Advanced Methods for Time Series Prediction Using Recurrent Neural Networks

Boné, Romuald; Cardot, Hubert

doi:10.5772/16015

Cited by 17 publications

(18 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The total number of inputs is p + m, the number of hidden neurons is q, while φ is the context unit associated with the recurrent connection that keeps the content of output from the previous training stage [9]. It represents the long-term memory of the network [34]. This type of structure is convenient to derive some special cases.…”

Section: Modeling Strategy and Datamentioning

confidence: 99%

Can Recurrent Neural Networks Predict Inflation in Euro Zone as Good as Professional Forecasters?

Šestanović

Arnerić

2021

Mathematics

View full text Add to dashboard Cite

This paper investigates whether a specific type of a recurrent neural network, in particular Jordan neural network (JNN), captures the expected inflation better than commonly used feedforward neural networks and traditional parametric time-series models. It also considers competing survey-based and model-based expected inflation towards ex-post actual inflation to find whose predictions are more accurate; predictions from survey respondents or forecasting modelers. Further, it proposes neural network modelling strategy when dealing with nonstationary time-series which exhibit long-memory property and nonlinear dependence with respect to lagged inputs and exogenous inputs as well. Following this strategy, overfitting problem was reduced until no improvement in forecasting accuracy of expected inflation is achieved. The main finding is that JNN predicts inflation in euro zone quite accurately within forecasting horizon of 2 years. Regarding rational expectation principle we have found a set of demand-pull and cost-push inflation characteristics as exogenous inputs which helps in reducing overfitting problem of recurrent neural network even more. The sample includes euro zone aggregated monthly observations from January 2000 to December 2019. The results also confirm that inflation expectations obtained from JNN are consistent with Survey of professional forecasters (SPF), and thus, monetary policy makers can use JNN as a complementary tool in shortcomings of other inflation expectations measures.

show abstract

Section: Modeling Strategy and Datamentioning

confidence: 99%

Can Recurrent Neural Networks Predict Inflation in Euro Zone as Good as Professional Forecasters?

Šestanović

Arnerić

2021

Mathematics

View full text Add to dashboard Cite

show abstract

“…1   is an integer. We used the following parameters: [4]. 500 values were used for training; the next 100 values were used for testing.…”

Section: Mackey-glass Chaotic Processmentioning

confidence: 99%

“…Neural networks are known as an effective and friendly tool for black-box modeling of unknown plant's dynamics [1]. Usually, neural networks are trained to perform single-step-ahead (SS) predictions, where the predictor uses some available input and output observations to estimate the variable of interest for the time step immediately following the latest observation [2][3][4]. However, recently there has been growing interest in multi-step-ahead (MS) predictions, where the values of interest must be predicted for some horizon in the future.…”

Section: Introductionmentioning

confidence: 99%

“…All of the above reasons prevent RNNs from becoming widely popular in industry. Meanwhile, RNNs have inspired a new family of methods for training FFNNs to perform MS predictions called direct methods [4]. Accumulated error is backpropagated through an unfolded through time FFNN in BPTT style that causes minimization of the MS prediction error.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Direct Method for Training Feed-Forward Neural Networks Using Batch Extended Kalman Filter for Multi-Step-Ahead Predictions

Chernodub

2013

Artificial Neural Networks and Machine Learning – ICANN 2013

View full text Add to dashboard Cite

This paper is dedicated to the long-term, or multi-step-ahead, time series prediction problem. We propose a novel method for training feed-forward neural networks, such as multilayer perceptrons, with tapped delay lines. Special batch calculation of derivatives called Forecasted Propagation Through Time and batch modification of the Extended Kalman Filter are introduced. Experiments were carried out on well-known timeseries benchmarks, the Mackey-Glass chaotic process and the Santa Fe Laser Data Series. Recurrent and feed-forward neural networks were evaluated.

show abstract

“…Since RNNs are able to simulate any open dynamical system, they have a broad spectrum of applications such as time series forecasting [2], control of plants [3], language modeling [4], speech recognition, neural machine translation [5] and other domains. The easiest way to create an RNN is adding the feedback connections to the hidden layer of multilayer perceptron.…”

Section: Introductionmentioning

confidence: 99%

Sampling-Based Gradient Regularization for Capturing Long-Term Dependencies in Recurrent Neural Networks

Chernodub

Nowicki

2016

Neural Information Processing

View full text Add to dashboard Cite

Abstract. Vanishing (and exploding) gradients effect is a common problem for recurrent neural networks which use backpropagation method for calculation of derivatives. We construct an analytical framework to estimate a contribution of each training example to the norm of the longterm components of the target functions gradient and use it to hold the norm of the gradients in the suitable range. Using this subroutine we can construct mini-batches for the stochastic gradient descent (SGD) training that leads to high performance and accuracy of the trained network even for very complex tasks. To check our framework experimentally we use some special synthetic benchmarks for testing RNNs on ability to capture long-term dependencies. Our network can detect links between events in the (temporal) sequence at the range 100 and longer.

show abstract

Advanced Methods for Time Series Prediction Using Recurrent Neural Networks

Cited by 17 publications

References 33 publications

Can Recurrent Neural Networks Predict Inflation in Euro Zone as Good as Professional Forecasters?

Can Recurrent Neural Networks Predict Inflation in Euro Zone as Good as Professional Forecasters?

Direct Method for Training Feed-Forward Neural Networks Using Batch Extended Kalman Filter for Multi-Step-Ahead Predictions

Sampling-Based Gradient Regularization for Capturing Long-Term Dependencies in Recurrent Neural Networks

Contact Info

Product

Resources

About