The Experts below are selected from a list of 297 Experts worldwide ranked by ideXlab platform
Jochen J. Steil - One of the best experts on this subject based on the ideXlab platform.
-
Modelling of parametrized processes via regression in the Model Space of neural networks
Neurocomputing, 2017Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:Abstract We consider the Modelling of parametrized processes, where the goal is to Model the process for new parameter value combinations. We compare the classical regression approach to a modular approach based on regression in the Model Space: First, for each process parametrization a Model is learned. Second, a mapping from process parameters to Model parameters is learned. We evaluate both approaches on two synthetic and two real-world data sets and show the advantages of the regression in the Model Space.
-
Time Series Classification in Reservoir- and Model-Space
Neural Processing Letters, 2017Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:We evaluate two approaches for time series classification based on reservoir computing. In the first, classical approach, time series are represented by reservoir activations. In the second approach, on top of the reservoir activations, a predictive Model in the form of a readout for one-step-ahead-prediction is trained for each time series. This learning step lifts the reservoir features to a more sophisticated Model Space. Classification is then based on the predictive Model parameters describing each time series. We provide an in-depth analysis on time series classification in reservoir- and Model-Space. The approaches are evaluated on 43 univariate and 18 multivariate time series. The results show that representing multivariate time series in the Model Space leads to lower classification errors compared to using the reservoir activations directly as features. The classification accuracy on the univariate datasets can be improved by combining reservoir- and Model-Space.
-
ANNPR - Time Series Classification in Reservoir- and Model-Space: A Comparison
Artificial Neural Networks in Pattern Recognition, 2016Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:Learning in the Space of Echo State Network (ESN) output weights, i.e. Model Space, has achieved excellent results in time series classification, visualization and Modelling. This work presents a systematic comparison of time series classification in the Model Space and the classical, discriminative approach with ESNs. We evaluate the approaches on 43 univariate and 18 multivariate time series. It turns out that classification in the Model Space achieves often better classification rates, especially for high-dimensional motion datasets.
-
ESANN - Modelling of parameterized processes via regression in the Model Space.
2016Co-Authors: Witali Aswolinskiy, Felix Reinhart, Jochen J. SteilAbstract:We consider the Modelling of parameterized processes, where the goal is to Model the process for new parameter value combinations. We compare the classical regression approach to a modular approach based on regression in the Model Space: First, for each process parametrization a Model is learned. Second, a mapping from process parameters to Model parameters is learned. We evaluate both approaches on a real and a synthetic dataset and show the advantages of the regression in the Model Space.
-
Impact of Regularization on the Model Space for Time Series Classification
2015Co-Authors: Witali Aswolinskiy, Felix Reinhart, Jochen J. SteilAbstract:Time series classification is an active research field and applicable in many domains, e.g. speech and gesture recognition. A recent approach to classify time series is based on Modelling each time series by an Echo State Network and then to classify the time series in the readout weight or Model Space of these networks. In this paper, we investigate the effect of Echo State Network regularization on the Model Space. The results show that regularization has a strong impact on the Model Space structure and the separability of the time series in the Model Space.
Witali Aswolinskiy - One of the best experts on this subject based on the ideXlab platform.
-
Learning in the Model Space of Neural Networks
2018Co-Authors: Witali AswolinskiyAbstract:Learning from time series is in demand in many domains including finance, medicine and industry. Recently, a novel method for time series classification was proposed based on the idea of training self-predictive Models on the time series and classifying in the Space of the learned Model parameters - the Model Space. In this thesis, learning in the Model Space of neural networks is investigated and extended. First, an empirical investigation of time series classification and clustering in the Model Space is conducted. Based on experiments on numerous time series datasets, key aspects are identified and improvements proposed. Then, the underlying concept is extended to transfer learning for time series. A novel approach for unsupervised transfer learning using self-predictive Modelling is proposed. Finally, a modular framework for Modelling parameterized processes is defined. The proposed approaches are successfully validated on synthetic and real-world datasets.
-
Modelling of parametrized processes via regression in the Model Space of neural networks
Neurocomputing, 2017Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:Abstract We consider the Modelling of parametrized processes, where the goal is to Model the process for new parameter value combinations. We compare the classical regression approach to a modular approach based on regression in the Model Space: First, for each process parametrization a Model is learned. Second, a mapping from process parameters to Model parameters is learned. We evaluate both approaches on two synthetic and two real-world data sets and show the advantages of the regression in the Model Space.
-
Time Series Classification in Reservoir- and Model-Space
Neural Processing Letters, 2017Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:We evaluate two approaches for time series classification based on reservoir computing. In the first, classical approach, time series are represented by reservoir activations. In the second approach, on top of the reservoir activations, a predictive Model in the form of a readout for one-step-ahead-prediction is trained for each time series. This learning step lifts the reservoir features to a more sophisticated Model Space. Classification is then based on the predictive Model parameters describing each time series. We provide an in-depth analysis on time series classification in reservoir- and Model-Space. The approaches are evaluated on 43 univariate and 18 multivariate time series. The results show that representing multivariate time series in the Model Space leads to lower classification errors compared to using the reservoir activations directly as features. The classification accuracy on the univariate datasets can be improved by combining reservoir- and Model-Space.
-
ANNPR - Time Series Classification in Reservoir- and Model-Space: A Comparison
Artificial Neural Networks in Pattern Recognition, 2016Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:Learning in the Space of Echo State Network (ESN) output weights, i.e. Model Space, has achieved excellent results in time series classification, visualization and Modelling. This work presents a systematic comparison of time series classification in the Model Space and the classical, discriminative approach with ESNs. We evaluate the approaches on 43 univariate and 18 multivariate time series. It turns out that classification in the Model Space achieves often better classification rates, especially for high-dimensional motion datasets.
