Just because you can analyse it doesn't mean that it is better. Deep learning theory is unbelievably garbage compared to the empirical results. In particular,

Just because you can analyse it doesn't mean that it is better. Deep learning theory is unbelievably garbage compared to the empirical results. In particular, please show me a worked example of a decision tree meta learning. Because its trivial to show this for DNNs.