Memory-Gated Recurrent Networks

The essence of multivariate sequential learning is all about how to extract dependencies in data. These data sets, such as hourly medical records intensive care units and multi-frequency phonetic time series, often exhibit not only strong serial the individual components (the "marginal" memory) but also non-negligible memories cross-sectional "joint" memory). Because complexity evolution joint distribution that underlies generating process, we take a data-driven approach construct novel recurrent network architecture, termed Memory-Gated Recurrent Networks (mGRN), with gates explicitly regulating two distinct types memories: marginal memory memory. Through combination comprehensive simulation studies empirical experiments on range public datasets, show our proposed mGRN architecture consistently outperforms state-of-the-art architectures targeting series.