: 15h00, ngày 14/07/2023 (Thứ Sáu)
: P104 D3
: Machine Learning và Data Mining
: Nguyễn Công Thịnh
: Grooo International
Tóm tắt báo cáo
A potential distributed machine learning system called federated learning (FL) trains models without utilizing shared local data while maintaining privacy. FL makes use of the idea of collaborative learning and creates models that protect privacy. However, FL's core components are not without their share of issues, such as the leakage of personal data, the dependability of uploading model parameters to the server, the expense of connectivity, etc. By combining federated learning with blockchain, the system addresses critical privacy, security, and trust challenges in collaborative machine learning. It offers robust data privacy guarantees, as sensitive information remains decentralized on participants' devices and is not exposed during the model training process. Additionally, the transparency and audibility provided by blockchain enhance accountability and trust among participants.
A review of the literature on the use of blockchain in federated learning was taken into account in this research, along with an analysis of the FL issues that can be addressed. In conclusion, the integration of federated learning with blockchain technology presents a promising approach to address privacy concerns and enable secure collaborative machine learning. Future research and development in this domain can explore practical implementations, scalability considerations, and optimization techniques to make this hybrid paradigm a reality in various applications and industries.