NAOMI: Non-Autoregressive Multiresolution Sequence Imputation
NAOMI: Non-Autoregressive Multiresolution Sequence Imputation
复制标题
DOI：
--
复制
发表时间：
2019-01
期刊：
影响因子：
--
通讯作者：
Yukai Liu;Rose Yu;Stephan Zheng;Eric Zhan;Yisong Yue
中科院分区：
文献类型：
--
作者：
Yukai Liu;Rose Yu;Stephan Zheng;Eric Zhan;Yisong Yue
从运动跟踪到物理系统的动力学，缺失值补偿是时空建模中的一个基本问题。深度自回归模型受到误差传播的影响，这对于输入长程序列来说是灾难性的。在本文中，我们采用非自回归方法，提出了一种新的深度生成模型：非自回归多分辨率推算(NAOMI)，用于对给定任意缺失模式的长程序列进行推算。Naomi利用时空数据的多分辨率结构，并使用分而治之的策略从粗粒度到细粒度递归解码。我们通过对抗性训练进一步增强了我们的模型。当在来自确定性和随机动力学系统的基准数据集上进行广泛评估时。Naomi在长距离序列的推算精度(与自回归对应的平均预测误差相比减少了60%)和泛化方面表现出了显著的改进。
Missing value imputation is a fundamental problem in spatiotemporal modeling, from motion tracking to the dynamics of physical systems. Deep autoregressive models suffer from error propagation which becomes catastrophic for imputing long-range sequences. In this paper, we take a non-autoregressive approach and propose a novel deep generative model: Non-AutOregressive Multiresolution Imputation (NAOMI) to impute long-range sequences given arbitrary missing patterns. NAOMI exploits the multiresolution structure of spatiotemporal data and decodes recursively from coarse to fine-grained resolutions using a divide-and-conquer strategy. We further enhance our model with adversarial training. When evaluated extensively on benchmark datasets from systems of both deterministic and stochastic dynamics. NAOMI demonstrates significant improvement in imputation accuracy (reducing average prediction error by 60% compared to autoregressive counterparts) and generalization for long range sequences.