RIC replaces single-pass label imitation with RL-driven iterative belief refinement, recovering cross-entropy optima while enabling adaptive halting via a value function.
Reading digits in natural images with unsupervised feature learning
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
GenCE is a strictly proper loss obtained by normalizing each sample's softmax against the batch predictions, outperforming cross-entropy in low-data and imbalanced regimes with better calibration and OOD detection.
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.
-
Generative Cross-Entropy: A Strictly Proper Loss for Data-Efficient Classification
GenCE is a strictly proper loss obtained by normalizing each sample's softmax against the batch predictions, outperforming cross-entropy in low-data and imbalanced regimes with better calibration and OOD detection.