English(EN) Conservative Dueling Bandits: Preference Learning under Baseline Constraints

新算法确保 AI 偏好学习保持在安全范围内

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-28 04:00

研究人员开发了保守的上下文对决土匪（conservative contextual dueling bandits）的新算法，旨在从成对比较中学习，同时遵守基线约束。这些算法 MaxInP-BC 和 BTIGW-BC 确保显示项的分数保持在可信基线的指定分数范围内，这是诸如使用人类偏好对语言模型进行微调等应用的关键功能。实验表明，这些保守方法满足了约束条件，而无约束的对应方法经常违反这些条件，且仅导致适度的遗憾增加。 AI

影响这些算法可以提高从人类反馈中学习的 AI 系统（如大型语言模型）的安全性和可靠性。

排序理由学术论文，详细介绍偏好学习的新算法。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 15 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 学术论文，详细介绍偏好学习的新算法。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, safety; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Rohan Deb, Mohammad Ghavamzadeh, Arindam Banerjee · 2026-10-28 04:00

保守派对决土匪：基线约束下的偏好学习

arXiv:2610.36708v1 Announce Type: cross Abstract: Contextual dueling bandits learn from pairwise comparisons, the same feedback used to fine-tune language models from human preferences. Standard algorithms explore aggressively and can show users two poor options. We study conserv…

本文如何被排名

报道来源 [1]

保守派对决土匪：基线约束下的偏好学习

相关实体

相关话题