Set Parameters operator for nested subprocess operator
Hello, let's get straight to topic of this post.
Below is the photo of the whole process, let me explain what I want to do:
1. I read my train data from .CSV file
2. Set role (label) for train data
3. Find optimal values for my Neural Net (learning rate, momentum, train cycles) with operator called "Optimize Parameters(Evolutionary)"
4. Apply model to test how well test data fits to train data
5. As you can see I pass output (lab) to operator called "Optimize Weights(Evolutionary)" which uses also Neural Net in its subprocess to find optimal weights then prints: output example set, optimized weights for every attribute and performance of this operator.
6. At the top you can see that "Optimizer Parameters(Evolutionary)" pass its optimal parameters (learning rate, momentum, train cycles) to operator called "Set Parameters".
My problem is that this "Set Parameters" operator can't assign parameters to Neural Net that is inside "Optimize Weights(Evolutionary)", because it can't see this Neural Net for some reason.
Thanks very much for anyone who know how to pass parameters to nested operators like "Optimize Weights(Evolutionary)" to Neural Net that is inside, because to achieve highest possible performance on my data I need first to optimize my Neural Net model, then pass its optimized parameters to Neural Net inside "Optimize Weights" operator.
Find more posts tagged with
Ok, problem solved. I was passing wrong parameters in "Set Parameter" operator. My source and target was incorrect. Anyway thanks for pointing to check tutorial about this operator, it really helped :smileyhappy:
Hi @Bambo
It would be beneficial if you could post the whole process here so it could be clear what's inside the operators, specifically, what's inside 'Set Parameters' and 'Optimize Weights'. Also maybe it could be helpful to study the example tutorila process for 'Set Parameters' operator which mught give some clues.
From what I see on a screenshot in your process structure, usage of 'Set Parameters' might not be mandatory. 'Optimize Parameters' operator already designed the way that it returns the optimized model, and you pass that model to 'Apply Model' operator input port, which means at least at this point the it applies the already optimized model. But again, to come up with more meaningful advise it would be helpful to see the whole process and insides of nested operators.
Hi @Bambo
It would be beneficial if you could post the whole process here so it could be clear what's inside the operators, specifically, what's inside 'Set Parameters' and 'Optimize Weights'. Also maybe it could be helpful to study the example tutorila process for 'Set Parameters' operator which mught give some clues.
From what I see on a screenshot in your process structure, usage of 'Set Parameters' might not be mandatory. 'Optimize Parameters' operator already designed the way that it returns the optimized model, and you pass that model to 'Apply Model' operator input port, which means at least at this point the it applies the already optimized model. But again, to come up with more meaningful advise it would be helpful to see the whole process and insides of nested operators.