用强化学习框架分析压缩中的探索与利用权衡。
Exploration-Exploitation Tradeoff in Universal Lossy Compression
- 将顺序压缩建模为多臂赌博机问题，优化探索与利用。
- 提出稳健的代价导向算法，适用于任意块长。
- 解释了原有方法在短块长下的性能局限。
通用压缩可学习数据源并以批处理模式（前向适应）或顺序模式（后向适应）进行自适应。本文将顺序模式重新建模为强化学习中的经典多臂赌博机问题，研究损失压缩中探索与利用的权衡。我们证明，此前提出的“自然类型选择”方案可视为一种重建导向的多臂赌博机算法，适用于顺序损失压缩，并从鲁棒性和短块长性能角度解释其局限性。随后，我们推导并分析了稳健的代价导向多臂赌博机算法，该算法可在任意块长下有效工作。
原文摘要 · Abstract (English)
Universal compression can learn the source and adapt to it either in a batch mode (forward adaptation), or in a sequential mode (backward adaptation). We recast the sequential mode as a multi-armed bandit problem, a fundamental model in reinforcement-learning, and study the trade-off between exploration and exploitation in the lossy compression case. We show that a previously proposed "natural type selection" scheme can be cast as a reconstruction-directed MAB algorithm, for sequential lossy compression, and explain its limitations in terms of robustness and short-block performance. We then derive and analyze robust cost-directed MAB algorithms, which work at any block length.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。