Skcikit learn has a very beautiful graph for choosing a right solution for different cases/tasks/datasets. http://scikit-learn.org/stable/tutorial/machine_learning_map/index.html
Is there any similar graphic or textual guide for ANNs (and Tensorflow as a particular case)? For example what type of NN should I choose, what activation and optimization fucntions should I use, what number of neurons and layers supposed to be optimal for the task etc?
I understand that the selection highly depends on many factors and any guide cannot cover all cases and details, but I mean some general directions like it was done for scikit learn.