Create Value and Insight Through Privacy-preserving Data Collaboration [...]
summary
Claims a deployed federated learning stack (based on open source FATE and a blockchain layer) that trains joint models without sharing raw data, using privacy intersection (PSI) for ID matching and homomorphic encryption to compute gradients on ciphertext. Demonstrates a bankruptcy/credit pilot with Hong Kong/China banks and alternative data providers, exposing prediction via simple REST APIs, and reports a model performance comparable to centralized training at AUC 95.5% using vertical federated learning.
AI-written summary · about
chapters (11)
- 0:00Privacy-preserving data collaboration
- 1:52OpenHive’s privacy tech
- 4:01Privacy-enhancing computation
- 7:29Federated learning types
- 9:39Federated learning use cases
- 15:37Privacy mechanisms in federated learning
- 19:55Federated learning in China
- 21:18Hong Kong HSBC pilot
- 23:27Enterprise API platform
- 24:31Regulatory support signals
- 25:14Building privacy ecosystem
related talks
29mBalancing Thin Line Between Data Intelligence & Privacy
42mFederated Learning: Unlocking the Values of Discrete Data for Enterprise
28mA Privacy Enhancing Federated Learning Framework for Multi-Party Distributed ..
40mDemystifying Decentralized Fair Models for Federated Machine Learning
01h 20mUnlocking societal benefits of AI with privacy protective technologies.
21mTowards Building a Responsible Data Economy
34mDecentralized Machine Learning
21mWhat, Why and How of Federated Machine Learning – Implementation Using...