Enhancing workplace productivity with secure AI using federated contrastive learning model for performance...

Area of Science:

• Artificial Intelligence
• Machine Learning
• Data Privacy

Background:

• Centralized AI data processing poses risks to employee privacy and is unsuitable for decentralized enterprise environments.
• Existing AI models struggle with scalability, data security, and inherent biases in centralized learning.
• Workplace productivity analysis is crucial but hampered by privacy and efficiency concerns in data handling.

Purpose of the Study:

• To develop a privacy-preserving AI model for decentralized workplace productivity analysis.
• To enhance AI model prediction accuracy, stability, and communication efficiency in federated learning.
• To address limitations of centralized data processing by proposing a secure and scalable federated learning framework.

Main Methods:

• Federated Contrastive Learning (FCL) framework integrating Contrastive Learning, Federated Averaging, and Homomorphic Encryption.
• Experimental analysis conducted on partitioned decentralized nodes simulating real-world federated learning scenarios.
• Utilized the Employee Performance and Productivity Dataset for model training and evaluation.

Main Results:

• The proposed FCL Model achieved a global accuracy of 98.9%, surpassing FedAvg (91.4%), LSTM (87.6%), and CNN (81.2%).
• Demonstrated high performance with precision (98.5%), recall (97.8%), and F1 score (97.9%).
• Significantly reduced data leakage by 97.2% and improved gradient compression efficiency by 95.2%, lowering communication overhead.

Conclusions:

• The FCL framework offers an efficient, scalable, and privacy-preserving solution for federated learning in workplace productivity analysis.
• This approach enables intelligent transformation of the workplace by moving beyond centralized data environments.
• The study highlights the potential of FCL for securing and adapting AI systems in the future of work.