A Guide to Cross-Validation for AI (with Dr. Tyler Bradshaw)

Chapter breakdowns:
0:00 Introduction
3:09 Overfitting vs. generalizability
6:24 Pitfalls of using one-time split method
6:44 Pitfall #1: Non-representative test set
11:24 Pitfall #2: Tuning to the test set
19:12 Cross-validation
20:10 Important note: in CV we are testing pipeline, not a single model
22:25 K-fold, folded test set
27:30 K-fold, hold-out test-set
32:56 Nested cross-validation
41:29 leave-one-out
41:45 random sampling
44:55 selecting an approach: pros and cons
46:48 Final thoughts
Reference paper:
pubs.rsna.org/doi/abs/10.1148...

[@AndrewJReader]

Great video Arman (and Tyler), very helpful!