Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Deep Neural Networks for YouTube Recommendations

research.google.com

Post reply on HN
Print view View original

1–10 of 102 posts

Re: Deep Neural Networks for YouTube Recommendations

#4

Post by taeric »

erichocean wrote: post #2

Recommendation systems are a really interesting topic to study/engineer on. I think there's a lot of unexplored/undiscovered techniques still remaining.

Oddly... I have grown rather weary of them. I have yet to get a surprise recommendation that I cared for. At best, I have seen people surprised that some good recommendations came out of a system.

Which usually just leads to curation systems being key. And they work well, until they are gamed. And they will be gamed.

Re: Deep Neural Networks for YouTube Recommendations

#5

Post by bfuller »

taeric wrote: post #4
erichocean wrote: post #2

Recommendation systems are a really interesting topic to study/engineer on. I think there's a lot of unexplored/undiscovered techniques still remaining.

Oddly... I have grown rather weary of them. I have yet to get a surprise recommendation that I cared for. At best, I have seen people surprised that some good recommendations came out of a system. Which usually just leads to curation systems being key. And they work well, until they are gamed. And they will be gamed.

I live with my elderly father. For some reason Facebook and Amazon think I am a geriatric.

Re: Deep Neural Networks for YouTube Recommendations

#6

Post by dominotw »

bfuller wrote: post #5
taeric wrote: post #4

Earlier quoted context omitted.

Oddly... I have grown rather weary of them. I have yet to get a surprise recommendation that I cared for. At best, I have seen people surprised that some good recommendations came out of a system. Which usually just leads to curation systems being key. And they work well, until they are gamed. And they will be gamed.

I live with my elderly father. For some reason Facebook and Amazon think I am a geriatric.

>For some reason Facebook and Amazon think I am a geriatric.

Heh. Me too. I suffer from joint/ligament issues and buy supplements for those.Now, Amazon thinks I am a geriatric and recommends me 'helping hand' sticks and incontinence products :\

Reminds me o this tweet

https://twitter.com/kibblesmith/status/724817086309142529

  Amazon is a $250 billion dollar company that reacts to you buying a vacuum by going THIS GUY LOVES BUYING VACUUMS HERE ARE SOME MORE VACUUMS

Re: Deep Neural Networks for YouTube Recommendations

#8

Post by erichocean »

taeric wrote: post #4
erichocean wrote: post #2

Recommendation systems are a really interesting topic to study/engineer on. I think there's a lot of unexplored/undiscovered techniques still remaining.

Oddly... I have grown rather weary of them. I have yet to get a surprise recommendation that I cared for. At best, I have seen people surprised that some good recommendations came out of a system. Which usually just leads to curation systems being key. And they work well, until they are gamed. And they will be gamed.

> Oddly... I have grown rather weary of them.

That's what makes the problem so interesting! Most recommendation systems are terrible, and those that aren't, are good only for the first 1-2 recommendations.

And then there's Google Search, which so thoroughly demolished existing search result recommendation systems (remember Altavista?) that they now own the market and are one of the most valuable companies in history.

When you finally solve a recommendation system problem in a way that actually works, it's a huge freaking deal!

Re: Deep Neural Networks for YouTube Recommendations

#9

Post by brianberns »

taeric wrote: post #4
erichocean wrote: post #2

Recommendation systems are a really interesting topic to study/engineer on. I think there's a lot of unexplored/undiscovered techniques still remaining.

Oddly... I have grown rather weary of them. I have yet to get a surprise recommendation that I cared for. At best, I have seen people surprised that some good recommendations came out of a system. Which usually just leads to curation systems being key. And they work well, until they are gamed. And they will be gamed.

I felt the same way until I tried Spotify's Discover Weekly recommendation system. I don't know what they're doing, but I've found many songs I really enjoy that way.

Re: Deep Neural Networks for YouTube Recommendations

#10

Post by Eliezer »

I noticed that YouTube's recommendations had suddenly gotten better! I wondered if they were using a new statistical approach, or had just started really optimizing at all because the old recommendations were extremely naive. I'm actually a little disappointed to find out that it might just be another deep learning thing. (Yes, it works, but I feel like you learn a little less about problem structure when what you read is mostly "we threw a generic function approximator and 30,000 hours of GPU time at it".)
Post reply on HN