Search papers, labs, and topics across Lattice.
1
0
2
8
Achieving fairness in synthetic data generation isn't just about bias correction—it's about ensuring utility parity across sensitive groups, and this paper reveals how to do it effectively.