arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2204.04512stat.MLcs.LGmath.FA
收藏

核类ε-熵的非渐近谱界

Non-asymptotic spectral bounds on the $\varepsilon$-entropy of kernel classes

Nazarbayev University(纳扎尔巴耶夫大学)

Rustem Takhanov

AI总结:

本文研究再生核希尔伯特空间单位球在L_p(ν)中ε-熵的非渐近谱界,证明p∈[1,∞]时的精确上下界,并揭示p>2时界还依赖截断Mercer级数的收敛速率,所得界比已有结果更紧且对特定核渐近最优。

摘要:

设 $K: \boldsymbolΩ\times \boldsymbolΩ$ 为定义在 ${\mathbb R}^n$ 的紧子集上的连续 Mercer 核,$\mathcal{H}_K$ 为与 $K$ 相关联的再生核希尔伯特空间(RKHS)。给定 $\boldsymbolΩ$ 上的有限测度 $ν$,我们研究 $\mathcal{H}_K$ 的单位球在空间 $L_p(ν)$ 中 ε-熵的上界与下界。该主题是基于核方法的现代统计理论中的一个重要方向。我们证明了当 $p\in [1,+\infty]$ 时的精确上界与下界。对于 $p\in [1,2]$,上界仅由相应积分算子 $ϕ\to \int_{\boldsymbolΩ} K(\cdot,{\mathbf y})ϕ({\mathbf y})dν({\mathbf y})$ 的特征值行为决定。相比之下,对于 $p>2$,这些界还额外依赖于截断 Mercer 级数在 $L_p(ν)$-范数下收敛到核 $K$ 的收敛速率。我们讨论了这些界的一系列推论,并证明它们比先前针对一般核的界要严格得多。此外,对于特定情形,例如球带核和盒子上的高斯核,当 $\varepsilon\to +0$ 时,我们的界是渐近紧的。

英文摘要:

Let $K: \boldsymbolΩ\times \boldsymbolΩ$ be a continuous Mercer kernel defined on a compact subset of ${\mathbb R}^n$ and $\mathcal{H}_K$ be the reproducing kernel Hilbert space (RKHS) associated with $K$. Given a finite measure $ν$ on $\boldsymbolΩ$, we investigate upper and lower bounds on the $\varepsilon$-entropy of the unit ball of $\mathcal{H}_K$ in the space $L_p(ν)$. This topic is an important direction in the modern statistical theory of kernel-based methods. We prove sharp upper and lower bounds for $p\in [1,+\infty]$. For $p\in [1,2]$, the upper bounds are determined solely by the eigenvalue behaviour of the corresponding integral operator $ϕ\to \int_{\boldsymbolΩ} K(\cdot,{\mathbf y})ϕ({\mathbf y})dν({\mathbf y})$. In constrast, for $p>2$, the bounds additionally depend on the convergence rate of the truncated Mercer series to the kernel $K$ in the $L_p(ν)$-norm. We discuss a number of consequences of our bounds and show that they are substantially tighter than previous bounds for general kernels. Furthermore, for specific cases, such as zonal kernels and the Gaussian kernel on a box, our bounds are asymptotically tight as $\varepsilon\to +0$.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