Visualization of approximation function learned through ANN for a regression problem.
The ANN has 5 hidden layers with 20 neurons in each layer.
Dear J. P. Panda ,
There was a discussion on a similar question that you can refer to: https://www.researchgate.net/post/How_do_you_get_output_equation_with_ANN_modelization_using_Matlab
P.S. : As far as I know, ANNs suggest a surrogate model at the end. In terms of interpretability, I think it's better to use traditional ML Regression algos.
with bests,
Mansurbek Abdullaev
Hello Jyoti, 5 hidden layers, 20 neurons in each, means that there are 100 neurons in hidden layers and they all have a certain formula, so the questions is.... is it a good strategy to try to extract them, instead, maybe turn to other metrics depending on the goal you are trying to achieve. For example if the goal is to explain variable influence on predictions, you can try calculating the Gini index Mean decrease for variables excluded in ANN .
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts