Probabilistic methods like naive bayes, decision trees and kernel based methods like SVM offer much simple ways to solve the same classification/regression problem while being interpretable. Derivative based methods still work as black box. I would argue to not skip learning any of these and always start with simpler solution whenever attempting any problem. Nice video though!
I will agree to that but the Goal for this video is to give a roadmap to quickly excel in ML in short time, more like what what are 20 percent things which will give 80 percent outcome. And all the derivative based methods are pre-requisite and stepping stone for each other thus makes learning very fast. While I don't recommend skipping anything, but I recommend the order where you get you most bang for your buck time wise.
Probabilistic methods like naive bayes, decision trees and kernel based methods like SVM offer much simple ways to solve the same classification/regression problem while being interpretable. Derivative based methods still work as black box. I would argue to not skip learning any of these and always start with simpler solution whenever attempting any problem.
Nice video though!
I will agree to that but the Goal for this video is to give a roadmap to quickly excel in ML in short time, more like what what are 20 percent things which will give 80 percent outcome. And all the derivative based methods are pre-requisite and stepping stone for each other thus makes learning very fast. While I don't recommend skipping anything, but I recommend the order where you get you most bang for your buck time wise.
Integration, differentiation, logarithm