TV-AIL achieves a horizon-independent imitation gap of O(min{1, sqrt(|S|/N)}) via stage-coupled dynamic programming analysis on locomotion-abstracted MDPs.
For the lower bound instances (refer to Assumption 2), we consider that the initial state distribution is uniform over all states
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2022 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Understanding Adversarial Imitation Learning in Small Sample Regime: A Stage-coupled Analysis
TV-AIL achieves a horizon-independent imitation gap of O(min{1, sqrt(|S|/N)}) via stage-coupled dynamic programming analysis on locomotion-abstracted MDPs.