How a Kaggle Grandmaster cheated at a machine learning competition and was caught by a 19 year old dev

How a Kaggle Grandmaster cheated at a machine learning competition and was caught by a 19 year old dev
0

I thought you all might find this interesting. Not just the human interest - 2,000 teams participated and have to be disappointed to find the winning team was a cheater.

But specifically, the way he cheated.

I won’t ruin the story for you. Give it a read. Does it surprise you that something like this got past the judges?

5 Likes

Vice declared this “worse than the college admissions scandal” which is a stretch, if I’m honest.

Competitions are results based: Steven Bradbury won a gold medal for speed skating in 2002 despite having been in last place–by quite a large distance–for the entire race [x]. Especially when it’s the first competition of it’s kind for the participating judges (as per the official announcement regarding the discovery of the fraud), and it is not required for the algorithms to be made public, it’s not at all surprising that no one looked for misconduct; I, too, try and assume the best of people and boy does it come back to bite you when you’re wrong.

The dude says they will return the prize money and H2O.ai is working with the NPO. So, the whole thing is a damn shame but at least some good is coming of it.

1 Like

Wow… All I can say is that I’ve learned a lot from his hack! Too bad he applied such good skills in such a bad way :frowning:

1 Like

Welcome to the community. Yes - it’s sad to see such talented people exhibit such moral failings. But I’m glad you learned a lot from it.

Are there any other hacks you’ve seen lately that you learned a lot from? If so, I’d encourage you to share them here so we can all learn from them and discuss them :slight_smile:

2 Likes

Sad that a guy with full time job at top AI company did this :frowning_face:

@Naveenkhasyap Agreed.

By the way, welcome back. I remember you from way back in the day.

Would you be interested in sharing some other ML-focused news stories here?