HIGHLIGHTS
SUMMARY
Security: On one hand, the resources of edge devices are insufficient to support large scale traditional privacy methods, e_g Homomorphic Encryption (HE). HFL is the union of samples, which is applicable when most features while few samples overlap e_g sharing diagnosis data between hospitals in different regions for training a more robust model to make accurate diagnoses. VFL is suitable when there are many samples overlapping and few features overlapping, e_g, banks and Internet companies sharing data to model client credit for risk control. FTL applies when both the samples and the features overlap . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.