Long short Term Memory(LSTM) has been widely used in sequencial problems. However, for the time series prediction problems, its complex structure limits its running speed and performance. In order to solve this problem, this paper simplified the standard LSTM model by reducing the number of gates and the parameters involved in gates computation. Experiments on univariate data set and multivariate data set show that the proposed simplified model not only has better accuracy, but also has higher running speed.
Paper
Full text
Simplifying Long Short-Term Memory for Fast Training and Time Series Prediction
Semantic Scholar · Physics · 2019
Abstract
Long short Term Memory(LSTM) has been widely used in sequencial problems. However, for the time series prediction problems, its complex structure limits its running speed and performance. In order to solve this problem, this paper simplified the standard LSTM model by reducing the number of gates and the parameters involved in gates computation. Experiments on univariate data set and multivariate data set show that the proposed simplified model not only has better accuracy, but also has higher running speed.