Dynamical system parameter identification using deep recurrent cell networks: Which gated recurrent unit and when?

Date

2021-01-01

Author

Cifdaloz, Oguzhan
Akagündüz, Erdem

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

170
views

0
downloads

In this paper, we investigate the parameter identification problem in dynamical systems through a deep learning approach. Focusing mainly on second-order, linear time-invariant dynamical systems, the topic of damping factor identification is studied. By utilizing a six-layer deep neural network with different recurrent cells, namely GRUs, LSTMs or BiLSTMs; and by feeding input/output sequence pairs captured from a dynamical system simulator, we search for an effective deep recurrent architecture in order to resolve the damping factor identification problem. Our study’s results show that, although previously not utilized for this task in the literature, bidirectional gated recurrent cells (BiLSTMs) provide better parameter identification results when compared to unidirectional gated recurrent memory cells such as GRUs and LSTM. Thus, indicating that an input/output sequence pair of finite length, collected from a dynamical system and when observed anachronistically, may carry information in both time directions to predict a dynamical systems parameter.

URI

https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85110437514&origin=inward
https://hdl.handle.net/11511/91470

Journal

Neural Computing and Applications

DOI

https://doi.org/10.1007/s00521-021-06271-5

Collections

Graduate School of Informatics, Article

Suggestions

OpenMETU
Core

Dynamics of Hopfield-Type Neural Networks with Modulo Periodic Unpredictable Synaptic Connections, Rates and Inputs Akhmet, Marat; Tleubergenova, Madina; Zhamanshin, Akylbek (2022-11-01) In this paper, we rigorously prove that unpredictable oscillations take place in the dynamics of Hopfield-type neural networks (HNNs) when synaptic connections, rates and external inputs are modulo periodic unpredictable. The synaptic connections, rates and inputs are synchronized to obtain the convergence of outputs on the compact subsets of the real axis. The existence, uniqueness, and exponential stability of such motions are discussed. The method of included intervals and the contraction mapping princip...
ADAPTIVE SYMMETRIC INTERIOR PENALTY GALERKIN METHOD FOR BOUNDARY CONTROL PROBLEMS BENNER, Peter; Yücel, Hamdullah (2017-01-01) We investigate an a posteriori error analysis of adaptive finite element approximations of linear-quadratic boundary optimal control problems under bilateral box constraints, which act on a Neumann boundary control. We use a symmetric interior Galerkin method as discretization technique. An efficient and reliable residual-type error estimator is introduced by invoking data oscillations. We then derive local upper and lower a posteriori error estimates for the boundary control problem. Adaptive mesh refineme...
Dynamical Gene-Environment Networks Under Ellipsoidal Uncertainty: Set-Theoretic Regression Analysis Based on Ellipsoidal OR Kropat, Erik; Weber, Gerhard Wilhelm; Belen, Selma (2008-09-12) We consider dynamical gene-environment networks under ellipsoidal uncertainty and discuss the corresponding set-theoretic regression models. Clustering techniques are applied for an identification of functionally related groups of genes and environmental factors. Clusters can partially overlap as single genes possibly regulate multiple groups of data items. The uncertain states of cluster elements are represented in terms of ellipsoids referring to stochastic dependencies between the multivariate data varia...
Learning Parameters of ptSTL Formulas with Backpropagation Ketenci, Ahmet; Aydın Göl, Ebru (2020-01-01) In this paper, a backpropagation based algorithm is presented to learn parameters of past time Signal Temporal Logic (ptSTL) formulas. A differentiable weight matrix over the parameter values and a loss function based on the mismatch value of the corresponding formulas over the labeled dataset are used in the algorithm. Analysis over a sample dataset shows that the algorithm solves the ptSTL parameter synthesis problem in an efficient way.
Multiple Model Adaptive Estimation Algorithm for Systems with Parameter Change Söken, Halil Ersin (2016-01-01) This paper presents an autonomous multiple model (AMM) estimation algorithm for systems with sudden parameter changes. Estimates of a bank of Kalman filters (KFs) are merged based on a newly defined likelihood function. The function is composed of two measures, one for weighting the filters during the steady state mode and the other for weighting when there is a change in the parameters. Compared to the interacting multiple model (IMM) method, the KFs do not interact but compete on the basis of the likeliho...

Citation Formats

O. Cifdaloz and E. Akagündüz, “Dynamical system parameter identification using deep recurrent cell networks: Which gated recurrent unit and when?,” Neural Computing and Applications, pp. 0–0, 2021, Accessed: 00, 2021. [Online]. Available: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85110437514&origin=inward.