Advanced encryption options for privacy-preserving ML
We discussed data encryption in Chapter 3 as a security control to protect data confidentiality for data in transit and at rest.
A more advanced approach to using cryptography is to apply encryption to machine learning processes to protect data confidentiality and privacy.
Secure multi-party computation (secure MPC)
Secure MPC is a cryptographic method allowing multiple parties to jointly compute a function without revealing their individual inputs. Here’s how it works:
- Data splitting with cryptography: Each participant’s data is divided into encrypted shares using cryptographic algorithms. These shares conceal any meaningful information.
- Distributed cryptographic computation: Parties perform computations on their encrypted shares and exchange encrypted intermediate results. Advanced cryptographic methods ensure that these computations and exchanges do not compromise the original data’s...
