Skip to main content

Table 1 Comparison of NAC features with related methods

From: Selective network discovery via deep reinforcement learning on embedded spaces

Method State space Action space Observability Learning goal Learning framework Policy training State embedding
NOL (LaRock et al. 2018) Large Dynamic Partial Vertex property MDP Online No
\(D^3TS\) (Murai et al. 2017) Large Dynamic Partial Vertex property Supervised Online No
GCPN (You et al. 2018) Small Fixed Full Graph property MDP Offline on given dataset No
NAC Large Dynamic Partial Vertex property MDP Offline and online on designed dataset Yes