-
ESANN - Modelling of parameterized processes via regression in the Model Space.
2016Co-Authors: Witali Aswolinskiy, Felix Reinhart, Jochen J. SteilAbstract:We consider the Modelling of parameterized processes, where the goal is to Model the process for new parameter value combinations. We compare the classical regression approach to a modular approach based on regression in the Model Space: First, for each process parametrization a Model is learned. Second, a mapping from process parameters to Model parameters is learned. We evaluate both approaches on a real and a synthetic dataset and show the advantages of the regression in the Model Space.
Bartosz łanucha - One of the best experts on this subject based on the ideXlab platform.
-
matrix representations of truncated toeplitz operators
Journal of Mathematical Analysis and Applications, 2014Co-Authors: Bartosz łanuchaAbstract:Abstract In this paper we describe matrix representations of truncated Toeplitz operators on the Model Space K B , where B is an infinite Blaschke product satisfying some additional conditions. Our results are extensions of that obtained by Cima, Ross and Wogen in 2008.
Giuseppe Di Giulio - One of the best experts on this subject based on the ideXlab platform.
-
exploring the Model Space and ranking a best class of Models in surface wave dispersion inversion application at european strong motion sites
Geophysics, 2012Co-Authors: Giuseppe Di Giulio, Alexandros Savvaidis, Matthias Ohrnberger, Marc Wathelet, Cecile Cornou, Brigitte Knapmeyerendrun, F Renalier, N Theodoulidis, Pierreyves BardAbstract:ABSTRACTThe inversion of surface-wave dispersion curve to derive shear-wave velocity profile is a very delicate process dealing with a nonunique problem, which is strongly dependent on the Model Space parameterization. When independent and reliable information is not available, the selection of most representative Models within the ensemble produced by the inversion is often difficult. We implemented a strategy in the inversion of dispersion curves able to investigate the influence of the parameterization of the Model Space and to select a “best” class of Models. We analyzed surface-wave dispersion curves measured at 14 European strong-motion sites within the NERIES EC-Project. We focused on the inversion task exploring the Model Space by means of four distinct parameterization classes composed of layers progressively added over a half-Space. The classes differ in the definition of the shear-wave velocity profile; we considered Models with uniform velocity as well as Models with increasing velocity with d...
-
Inversion of surface wave dispersion at European strong motion sites using a multi-Model parameterization and an information-theoretic approach
2010Co-Authors: Giuseppe Di Giulio, Matthias Ohrnberger, Marc Wathelet, Cecile Cornou, F Renalier, A. Savvaidi, N. Theodoulidi, B. Endrun, P.y. BardAbstract:Within the scope of the EC-projects NERIES and ITSAK-GR we have applied a procedure able to combine a multi-Model Space parameterization and an information theoretic approach in analysis of dispersion curve inversion. In detail we considered the dispersion curve assessed at 14 strong motion European sites. At each site we investigated the Model Space through four different parameterization groups within the wavelength range estimated by actual dispersion curves. In order to explore the influence of Model Space we increased progressively the number of layers for each parameterization. We therefore addressed the Model evaluation among a set of competing Models obtained by inversion following the corrected Akaike's Information Criterion (AICc). By using such information-theoretic approach, we found an acceptable agreement between the inverted shear-velocity profiles of the best Models and the available borehole results.
René Felix Reinhart - One of the best experts on this subject based on the ideXlab platform.
-
Modelling of parametrized processes via regression in the Model Space of neural networks
Neurocomputing, 2017Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:Abstract We consider the Modelling of parametrized processes, where the goal is to Model the process for new parameter value combinations. We compare the classical regression approach to a modular approach based on regression in the Model Space: First, for each process parametrization a Model is learned. Second, a mapping from process parameters to Model parameters is learned. We evaluate both approaches on two synthetic and two real-world data sets and show the advantages of the regression in the Model Space.
-
Time Series Classification in Reservoir- and Model-Space
Neural Processing Letters, 2017Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:We evaluate two approaches for time series classification based on reservoir computing. In the first, classical approach, time series are represented by reservoir activations. In the second approach, on top of the reservoir activations, a predictive Model in the form of a readout for one-step-ahead-prediction is trained for each time series. This learning step lifts the reservoir features to a more sophisticated Model Space. Classification is then based on the predictive Model parameters describing each time series. We provide an in-depth analysis on time series classification in reservoir- and Model-Space. The approaches are evaluated on 43 univariate and 18 multivariate time series. The results show that representing multivariate time series in the Model Space leads to lower classification errors compared to using the reservoir activations directly as features. The classification accuracy on the univariate datasets can be improved by combining reservoir- and Model-Space.
-
ANNPR - Time Series Classification in Reservoir- and Model-Space: A Comparison
Artificial Neural Networks in Pattern Recognition, 2016Co-Authors: Witali Aswolinskiy, René Felix Reinhart, Jochen J. SteilAbstract:Learning in the Space of Echo State Network (ESN) output weights, i.e. Model Space, has achieved excellent results in time series classification, visualization and Modelling. This work presents a systematic comparison of time series classification in the Model Space and the classical, discriminative approach with ESNs. We evaluate the approaches on 43 univariate and 18 multivariate time series. It turns out that classification in the Model Space achieves often better classification rates, especially for high-dimensional motion datasets.
