Time series prediction with simple recurrent neural networks

Abdulkarim, Salihu Aish

doi:10.4314/bajopas.v9i1.4

Cited by 14 publications

(7 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Unlike the JORDAN cell, each time step in the ELMAN cell calls the previous hidden state (Figure 7b). In 2016, a combination of both JORDAN and ELMAN cells in one cell named multi-recurrent neural network (MRNN) was evaluated by Abdulkarim (2016). In this cell structure, at each time step, both previous output and hidden states are presented to the cell (Figure 7c).…”

Section: Taxonomy Of Rnn Cellsmentioning

confidence: 99%

What is the best RNN-cell structure for forecasting each time series behavior?

Khaldi¹,

Afia²,

Chiheb³

et al. 2022

Preprint

View full text Add to dashboard Cite

It is unquestionable that time series forecasting is of paramount importance in many fields. The most used machine learning models to address time series forecasting tasks are Recurrent Neural Networks (RNNs).Typically, those models are built using one of the three most popular cells, ELMAN, Long-Short Term Memory (LSTM), or Gated Recurrent Unit (GRU) cells, each cell has a different structure and implies a different computational cost. However, it is not clear why and when to use each RNN-cell structure.Actually, there is no comprehensive characterization of all the possible time series behaviors and no guidance on what RNN cell structure is the most suitable for each behavior. The objective of this study is two-fold: it presents a comprehensive taxonomy of all-time series behaviors (deterministic, random-walk, nonlinear, long-memory, and chaotic), and provides insights into the best RNN cell structure for each time series behavior. We conducted two experiments: (1) The first experiment evaluates and analyzes the role of each component in the LSTM-Vanilla cell by creating 11 variants based on one alteration in its basic architecture (removing, adding, or substituting one cell component). ( 2) The second experiment evaluates and analyzes the performance of 20 possible RNN-cell structures. Our results showed that the MGU-SLIM3 cell is the most recommended for deterministic and nonlinear behaviors, the MGU-SLIM2 cell is the most suitable for random-walk behavior, FB1 cell is advocated for long-memory behavior, and LSTM-SLIM1 for chaotic behavior.

show abstract

Section: Taxonomy Of Rnn Cellsmentioning

confidence: 99%

What is the best RNN-cell structure for forecasting each time series behavior?

Khaldi¹,

Afia²,

Chiheb³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…RNNs represent one of the most optimal choices when working with data organized in time series models. Their work principle is based on combining nonlinear activation functions in a recurrent structure, which makes prediction possible and provides improved prediction accuracy, as stated in [1]. In contrast to the standard Neural Networks, which are usually represented using feedforward architectures, RNNs allow the information to be transferred both forward and backward, with the help of their feedback connections.…”

Section: Lstm-based Forecasting Of Electricity Consumption and Producmentioning

confidence: 99%

Modeling Electricity Consumption and Production in Smart Homes using LSTM Networks

Bachici

Gellért

2020

International Journal of Advanced Statistics and IT&C for Economics and Life Sciences

View full text Add to dashboard Cite

This paper presents a forecasting method of the electricity consumption and production in a household equipped with photovoltaic panels and a smart energy management system. The prediction is performed with a Long Short-Term Memory recurrent neural network. The datasets collected during five months in a household are used for the evaluations. The recurrent neural network is configured optimally to reduce the forecasting errors. The results show that the proposed method outperforms an earlier developed Multi-Layer Perceptron, as well as the Autoregressive Integrated Moving Average statistical forecasting algorithm.

show abstract

“…For time series forecasting, this work has explored various machine learning methods. These methods are divided into traditional statistical methods and deep learning-based techniques, such as Auto-Regressive Integrated Moving Average (ARIMA) [63], and recurrent neural networks including long short-term memory (LSTM) networks [64], bi-direction recurrent neural networks, Elman networks, Jordan networks [65], and recursive neural networks. Particularly, we have explored the ability of different network configurations to capture the behavior of the data, and their performance, measured in terms of the root mean squared error (RMSE), is shown in Table 4.…”

Section: A System Modelmentioning

confidence: 99%