Holistic nn modeling?

Can Tensorflow\ Keras be used for a synchronous \ holistic modeling off NN architecture\ structure, types of neurons, weights etc? Are there examples to show that?

Do you have any non Keras example/reference? Just to understand a little bit the topic.

We did some holistic nn modeling for some automotive companies, but when the know-how owners experienced the power and value of such models, they were not open for any publishing of this work…
Now we look for a common platform for our holistic modeling, and we just wonder if TensorFlow\ Keras could do the job…

Are you talking about this:

https://arxiv.org/abs/2110.15829

Hmm… we just try to optimize\ train the architecture, the size, and the weights of a model in an integrated “holistic” process, and I wonder if Tensorflow\ Keras can provide special support for such a strategy.
The resulting effects (eg “adversarial robustness, parameter sparsity, and output stability”) may be partially the same, but there is more…

Do you have any specific paper reference?

1 Like

P.s. If you are just looking instead for AutML you could check:

For cloud solutions:

thanks for your hints…