Continuous-time Autoencoders for Regular and Irregular Time Series Imputation

Time series imputation is one of the most fundamental tasks for time series. Real-world time series datasets are frequently incomplete (or irregular with missing observations), in which case imputation is strongly required.