep neural network (LSTM-CNN) for human activity recognition.
Main Results:
- Federated learning models demonstrated accuracy comparable to centralized methods.
- Performance differences between federated and centralized models were minimal (under 2.7%).
- Federated learning effectively protected sensitive human data.
Conclusions:
- Federated learning is a viable alternative to traditional machine learning for human factors applications.
- FL provides comparable accuracy with enhanced data privacy.
- This research advances privacy-preserving methods for sensitive human-subject data.
