Introduction to Adversarial Validation In Machine Learning.

[@muhammadawon8164]

The knowledge you possess is a gold mine. Thank you for sharing it with the world.

[@arvindodangeh]

You are explaining concept in amazing way and the fact that you are using body language and interesting gesture ,It's quit helpful .

[@odysseus9672]

"If your model can't tell these apart, that means they come from same distribution." Not quite. It could also be that your model is underpowered/inappropriate to the task. This is one of those cases where a positive result (the data can be distinguished) is conclusive, but a negative one is just indicative.

[@Mutual_Information]

Nice to see you on KZbin! I follow you on Twitter - you have a habit of positing quality info on ML, so I’m quick to subscribe. This video is quality. Good topic choice, well edited, no fluffy filler language and a colorful and flexible form of plots. I look at a lot of ML KZbin channels (I have one too), so I believe I have a good sense of what works and what doesn’t. To me, this has all the ingredients of a successful channel.

[@malik4islam]

Thanks for sharing! Any recommended textbook or publications, please.

[@viniciusmonteirodelira9872]

Keep it up with the great work!!!

[@lakeguy65616]

following your example, if you build & train a model with the y column "adversarial variable target", if the model trains well, doesn't that mean there is a significant difference between the training and test datasets? If, after many epochs, the loss fails to decline, doesn't that mean there is nothing that distinguishes the training data from the test data? If true, what's the point of determining the Area Under the Curve? (FYI, this is another GREAT video!)

[@iftikhar58]

happy to see you on youtube

[@underfitted]

Thanks

[@harish00784]

Amazing one, can't find such an info anywhere else. Thanks Santiago 👍🏼. I have to find such a dataset and work on it..

[@underfitted]

Happy to help!

[@nedafiroz514]

Thank you for so elaborate discussion, I love your channel ,and grateful for your illustrations as a novice learner

[@underfitted]

Thanks!

[@morthim]

really good video. it was concise, helpful, articulate, and pragmatic. gave a thumbs up and added it to a playlist to help with the algorythm. thank you for sharing this.

[@underfitted]

Thanks

[@greyhat_gaming]

Very helpful! Thank you!

[@a7med7x7]

amazing explantation!

[@funshoafolabi9586]

Good one, your videos and posts on twitter have been really helpful. can you recommend a book to learn more about machine learning?

[@AbhishekSinghSambyal]

Probabilistic machine learning by Murphy. The best book but a little advance.

[@underfitted]

Thanks, Funsho!

[@TheProtein83]

Nice video, thank you! Can you explain why we wouldn't reshuffle the data? I mean, usually you pick train and test set randomly. The procedure is repeated several times. This should avoid the problem you just described, right?

[@underfitted]

You can't always reshuffle the data. Look at this video: kzbin.info/www/bejne/h5DYl2x_oJeGaJo

[@itsm0saan]

Can please provide good resources to learn statistics and probability, Thanks^^

[@tex1297]

Actually you culd include te target column too. Maybe there is the difference between your sets.

[@abdelrahmanehab7304]

Tank you

[@premierjoseph9871]

Wow just wow🙌🏼. 🌟🌟🌟🌟🌟