Efficient Privacy-preserving Machine Learning in Hierarchical Distributed System.

Area of Science:

• Distributed machine learning
• Privacy-preserving techniques
• Hierarchical systems

Background:

• Massive data growth necessitates distributed machine learning.
• Raw data aggregation is infeasible due to privacy and physical constraints.
• Existing privacy-preserving methods face computational and architectural challenges in complex systems.

Purpose of the Study:

• To propose an efficient privacy-preserving machine learning scheme for hierarchical distributed systems.
• To reduce computational overheads and enhance learning speed.
• To provide comprehensive privacy protection across all system layers.

Main Methods:

• Modification and improvement of collaborative learning algorithms.
• Development of an asynchronous strategy for enhanced efficiency.
• Evaluation using extensive experiments on real-world data.

Main Results:

• The proposed scheme reduces learning overhead.
• Comprehensive privacy protection is achieved for each hierarchical layer.
• The asynchronous strategy further boosts learning efficiency.

Conclusions:

• The developed scheme offers an efficient and secure solution for privacy-preserving machine learning in hierarchical systems.
• The approach balances learning efficacy with robust privacy guarantees.
• Experimental validation confirms the scheme's performance on real-world datasets.