×

NAOMI

swMATH ID: 43099
Software Authors: Yukai Liu, Rose Yu, Stephan Zheng, Eric Zhan, Yisong Yue
Description: NAOMI: Non-Autoregressive Multiresolution Sequence Imputation. 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
Homepage: https://arxiv.org/abs/1901.10946
Source Code:  https://github.com/felixykliu/NAOMI
Dependencies: Python
Related Software: E2EGAN; GP-VAE; matrix-completion; BRITS; GitHub; imputeTS; missForest; CRAN
Cited in: 1 Publication

Cited in 1 Serial

1 Machine Learning

Citations by Year