Time-Series Forecasting of Indoor Temperature Using Pre-trained Deep Neural Networks

Romeu, Pablo; Zamora-Martínez, Francisco; Botella-Rocamora, Paloma; Pardo, Juan

doi:10.1007/978-3-642-40728-4_57

Cited by 46 publications

(27 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Stacked denoising auto-encoders (SDAEs) [32] have been found useful and simpler, because they are based on standard GD algorithms and almost any ANN toolkit can deal with them. This paper is an extension of a previous work [27] where minor differences has been observed comparing standard ANNs with deep ANNs pre-trained with SDAEs for time series forecasting. A new set of experiments, with different and larger data, besides an analysis and discussion of the learned models, comparing deep and shallow architectures, has been done for this extension.…”

Section: Introductionmentioning

confidence: 89%

Stacked Denoising Auto-Encoders for Short-Term Time Series Forecasting

Romeu

Zamora-Martínez

Botella-Rocamora

et al. 2015

Springer Series in Bio-/Neuroinformatics

Self Cite

View full text Add to dashboard Cite

Abstract. In this chapter, a study of deep learning of time-series forecasting techniques is presented. Using Stacked Denoising Auto-Encoders, it is possible to disentangle complex characteristics in time series data. The effects of complete and partial fine-tuning are shown. SDAE prove to be able to train deeper models, and consequently to learn more complex characteristics in the data. Hence, these models are able to generalize better. Pre-trained models show a better generalization when used without covariates. The learned weights show to be sparse, suggesting future exploration and research lines.

show abstract

Section: Introductionmentioning

confidence: 89%

Stacked Denoising Auto-Encoders for Short-Term Time Series Forecasting

Romeu

Zamora-Martínez

Botella-Rocamora

et al. 2015

Springer Series in Bio-/Neuroinformatics

Self Cite

View full text Add to dashboard Cite

show abstract

“…Some types of deep ANNs are constructed by stacking multiple building blocks, such as auto-encoders [27]. The effectiveness of these auto-encoder ANNs have been demonstrated in temperature prediction [28], weather forecasting [29], prediction of traffic flows [30] or prediction of power consumption in a data centre [31]. The comparison presented in [13] shows that FARIMA processes and ANNs have similar approximation errors.…”

Section: Related Workmentioning

confidence: 99%

Traffic prediction methods for quality improvement of adaptive video

Biernacki

2017

Multimedia Systems

View full text Add to dashboard Cite

During the past years, adaptive video based on HTTP has become very popular. Streaming of the adaptive video relies heavily on an estimation of end-to-end network throughput, which can be challenging especially in mobile networks, where the capacity highly fluctuates. In this work, we propose to predict the network throughput using its past measurements. As the analysis shows, the network throughput forms a long range-dependent process; thus, for the throughput prediction, we apply a fractional ARIMA process and artificial neural networks. Our approach does not require any modifications to the network infrastructure or the TCP stack. The predictions are performed for data traces obtained from measurements of throughput of a real mobile network. As the experiment shows, the obtained traffic models are able to enhance the performance of an adaptive streaming algorithm. Compared to the throughput predictors employed in contemporary systems dedicated to adaptive video streaming, the proposed technique obtains better results when taking into account effectiveness of network capacity utilisation and stability of video play-out.

show abstract

“…data from several correlated signals (multivariate forecasting). Artificial neural networks (ANNs) have been widely applied to this task [9], [12].…”

Section: Introductionmentioning

confidence: 99%

A 24h forecast of solar irradiance using echo state neural networks

Kmeť

Kmetova

2015

Proceedings of the 16th International Conference on Engineering Applications of Neural Networks (INNS)

View full text Add to dashboard Cite

Artificial neural networks have demonstrated to be good at timeseries forecasting problems, being widely studied in literature. In this study an artificial neural network model is introduced for modelling the solar irradiance. Forecasting of solar irradiance is in general significant for planning the operations of power plants which convert renewable energies into electricity. In particular, the possibility to predict the solar irradiance (up to 24h or even more) can become fundamental in making power dispatching plans. In this paper, a practical method for solar irradiance forecast using artificial neural network is presented. The proposed echo state neural networks model makes it possible to forecast the solar irradiance on the base of 24h using the present values of the mean hours solar irradiance, air temperature, humidity, and the wind speed. An experimental database of solar irradiance, air temperature, humidity, wind speed data (from October 17th 2013 to May 11th 2015) has been used. The database has been collected in Lucenec (48.33N 19.67E), Slovakia.The results indicate that the proposed model performs well, while the correlation coefficient between measured and forecasted solar irradiance is in the range 0.94 − 0.97.

show abstract

Time-Series Forecasting of Indoor Temperature Using Pre-trained Deep Neural Networks

Cited by 46 publications

References 10 publications

Stacked Denoising Auto-Encoders for Short-Term Time Series Forecasting

Stacked Denoising Auto-Encoders for Short-Term Time Series Forecasting

Traffic prediction methods for quality improvement of adaptive video

A 24h forecast of solar irradiance using echo state neural networks

Contact Info

Product

Resources

About