ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift
Cited 63 time in scopus Share share facebook twitter linkedin kakaostory
Authors
Taesung Kim, Jinhee Kim, Yunwon Tae, Cheonbok Park, Jang-Ho Choi, Jaegul Choo
Issue Date
2022-04
Citation
International Conference on Learning Representations (ICLR) 2022, pp.1-25
Language
English
Type
Conference Paper
Abstract
Statistical properties such as mean and variance often change over time in time series, i.e., time-series data suffer from a distribution shift problem. This change in temporal distribution is one of the main challenges that prevent accurate timeseries forecasting. To address this issue, we propose a simple yet effective normalization method called reversible instance normalization (RevIN), a generally-applicable normalization-and-denormalization method with learnable affine transformation. The proposed method is symmetrically structured to remove and restore the statistical information of a time-series instance, leading to significant performance improvements in time-series forecasting, as shown in Fig. 1. We demonstrate the effectiveness of RevIN via extensive quantitative and qualitative analyses on various real-world datasets, addressing the distribution shift problem.
KSP Keywords
Affine Transformation, Normalization method, Over time, Real-world, Statistical information, Statistical properties, Time series data, Time-series forecasting, temporal distribution