arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2502.06164cs.LGstat.ML
收藏

函数型复杂度自适应时序张量分解

Functional Complexity-adaptive Temporal Tensor Decomposition

College of Information Science and Electronic Engineering, Zhejiang University(浙江大学信息科学与电子工程学院);Zhejiang Provincial Key Laboratory of Multi-Modal Communication Networks and Intelligent Information Processing(浙江省多模态通信网络与智能信息处理重点实验室);Microsoft Research Asia(微软亚洲研究院)

Panqi Chen, Lei Cheng, Jianlong Li, Weichang Li, Weiqing Liu, Jiang Bian, Shikai Fang

AI总结:

针对多模式连续索引与复杂度自适应难题,提出CATTE,结合Fourier特征、neural ODE和稀疏先验及免采样变分推断,提升预测与抗噪性能。

摘要:

张量分解是分析多维数据的基础工具,它通过学习低秩因子来表示高阶交互。尽管近期关于时序张量分解的工作通过在潜在因子中引入连续时间戳而取得了显著进展,但它们仍难以处理不仅在时间模式、而且在其他模式(例如气候数据中的空间坐标)也具有连续索引的一般张量数据。此外,在函数型时序张量模型中,模型复杂度的自适应挑战在很大程度上尚未得到探索,现有方法在这一设定下并不适用。为克服这些局限,我们提出函数型 Complexity-Adaptive Temporal Tensor dEcomposition(CATTE)。我们的方法将连续空间索引编码为可学习的 Fourier 特征,并在潜在空间中采用 neural ODE 来学习因子的时间轨迹。为实现模型复杂度的自动适应,我们在因子轨迹上引入了一种诱导稀疏性的先验。我们开发了一种高效的变分推断方案,该方案具有解析的证据下界,可实现无需采样的优化。通过在合成数据集和真实世界数据集上的大量实验,我们证明 CATTE 不仅能够揭示函数型时序张量的潜在秩,而且在预测性能和抗噪声鲁棒性方面显著优于现有方法。

英文摘要:

Tensor decomposition is a fundamental tool for analyzing multi-dimensional data by learning low-rank factors to represent high-order interactions. While recent works on temporal tensor decomposition have made significant progress by incorporating continuous timestamps in latent factors, they still struggle with general tensor data with continuous indexes not only in the temporal mode but also in other modes, such as spatial coordinates in climate data. Moreover, the challenge of self-adapting model complexity is largely unexplored in functional temporal tensor models, with existing methods being inapplicable in this setting. To address these limitations, we propose functional \underline{C}omplexity-\underline{A}daptive \underline{T}emporal \underline{T}ensor d\underline{E}composition (\textsc{Catte}). Our approach encodes continuous spatial indexes as learnable Fourier features and employs neural ODEs in latent space to learn the temporal trajectories of factors. To enable automatic adaptation of model complexity, we introduce a sparsity-inducing prior over the factor trajectories. We develop an efficient variational inference scheme with an analytical evidence lower bound, enabling sampling-free optimization. Through extensive experiments on both synthetic and real-world datasets, we demonstrate that \textsc{Catte} not only reveals the underlying ranks of functional temporal tensors but also significantly outperforms existing methods in prediction performance and robustness against noise.

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。

↑