Search papers, labs, and topics across Lattice.
1
0
2
Generate more robust risk scenarios: GAR uses adversarial training to create generative models that are resilient to worst-case policy discrepancies, outperforming traditional methods in preserving downstream risk.
