Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
One TD update is sufficient for linear convergence in regularized MDPs, and off-policy sample complexity can be achieved without trajectory resets or generative models.
