Privacy concerns arise with the prevalent use of AI-based chatbots, as many chatbot developers do not clearly explain how they collect and use user data. At the last ISC2024 conference, Dr Bharath Ramesh from Western Sydney University, Australia presented his research on scalable decentralised federated learning framework for confidential chatbots. His research aims to address the privacy concerns associated with chatbot interactions by leveraging blockchain technology.
Federated Learning is a method of training AI models without directly accessing user data. It enables IoT nodes, sensors and AI systems to learn from each other without sharing their data. However, the centralised server orchestrating federated learning is vulnerable to issues such as single point of failure or model inversion attacks.
Dr Bharath’s research focuses on replacing the centralised server with blockchain technology, where smart contracts facilitate the learning process. This approach aims to enhance privacy in machine learning applications. By utilising blockchain, the framework ensures a more secure and decentralised method for federated learning, mitigating the risks associated with central servers.
Dr. Bharath's impressive research had earned him first prize in the research abstract presentation at the 6th Blockchain International Scientific Conference ISC2024. Watch his video presentation to gain more insights into his innovative work!
Full paper: doi.org/10.315...
Demo video: • Distributed Federated ...
Su, Hongxu, Cheng Xiang, and Bharath Ramesh. 2024. “Towards Confidential Chatbot Conversations: A Decentralised Federated Learning Framework.” The Journal of The British Blockchain Association, February.
Негізгі бет Towards Confidential Chatbots: A Scalable Decentralised Federated Learning Framework"
Пікірлер