Recurrent sparse support vector regression machines trained by active learning in the time-domain

Čeperić, Vladimir; Gielen, Georges; Barić, Adrijan

doi:10.1016/j.eswa.2012.03.031

Cited by 15 publications

(8 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(11) the term u(n) and the uncertainty of its random part might lead to misleading conclusions and an inaccurate comparison of the results. Therefore, it is decided to test the algorithms on clearly defined data sets that are easily reproducible, based on the data generation procedure described in [17]. The data generation in this experiment is based on pseudo-random number generation with the following setup [17]:…”

Section: Th-order Nonlinear Autoregressive Moving Average Systemmentioning

confidence: 99%

Reducing Complexity of Echo State Networks with Sparse Linear Regression Algorithms

Čeperić

Barić

2014

2014 UKSim-AMSS 16th International Conference on Computer Modelling and Simulation

View full text Add to dashboard Cite

In this paper the use of sparse linear regression algorithms in echo state networks (ESN) is presented for reducing the number of readouts and improving the robustness and generalization properties of ESNs. Three data sets with overall 80 tests are used to validate the use of sparse linear regression algorithms for echo state networks. It is shown that it is possible to increase accuracy on the test data sets, not used in the ESN training phase, and in the same time reduce the overall number of the required readouts when compared to the standard approach of using ridge linear regression on the echo state network readouts.

show abstract

Section: Th-order Nonlinear Autoregressive Moving Average Systemmentioning

confidence: 99%

Reducing Complexity of Echo State Networks with Sparse Linear Regression Algorithms

Čeperić

Barić

2014

2014 UKSim-AMSS 16th International Conference on Computer Modelling and Simulation

View full text Add to dashboard Cite

show abstract

“…We analyze the performances of SOPNN on multiple superimposed oscillations (MSO) task that has already been attempted with varying degrees of success by using ANN and SVM (Xue, Yang & Haykin, 2007;Schmidhuber, Wierstra, Gagliolo & Gomez, 2007;Holzmann & Hauser, 2010;Ceperic, Gielen & Baric, 2012). The following multiple sinusoids are modeled: where n=1,…,700.…”

Section: Multiple Superimposed Oscillations Modelingmentioning

confidence: 99%

“…Note that the frequencies of the sinusoids are not integer multiples of each other. As described in (Ceperic, Gielen & Baric, 2012) the first 400 samples (n=1,...,400) are used to train the model, while the rest of data (n=401,...,700) is used to test the model. The data is generated in double floating point (FP) number precision.…”

Section: Multiple Superimposed Oscillations Modelingmentioning

confidence: 99%

“…In the worst case the maximum total number of 100 FP additions and 100 FP multiplications would be necessary to calculate the SOPNN output, which is far below the computational complexities of the corresponding models described in the above mentioned references. As in (Ceperic, Gielen & Baric, 2012), we limited the maximum dimension (D) of the input space to 50 what corresponds to maximum 50 delayed output samples in the recurrent configuration. Table 3 shows the mean square error (MSE) and the normalized root mean square error (NRMSE) (Holzmann & Hauser, 2010) on the test data set, obtained by the non-optimized GMDH model (SOPNN) and by the same model after it has been optimized by the proposed LM algorithm (SOPNN-LM).…”

Section: Multiple Superimposed Oscillations Modelingmentioning

confidence: 99%

“…The huge complexity of their models cannot be compared with the simplicity of SOPNN models. Ceperic, Gielen & Baric, 2012, use recurrent sparse support vector regression machines trained by active learning in time-domain combined with the optimization of SVM hyper parameters to model the MSO data sets. They reported the following errors:…”

Section: Tablementioning

confidence: 99%

See 2 more Smart Citations

Optimization of self-organizing polynomial neural networks

Marić

2013

Expert Systems with Applications

View full text Add to dashboard Cite

The main disadvantage of self-organizing polynomial neural networks (SOPNN) automatically structured and trained by the group method of data handling (GMDH) algorithm is a partial optimization of model weights as the GMDH algorithm optimizes only the weights of the topmost (output) node. In order to estimate to what extent the approximation accuracy of the obtained model can be improved the particle swarm optimization (PSO) has been used for the optimization of weights of all node-polynomials. Since the PSO is generally computationally expensive and time consuming a more efficient Levenberg-Marquardt (LM) algorithm is adapted for the optimization of the SOPNN. After it has been optimized by the LM algorithm the SOPNN outperformed the corresponding models based on artificial neural networks (ANN) and support vector method (SVM). The research is based on the meta-modeling of the thermodynamic effects in fluid flow measurements with time-constraints. The outstanding characteristics of the optimized SOPNN models are also demonstrated in learning the recurrence relations of multiple superimposed oscillations (MSO). KeywordsPolynomial neural networks GMDH Levenberg-Marquardt algorithm Particle swarm optimization Time series modeling 3

show abstract