I think you’ve got the L1 and L2 regularization switched at 6:00, since L1 reduces weights of the less important features thus making it a good approach for feature selection.
Interesting, that’s great to know! We’ll take a look at the lecture and see what we might need to do about getting anything corrected. Thanks for the heads up!
Sorry to reply to an old question but I’m a bit confused about this too. On the prior slide you say L2 measures complexity using squares of the weights and L1 measures absolute values but then you seem to say it is the other way around on the following slide @6 minutes
This is indeed wrong and should definitely be corrected!