Modeling systems with machine learning based differential equations

The prediction of behavior in dynamical systems, is frequently associated to the design models. When a time series obtained from observing system available, task can be performed by designing model these observations without additional assumptions or assuming preconceived structure model, with help information about system. In second case, it question adequately combining theory and subsequently optimizing mixture. this work, we proposes time-continuous models systems as solutions differential equations, non-uniformly sampled noisy observations, using machine learning techniques. proposed approach, for models, simple interpret implement computationally its performance shown both, several simulated data sets experimental Hare–Lynx population Coronavirus 2019 outbreak. results suggest usefulness case synthetic real data, uniformly sampled.