DS-Compatible Log-Linear Reliability with KL-Prox EM: Monotone Ascent, Identifiability, and Generalization

Shiva Koreddi, Sravani Sowrupilli
Proceedings of The 37th International Conference on Algorithmic Learning Theory, PMLR 313:1-17, 2026.

Abstract

We study context-conditioned reliability in Dawid–Skene (DS) models and propose a DS-compatible parameterization in which a log-linear correction to confusion logits is softmax-renormalized over reported labels, yielding valid, interpretable confusion matrices conditioned on covariates. We derive a KL-proximal (mirror-descent) update for confusion matrices that warm-starts at DS and provably yields monotone ascent of a standard EM surrogate. Under diagonal-dominance at warm start and mild covariate excitation, we prove identifiability up to permutation; for the correction head we give a finite-sample generalization bound of order $O(\sqrt{d\log(dK)/n})$ via Rademacher complexity. The formulation drops into vanilla EM, preserves DS interpretability, and supports physics-inspired priors (e.g., monotonicity) without breaking guarantees. Empirical validation on multi-agent label fusion confirms monotone convergence and calibration improvements with minimal overhead.

Cite this Paper

BibTeX
@InProceedings{pmlr-v313-koreddi26a, title = {DS-Compatible Log-Linear Reliability with KL-Prox EM: Monotone Ascent, Identifiability, and Generalization}, author = {Koreddi, Shiva and Sowrupilli, Sravani}, booktitle = {Proceedings of The 37th International Conference on Algorithmic Learning Theory}, pages = {1--17}, year = {2026}, editor = {Telgarsky, Matus and Ullman, Jonathan}, volume = {313}, series = {Proceedings of Machine Learning Research}, month = {23--26 Feb}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v313/main/assets/koreddi26a/koreddi26a.pdf}, url = {https://proceedings.mlr.press/v313/koreddi26a.html}, abstract = {We study context-conditioned reliability in Dawid–Skene (DS) models and propose a DS-compatible parameterization in which a log-linear correction to confusion logits is softmax-renormalized over reported labels, yielding valid, interpretable confusion matrices conditioned on covariates. We derive a KL-proximal (mirror-descent) update for confusion matrices that warm-starts at DS and provably yields monotone ascent of a standard EM surrogate. Under diagonal-dominance at warm start and mild covariate excitation, we prove identifiability up to permutation; for the correction head we give a finite-sample generalization bound of order $O(\sqrt{d\log(dK)/n})$ via Rademacher complexity. The formulation drops into vanilla EM, preserves DS interpretability, and supports physics-inspired priors (e.g., monotonicity) without breaking guarantees. Empirical validation on multi-agent label fusion confirms monotone convergence and calibration improvements with minimal overhead.} }
Endnote
%0 Conference Paper %T DS-Compatible Log-Linear Reliability with KL-Prox EM: Monotone Ascent, Identifiability, and Generalization %A Shiva Koreddi %A Sravani Sowrupilli %B Proceedings of The 37th International Conference on Algorithmic Learning Theory %C Proceedings of Machine Learning Research %D 2026 %E Matus Telgarsky %E Jonathan Ullman %F pmlr-v313-koreddi26a %I PMLR %P 1--17 %U https://proceedings.mlr.press/v313/koreddi26a.html %V 313 %X We study context-conditioned reliability in Dawid–Skene (DS) models and propose a DS-compatible parameterization in which a log-linear correction to confusion logits is softmax-renormalized over reported labels, yielding valid, interpretable confusion matrices conditioned on covariates. We derive a KL-proximal (mirror-descent) update for confusion matrices that warm-starts at DS and provably yields monotone ascent of a standard EM surrogate. Under diagonal-dominance at warm start and mild covariate excitation, we prove identifiability up to permutation; for the correction head we give a finite-sample generalization bound of order $O(\sqrt{d\log(dK)/n})$ via Rademacher complexity. The formulation drops into vanilla EM, preserves DS interpretability, and supports physics-inspired priors (e.g., monotonicity) without breaking guarantees. Empirical validation on multi-agent label fusion confirms monotone convergence and calibration improvements with minimal overhead.
APA
Koreddi, S. & Sowrupilli, S.. (2026). DS-Compatible Log-Linear Reliability with KL-Prox EM: Monotone Ascent, Identifiability, and Generalization. Proceedings of The 37th International Conference on Algorithmic Learning Theory, in Proceedings of Machine Learning Research 313:1-17 Available from https://proceedings.mlr.press/v313/koreddi26a.html.

Related Material