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

11–20 of 102 posts

Re: Deep Neural Networks for YouTube Recommendations

#11

Post by iverjo »

brianberns wrote: post #9
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 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.

Spotify's Discover Weekly is based on collaborative filtering. Some of the techniques they use are described in these slides: http://www.slideshare.net/MrChrisJohnson/collaborative-filte...

Re: Deep Neural Networks for YouTube Recommendations

#12

Post by placeybordeaux »

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.

I am currently looking for a decent comprehensive study on them. So far I have found this slide deck from netflix to be pretty good [1]. Do you have any suggestions?

[1] http://www.slideshare.net/xamat/recommender-systems-machine-...

Re: Deep Neural Networks for YouTube Recommendations

#13

Post by thomasahle »

Eliezer wrote: post #10

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 re…

Mine seem like they have gotten a lot more click-baity. I wonder if that's an artifact of their optimization goal.

Re: Deep Neural Networks for YouTube Recommendations

#14

Post by Rotonde »

Eliezer wrote: post #10

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 re…

That's curious, because in my opinion YouTube recommendations haven't been any good since 2009. Instead of getting interesting, strange and niche content, I'm bombarded with videos that have >100k views, feature clickbait titles and thumbnails and are generally incredibly low effort content.

Methods that work better for a population as a whole might not work better for a large subset of that population, and might even cause users to stop using features entirely. The lack of transparency in recommendation algorithms combined with the homogenizing effect of distributing low-quality content this way is something I find somewhat depressing.

Re: Deep Neural Networks for YouTube Recommendations

#15

Post by artifaxx »

thomasahle wrote: post #13
Eliezer wrote: post #10

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 re…

Mine seem like they have gotten a lot more click-baity. I wonder if that's an artifact of their optimization goal.

If people fall for the click bait(they do), then they get more views from optimizing for it. It would be more surprising if an optimization algorithm for views didn't favor this.

Re: Deep Neural Networks for YouTube Recommendations

#16

Post by erichocean »

iverjo wrote: post #11
brianberns wrote: post #9

Earlier quoted context omitted.

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.

Spotify's Discover Weekly is based on collaborative filtering. Some of the techniques they use are described in these slides: http://www.slideshare.net/MrChrisJohnson/collaborative-filte...

That's a major technique in one of the social platforms I'm working on, too. Glad to see it working so well at scale!

Re: Deep Neural Networks for YouTube Recommendations

#18

Post by robinduckett »

Rotonde wrote: post #14
Eliezer wrote: post #10

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 re…

That's curious, because in my opinion YouTube recommendations haven't been any good since 2009. Instead of getting interesting, strange and niche content, I'm bombarded with videos that have >100k views, feature clickbait titles and thumbnails and are generally incredibly low effort content. Methods that work better for a population as a whole might not work better for a large subset of that population, and might eve…

Maybe those are just the videos that you're statistically more likely to watch through to the end based on your viewing history...

Re: Deep Neural Networks for YouTube Recommendations

#20

Post by AndrewKemendo »

pcovington wrote: post #17

Author here - happy to answer questions about the techniques in the paper. We're super excited to finally share this work externally. Feedback about YouTube recommendations in general also welcome.

I'm most curious about this part:

n conjugation with other product areas across Google, YouTube has undergone a fundamental paradigm shift to- wards using deep learning as a general-purpose solution for nearly all learning problems.

Can you talk about how this works in practice? Is the deep learning group separate from other teams and then tackles problems from different areas as needed, or are there deep learning engineers in each project area that are building nets for each different area? Is the ML team also redesigning product architecture by building products around reinforcement learning?

Post reply on HN