ANN models and the identifiability problem

Dear community,

In general, ANN models are non-identified. But, we attempt to address the identifiability problem by imposing some constraints. How can I impose these constraints using tf.keras.constraints?

Someone already did this?


If you are looking for how to just handle some constrains optimizzation you could take a look at:

Google AI Blog: Setting Fairness Goals with the TensorFlow Constrained Optimization Library

1 Like