两篇新研究论文探讨了多类别分类设定下 Bandit 学习算法的样本复杂度。第一篇论文引入了“Bandit DS 维度”来表征 Bandit 反馈下的 PAC 学习样本复杂度，并提出了一种名为 ListCascade 的新算法。第二篇论文专注于稀疏上下文 Bandit，提出了一种通过利用奖励向量的稀疏性和分析决策估计系数 (DEC) 来实现改进样本复杂度界限的算法。两篇论文都旨在为标签信息不完整的学习场景提供更严格的理论保证。 AI
影响 这些论文为 Bandit 学习提供了理论上的进展，有望为反馈有限的任务带来更高效的算法。
排序理由 该集群包含两篇在 arXiv 上发表的学术论文，详细介绍了机器学习算法的理论进展。
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
arXiv:2605.25678v1 Announce Type: cross Abstract: We study the problem of multiclass PAC learning with bandit feedback in the realizable setting. In this framework, there is an unknown data distribution over an instance space $\mathcal{X}$ and a label space $\mathcal{Y}$, as in c…
arXiv:2605.29645v1 Announce Type: cross Abstract: We study contextual bandits in the stochastic i.i.d.\ setting, where a learner observes contexts drawn from an unknown distribution, selects actions from a finite set $A$, and aims to identify an approximately optimal policy from …
We study contextual bandits in the stochastic i.i.d.\ setting, where a learner observes contexts drawn from an unknown distribution, selects actions from a finite set $A$, and aims to identify an approximately optimal policy from a given class based on bandit feedback. Motivated …
We study the problem of multiclass PAC learning with bandit feedback in the realizable setting. In this framework, there is an unknown data distribution over an instance space $\mathcal{X}$ and a label space $\mathcal{Y}$, as in classical multiclass PAC learning, but the learner …
We study the problem of multiclass PAC learning with bandit feedback in the realizable setting. In this framework, there is an unknown data distribution over an instance space $\mathcal{X}$ and a label space $\mathcal{Y}$, as in classical multiclass PAC learning, but the learner …