RIC replaces single-pass label imitation with RL-driven iterative belief refinement, recovering cross-entropy optima while enabling adaptive halting via a value function.
Recurrent networks, hidden states and beliefs in partially observable environments.arXiv preprint arXiv:2208.03520
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Do Not Imitate, Reinforce: Iterative Classification via Belief Refinement
RIC replaces single-pass label imitation with RL-driven iterative belief refinement, recovering cross-entropy optima while enabling adaptive halting via a value function.