• Corpus ID: 246210083

On the adaptation of recurrent neural networks for system identification

  title={On the adaptation of recurrent neural networks for system identification},
  author={Marco Forgione and Aneri Muni and Dario Piga and Marco Gallieri},
This paper presents a transfer learning approach which enables fast and efficient adaptation of Recurrent Neural Network (RNN) models of dynamical systems. A nominal RNN model is first identified using available measurements. The system dynamics are then assumed to change, leading to an unacceptable degradation of the nominal model performance on the perturbed system. To cope with the mismatch, the model is augmented with an additive correction term trained on fresh data from the new dynamic… 

Figures and Tables from this paper

Few-Shot Learning by Dimensionality Reduction in Gradient Space

The advantages of SubGD are experimentally corroborated on three distinct dynamical systems problem settings, outperforming popular few-shot learning methods both in terms of sample efficiency and performance.



Nonlinear state-space identification using deep encoder networks

A method that approximates the simulation loss by splitting the data set into multiple independent sections similar to the multiple shooting method allows for the use of stochastic gradient optimization methods which scale well with data set size and has a smoothing effect on the non-convex cost function.

Fast Adaptation with Linearized Neural Networks

This work proposes a technique for embedding these inductive biases of linearizations of neural networks into Gaussian processes through a kernel designed from the Jacobian of the network, which develops significant computational speed-ups based on matrix multiplies, including a novel implementation for scalable Fisher vector products.

Continuous-time system identification with neural networks: model structures and fitting criteria

Deep Identification of Nonlinear Systems in Koopman Form

The present paper treats the identification of nonlinear dynamical systems using Koopman-based deep state-space encoders using the the deepSI toolbox in Python to lower the computational need of the simulation error-based training.

Echo State Networks: analysis, training and predictive control

Numerical results concerning the predictive control of a nonlinear process for pH neutralization confirm the effectiveness of the proposed algorithms for the identification, dimensionality reduction, and the control design for ESNs.

Integrated Neural Networks for Nonlinear Continuous-Time System Identification

This letter introduces a novel neural network architecture, called Integrated Neural Network (INN), for direct identification of nonlinear continuous-time dynamical models in state-space representation and its effectiveness is assessed against the Cascaded Tanks System benchmark.

Deep Convolutional Networks in System Identification

This paper establishes connections between the deep learning and the system identification communities and explores the explicit relationships between the recently proposed temporal convolutional network (TCN) and two classic system identification model structures; Volterra series and block-oriented models.

dynoNet: A neural network architecture for learning dynamical systems

  • Marco ForgioneD. Piga
  • Computer Science
    International Journal of Adaptive Control and Signal Processing
  • 2021
The back‐propagation behavior of the linear dynamical operator with respect to both its parameters and its input sequence is defined, which enables end‐to‐end training of structured networks containinglinear dynamical operators and other differentiable units, exploiting existing deep learning software.

LSTM: A Search Space Odyssey

This paper presents the first large-scale analysis of eight LSTM variants on three representative tasks: speech recognition, handwriting recognition, and polyphonic music modeling, and observes that the studied hyperparameters are virtually independent and derive guidelines for their efficient adjustment.