Robbie with a Generative Model
Generative Model
e,T-optimal
(|A|T/e)
O(T)
Sparse Sampling
e/T-classification
O(|A|
T
)
RLGen
e-local-optimal
O(T
2
)
various
&mu = opt. dist
Te,T-optimal
PSDP
basic setting
direct experience
reset model
precise description
full table