Evaluation of Deep Learning Models for Multi-Step Ahead Time Series Prediction

Time series prediction with neural networks has been the focus of much research in past few decades. Given recent deep learning revolution, there attention using models for time prediction, and hence it is important to evaluate their strengths weaknesses. In this paper, we present an evaluation study that compares performance multi-step ahead prediction. The methods comprise simple recurrent networks, long short-term memory (LSTM) bidirectional LSTM encoder-decoder convolutional networks. We provide a further comparison use stochastic gradient descent adaptive moment estimation (Adam) training. on univariate multi-step-ahead from benchmark time-series datasets results related literature. show network provides best accuracy given problems.