3 ms·
Personally, I don't agree with a lot of Dr. Harell's definitions for SL and ML to begin with. He mostly talks about clinical trials, a field I don't have much k
by nafizh 7y ago
Personally, I don't agree with a lot of Dr. Harell's definitions for SL and ML to begin with. He mostly talks about clinical trials, a field I don't have much knowledge about. But a lot of the arguments seem to attack a strawman defined as ML.
- anthony_doan 7y agoOh I can see that. I think personally they're just pro and con for both. They're enough problems out there for both of them to coexist. With that said, I think the future of AI is going to be a hybrid of SL/ML as seen on m4 time series competition. ML may be just a stepping stone, like the two AI winter with expert system dying out. Or perhaps ML is going to just advance like NN did and went deep learning. Dr. Leo Breiman is correct in harshly criticizing SL but I believe both sides are going to have to take harsh objective criticisms for them to move forward.