You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, it is possible to automatically construct one Simulink model for one training iteration. However, for each counterexample we might need to keep the previous neural net blocks either for testing or for debugging reasons.
This implies that we should either
Store each model/block as a new file and iterate over these files
Inside each file create a copy of the closed-loop system and replace the NN-block with the new NN (obtained after retraining). How to create a copy of the closed-loop programmatically via Matlab might not be easy (To-DO: add a new issue and link it)
The text was updated successfully, but these errors were encountered:
nikos-kekatos
changed the title
Constructing *several* new Simulink models/blocks in a loop (for counterexamples)
Constructing several new Simulink models/blocks in a loop (for counterexamples)
May 6, 2020
Currently, it is possible to automatically construct one Simulink model for one training iteration. However, for each counterexample we might need to keep the previous neural net blocks either for testing or for debugging reasons.
This implies that we should either
Store each model/block as a new file and iterate over these files
Inside each file create a copy of the closed-loop system and replace the NN-block with the new NN (obtained after retraining). How to create a copy of the closed-loop programmatically via Matlab might not be easy (To-DO: add a new issue and link it)
The text was updated successfully, but these errors were encountered: