Where do you think Split Learning comes? https://splitlearning.github.io/
-
-
-
Great question! The model is another dimension we can slice along in order to find a partitioning that satisfies our security/privacy constraints. EL may e.g. be used only for fine-tuning certain layers, and for inference we may choose to perform preprocessing in plaintext.
- Još 3 druga odgovora
Novi razgovor -
-
-
What's your definition on encrypted learning ? (HE?) Because you do bring the data to the model, just in an encrypted way (the unencrypted model potentially contain the data)
-
Yes exactly: homomorphic encryption (HE), secure multi-party computations (MPC), garbled circuits (GC), or a hybrid between them -- finding the best mix is still an active area of research, and some may fit more naturally a given concrete task
- Još 2 druga odgovora
Novi razgovor -
-
-
Very nice nutshelling!
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
- AutoML in simulation when you have no data and can’t design a network
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.