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

利用基于联邦对比学习模型的安全人工智能提升工作场所生产力,实现性能优化

阅读:1

Abstract

Artificial intelligence (AI) adoption is now heading towards a fast speed moving us towards a more productive future where intelligent systems automate complex works, optimize decisions and personalize work arena. But AI powered transformation realization brings problems, such as securing employee data privacy, making the solution scalable for various organizational environments and fights against biases of centralized learning methodology. Centralized data processing is common today, and it means that sensitive information is vulnerable to security weaknesses, and it is not good for the decentralized data ecosystem in large enterprises. To deal with these limitations, this research suggests Federated Contrastive Learning (FCL) framework for the secure and efficient workplace productivity analysis using the Employee Performance and Productivity Dataset. The goal is to design a privacy preserving AI model for decentralized learning while preserving data security, but at the same time improving model prediction accuracy, stability and communication efficiency. With Contrastive Learning, Federated Averaging for iterating over centralized update and Homomorphic Encryption for training model in the secure way we propose the model. We conduct experimental analysis on partitioned decentralized nodes that mimic real federated learning. I also showed that the Proposed FCL Model achieves global accuracy of 98.9% that outperforms FedAvg (91.4%), LSTM (87.6%) and CNN (81.2%), and has precision (98.5%), recall (97.8%), and F1 score (97.9%). Moreover, the model also reduced data leakage by 97.2% and increased gradient compression efficiency to 95.2%, which greatly lowers communication overhead. The results of these results demonstrate the efficiency of the FCL framework for achieving privacy preserving, scalable and high-performance federated learning for workplace productivity analysis. The implications of this study for securing, adapting intelligent systems in the future of work, skipping the centralized data environment, while driving the intelligent transformation of the workplace, are insightful and compelling.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。