You’ve heard a lot about Neural Networks (or “A.I.” as your company’s marketing team likes to call it), and how they are solution to all of your problems. Unfortunately, they’re also finicky and complex. And for most problems that most people deal with, they’re not necessarily much better than easier-to-use algorithms such as Gradient Boosted Trees (GBTs). We will explain (briefly) what GBTs are, compare neural networks vs. GBTs on some benchmark problems, demonstrate where Neural Nets are necessary (images/sound/text), and share some tips for making GBTs working even on those cases.
This talk is meant for people with a cursory knowledge of machine learning, but who aren’t necessarily experts.