n-takeishi/deepgreybox

Codes for "Deep Grey-Box Modeling With Adaptive Data-Driven Models Toward Trustworthy Estimation of Theory-Driven Models"

Takeishi and Kalousis, Deep Grey-Box Modeling With Adaptive Data-Driven Models Toward Trustworthy Estimation of Theory-Driven Models, arXiv:2210.13103.

Experiments

Toy dataset experiment (in Section 3.4 of the paper)

First, run the data-making scripts, the training script, and the prediction script by:

bash expt.sh toy1 demo adaptive 0 1.00e-02 cpu

The fourth argument of expt.sh is the random seed value, and the fifth argument is the value of $\lambda$.

Then, use toy1/notebooks/inspect_model.ipynb to inspect the results.

Controlled pendulum

bash expt.sh pendulum demo adaptive 0 1.00e-02 cpu

Then, use pendulum/notebooks/inspect_model.ipynb.

Reaction-diffusion system

bash expt.sh reaction-diffusion demo adaptive 0 1.00e-06 cuda:0

Then, use reaction-diffusion/notebooks/inspect_model.ipynb.

Predator-prey system

bash expt.sh predator-prey demo adaptive 0 1.00e-03 cpu

Then, use predator-prey/notebooks/inspect_model.ipynb.

Author

Naoya Takeishi - https://ntake.jp/