Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
1
0
0
0
Multi-policy RL can outperform single policies in non-stationary settings not due to theoretical limitations, but due to practical function approximation and learning dynamics, especially when phase durations are long or phases are highly heterogeneous.