Mind the Gap: Structure-Aware Consistency in Preference Learning

• URL: http://arxiv.org/abs/2604.27733v1
• Date: Thu, 30 Apr 2026 11:24:04 GMT
• Title: Mind the Gap: Structure-Aware Consistency in Preference Learning
• Abstract summary: Preference learning has become the foundation of aligning Large Language Models with human intent.<n>We show that for equicontinuous hypothesis sets typical of neural networks, standard surrogates are theoretically inconsistent.<n>We derive rigorous $H$-consistency bounds that depend on enforcing a separation margin.<n>We extend this to Structure-Aware $H$-consistency, introducing a novel objective (SA-DPO) that adapts the margin based on the semantic distance between responses to handle synonyms and hard pairs.
• Score: 42.67092904252001
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Preference learning has become the foundation of aligning Large Language Models (LLMs) with human intent. Popular methods, such as Direct Preference Optimization (DPO), minimize surrogate losses as proxies for the intractable pairwise ranking loss. However, we demonstrate that for the equicontinuous hypothesis sets typical of neural networks, these standard surrogates are theoretically inconsistent, yielding vacuous generalization guarantees. To resolve this, we formulate LLM alignment within a margin-shifted ranking framework. We derive rigorous $H$-consistency bounds that depend on enforcing a separation margin $γ$. Crucially, we extend this to Structure-Aware $H$-consistency, introducing a novel objective (SA-DPO) that adapts the margin based on the semantic distance between responses to handle synonyms and hard pairs. Finally, we analyze the trade-off between consistency and model limitations via the Margin-Capacity Profile, proving that heavy-tailed surrogates (such as the Polynomial Hinge family) offer superior consistency guarantees for capacity-bounded models compared to the standard logistic loss used in DPO.

Related papers

• Ratio-Variance Regularized Policy Optimization [64.95520246570446]
  We show that explicitly constraining the policy ratio variance provides a principled local approximation to trust-region constraints.<n>We introduce $bf R2bf VPO$ (Ratio-Variance Regularized Policy Optimization), which implements this constraint via a primal-dual optimization framework.
  arXiv  Detail & Related papers  (2026-05-26T09:53:42Z)
• Too Correct to Learn: Reinforcement Learning on Saturated Reasoning Data [55.84428098924793]
  We propose Constrained Uniform Top-K Sampling (CUTS), a parameter-free decoding strategy enforcing structure-preserving exploration.<n>We integrate this into Mixed-CUTS, a training framework synergizing exploitative and exploratory rollouts to amplify intra-group advantage variance.<n> Notably, Mixed-CUTS improves Pass@1 accuracy on the challenging AIME25 benchmark by up to 15.1% over standard GRPO.
  arXiv  Detail & Related papers  (2026-04-20T16:43:28Z)
• Bulk-Calibrated Credal Ambiguity Sets: Fast, Tractable Decision Making under Out-of-Sample Contamination [8.826173150779145]
  Distributionally robust optimisation (DRO) minimises the worst-case expected loss over an ambiguity set.<n>We show how IP credal sets translate into DRO objectives with interpretable tolerance levels.
  arXiv  Detail & Related papers  (2026-01-29T06:37:36Z)
• Fundamental Novel Consistency Theory: $H$-Consistency Bounds [19.493449206135296]
  In machine learning, the loss functions optimized during training often differ from the target loss that defines task performance.<n>We present an in-depth study of the target loss estimation error relative to the surrogate loss estimation error.<n>Our analysis leads to $H$-consistency bounds, which are guarantees accounting for the hypothesis set $H$.
  arXiv  Detail & Related papers  (2025-12-28T11:02:20Z)
• NDCG-Consistent Softmax Approximation with Accelerated Convergence [67.10365329542365]
  We propose novel loss formulations that align directly with ranking metrics.<n>We integrate the proposed RG losses with the highly efficient Alternating Least Squares (ALS) optimization method.<n> Empirical evaluations on real-world datasets demonstrate that our approach achieves comparable or superior ranking performance.
  arXiv  Detail & Related papers  (2025-06-11T06:59:17Z)
• Supervised Optimism Correction: Be Confident When LLMs Are Sure [91.7459076316849]
  We establish a novel theoretical connection between supervised fine-tuning and offline reinforcement learning.<n>We show that the widely used beam search method suffers from unacceptable over-optimism.<n>We propose Supervised Optimism Correction, which introduces a simple yet effective auxiliary loss for token-level $Q$-value estimations.
  arXiv  Detail & Related papers  (2025-04-10T07:50:03Z)
• $i$REPO: $i$mplicit Reward Pairwise Difference based Empirical Preference Optimization [12.266207199002604]
  Large Language Models (LLM) can sometimes produce outputs that deviate from human expectations. We propose a novel framework named $i$REPO, which utilizes implicit Reward pairwise difference regression for Empirical Preference Optimization. We show that $i$REPO effectively achieves self-alignment using soft-label, self-generated responses and the logit of empirical AI annotators.
  arXiv  Detail & Related papers  (2024-05-24T05:42:11Z)
• Distributional Robustness and Regularization in Reinforcement Learning [62.23012916708608]
  We introduce a new regularizer for empirical value functions and show that it lower bounds the Wasserstein distributionally robust value function. It suggests using regularization as a practical tool for dealing with $textitexternal uncertainty$ in reinforcement learning.
  arXiv  Detail & Related papers  (2020-03-05T19:56:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.