Search papers, labs, and topics across Lattice.
1
1
3
1
Steering worst-case trajectories with an adversarial network and Boltzmann reweighting dynamics ensembles yields a surprisingly stable and efficient approach to robust RL under dynamics uncertainty.