Homotopy-based training of NeuralODEs for accurate dynamics discovery
Abstract Neural Ordinary Differential Equations (NeuralODEs) present an attractive way to
extract dynamical laws from time series data, as they bridge neural networks with the …
extract dynamical laws from time series data, as they bridge neural networks with the …
Differentiable homotopy methods for gradually reinforcing the training of fully connected neural networks
Deep fully connected neural networks (FCNNs) are the workhorses of deep learning and are
broadly applicable due to their “agnostic” structure. Generally, the learning capability of …
broadly applicable due to their “agnostic” structure. Generally, the learning capability of …
Control Refinement Using Particle Methods
S Zia, A Qayyum, FM Malik, MB Malik - IEEE Access, 2022 - ieeexplore.ieee.org
The paper presents a control scheme for the real-time tracking problem of nonlinear systems
subjected to hard nonlinearities. The proposed tracking controller introduces a refining …
subjected to hard nonlinearities. The proposed tracking controller introduces a refining …
Analysis and Control of Linear Time Periodic System using Normal Forms
S Cherangara Subramanian, S Redkar - International Journal of Dynamics …, 2022 - Springer
Multiple techniques have been developed in the past towards stability and control of linear
time periodic systems. Though the method of normal forms was predominantly applied to …
time periodic systems. Though the method of normal forms was predominantly applied to …