Deep Learning of Dynamic Systems using System Identification Toolbox(TM)
This work provides incremental improvements for engineers and researchers using MATLAB for dynamic system modeling, facilitating easier integration of deep learning techniques.
The paper describes enhancements to MATLAB's System Identification Toolbox over three years, focusing on integrating deep learning architectures and training techniques to use deep neural networks as building blocks for nonlinear models, including neural state-space models with auto-encoding features for reduced-order modeling of large systems.
MATLAB(R) releases over the last 3 years have witnessed a continuing growth in the dynamic modeling capabilities offered by the System Identification Toolbox(TM). The emphasis has been on integrating deep learning architectures and training techniques that facilitate the use of deep neural networks as building blocks of nonlinear models. The toolbox offers neural state-space models which can be extended with auto-encoding features that are particularly suited for reduced-order modeling of large systems. The toolbox contains several other enhancements that deepen its integration with the state-of-art machine learning techniques, leverage auto-differentiation features for state estimation, and enable a direct use of raw numeric matrices and timetables for training models.