arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2306.06262stat.MLcs.LG
收藏

基于谱间隙的确定性张量补全

Spectral gap-based deterministic tensor completion

Western Washington University(西华盛顿大学);Harbor Branch Oceanographic Institute(港湾分支海洋学研究所);Florida Atlantic University(佛罗里达大西洋大学);University of California, Irvine(加州大学欧文分校)

Kameron Decker Harris, Oscar López, Angus Read, Yizhe Zhu

AI总结:

该研究针对确定性采样下张量补全理论有限的问题,推导了Poisson损失与原子范数最小化方法的更紧泛化误差界,改进了秩依赖关系,验证了实用算法重构误差与谱间隙的关联,为张量算法研究提供新视角。

摘要:

张量补全是应用于推荐系统及其他存在缺失数据领域的核心机器学习算法。尽管矩阵情形的相关理论已较为成熟,但张量问题的理论成果仍较为有限,尤其是在采样模式为确定性的情况下。本文对两种张量补全方法——Poisson损失(泊松损失)与原子范数最小化——的解的泛化误差给出了界,在目标张量秩的维度上得到了更紧的界。若真实张量为t阶且CP秩为r,对r的依赖度从arXiv:1910.10692中的$r^{2(t-1)(t^2-t-1)}$改进为$r^{2(t-1)(3t-5)}$。我们的误差界由采样稀疏模式的谱间隙确定性地控制。我们还证明了张量原子范数的若干新性质,在随机采样方案下,将秩依赖度从arXiv:1711.04965中的$r^{3t-3}$降至$r^{3t-5}$。存在的一个局限是,原子范数最小化虽在理论上有研究价值,但对应的算法效率较低。不过数值实验表明,在实用的max-quasinorm(最大拟范数)、岭惩罚以及Poisson损失最小化算法中,重构误差与谱间隙存在依赖关系。这种从谱间隙出发的视角为张量算法的进一步研究提供了有前景的方向。

英文摘要:

Tensor completion is a core machine learning algorithm used in recommender systems and other domains with missing data. While the matrix case is well-understood, theoretical results for tensor problems are limited, particularly when the sampling patterns are deterministic. Here we bound the generalization error of the solutions of two tensor completion methods, Poisson loss and atomic norm minimization, providing tighter bounds in terms of the target tensor rank. If the ground-truth tensor is order $t$ with CP-rank $r$, the dependence on $r$ is improved from $r^{2(t-1)(t^2-t-1)}$ in arXiv:1910.10692 to $r^{2(t-1)(3t-5)}$. The error in our bounds is deterministically controlled by the spectral gap of the sampling sparsity pattern. We also prove several new properties for the atomic tensor norm, reducing the rank dependence from $r^{3t-3}$ in arXiv:1711.04965 to $r^{3t-5}$ under random sampling schemes. A limitation is that atomic norm minimization, while theoretically interesting, leads to inefficient algorithms. However, numerical experiments illustrate the dependence of the reconstruction error on the spectral gap for the practical max-quasinorm, ridge penalty, and Poisson loss minimization algorithms. This view through the spectral gap is a promising window for further study of tensor algorithms.

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。

↑