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

31–40 of 102 posts

Re: Deep Neural Networks for YouTube Recommendations

#31

Post by andreyk »

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.

"To correct for this, we feed the age of the training example as a feature during training"

Does this mean something different from feeding the age of the video, relative to when the training example was recorded? Feeding in the age of the video seems like a fairly obvious idea and like it should train the network to favor newer videos. If it actually means how long ago the training example was recorded that is rather strange, as I don't see how that would be needed on top of the video age. Neat graph, there.

I am often annoyed at how overly focused online recommendations systems are for my overly specific recent trends, rather than broader interests I display over months or years of using a product (looking at you Amazon). It seems like it should be relatively easy to learn 'this guy likes little video essays about art and science and sometimes fun talk shows' and yet YouTube has been pretty bad at recommending such video-essay style content to me. Perhaps this will improve it, although I wonder how much the recent history features end up overwhelming overall years-long type data about what interests me broadly and not just yesterday.

As an aside, is it really "Deep Neural Networks for YouTube Recommendations" if you are using 5-ish layers of embedding, ReLu units, and output? A bit humorous, that.

Re: Deep Neural Networks for YouTube Recommendations

#32

Post by 5olidor »

alphydan wrote: post #24
robinduckett wrote: post #18

Earlier quoted context omitted.

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

Precisely. But that might be a local minimum. "Show him boobs and action trailers" is guaranteed to make him stay another 40min. But perhaps there is a more risky strategy that takes longer to craft and actually delivers hours and hours of content to the user (but needs to fail longer before getting there).

It seems like reinforcement learning would be useful, i.e. at a high level, forming a policy for recommendations would require balancing exploration (experimenting with more risky recommendations) vs. exploitation (showing you recommendations that it knows will likely lead to clicks) and using the click-throughs, time spent watching the video, etc. as reward signals.

Does anyone know whether RL is used for recommendation in practical settings, and if so what is the current state of the art?

Re: Deep Neural Networks for YouTube Recommendations

#33

Post by pcovington »

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…

YouTube has used machine learning in recommendations for many years.

We have struggled with interpretability, both while debugging mistakes made by the system and exposing plausible "reasons" to users.

There was a fascinating discussion [1] about interpretability during a deep learning panel at KDD this year.

[1] https://www.youtube.com/watch?v=furfdqtdAvc#t=54m25s

Re: Deep Neural Networks for YouTube Recommendations

#34

Post by firasd »

rainy-day wrote: post #28

I don't feel like anyone has gotten recommendations right, even though one seemingly obvious approach has not been tried by anyone: allow ratings of favorite works across all media: movies, tv shows, books, music, radio programs, youtube videos. Make a very easy, efficient UI to add ratings. This way you will avoid superficial matches: if I just watched an excellent steampunk cartoon, let's offer a zillion of throwaw…

You're kind of suggesting people go in reverse. Ratings were the initial way these things worked but then they moved to more implicit signals. Netflix used to be all about star ratings back in the day; now they want to measure what you're actually watching.

I think the issue of a system determining whether you like the steampunk genre vs the quality of only that particular steampunk video is separate from the issue of ratings.

Re: Deep Neural Networks for YouTube Recommendations

#35

Post by pcovington »

AndrewKemendo wrote: post #20
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 en…

There are many close collaborations between product and research, as well as direct exchanges between different product areas. Close collaboration is key because those working directly on the product understand best the data, serving system and fundamental constraints.

A recent article [1] revealed how engineers are trained in ML across Google.

[1] https://backchannel.com/how-google-is-remaking-itself-as-a-m...

Re: Deep Neural Networks for YouTube Recommendations

#36

Post by makeee »

seanwilson wrote: post #25
dsco wrote: post #22

One of the very best recommendation engines I've encountered is the "Discover Weekly" playlist from Spotify. It's helped me reconsider my relationship to music which I basically thought was dead since I had hit a rut on exploring new artists. There's an interesting presentation of how it's created on SlideShare http://www.slideshare.net/MrChrisJohnson/from-idea-to-execut...

> One of the very best recommendation engines I've encountered is the "Discover Weekly" playlist from Spotify. The addition of Discover Weekly really confused me. Shouldn't the features that create a radio station from an artist or a playlist fill this need already? Why is it only updated weekly? I haven't tried other services much but it feels like Spotify isn't doing as much as they can with recommendations.

I think it's because people are used to listening to their Playlists. It feels more natural to check your "Discover Weekly" playlist then a whole new section within Spotify. And I think the decision to only update it ever Monday was pretty genius. Most people aren't particularly excited when Monday rolls around.. but when they think about the fact that it's Monday they are likely to remember they have a brand new Discover Weekly playlist to listen to. It's one of the good things about their Monday and becomes a habit over time.

Re: Deep Neural Networks for YouTube Recommendations

#37

Post by hacker42 »

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.

Do you study the phenomenon of information bubbles at Google? Let's say, a German user just happens to watch some right-wing populist video claiming that we need to stop Merkel's refugee politics. The next day the user might receive plenty of recommendations in their feed that confirm the message in the first video. They happen to stumble upon a video of some party convention by an uprising German populist party, and everything makes sense now! Video by video the user gets dragged into a right-wing ideology.

That is an information bubble. The algorithm cannot detect low quality or populism, neither can it recommend opposite standpoints, and at the end of the day it has a real effect on a country's politics and the well-being of many people.

Do you have means of quantifying such effects? What are possible countermeasures?

If you cannot talk about that, then this would be my feedback: Perhaps you could train a language model to find opposing views in video titles and tags and then diversify the video recommendations based on that.

Re: Deep Neural Networks for YouTube Recommendations

#38

Post by eva1984 »

pcovington wrote: post #29
iverjo wrote: post #26

Earlier quoted context omitted.

Would love to see some more details on how you represent videos as feature vectors. Do you only use metadata provided by the uploaders (e.g. title and tags), or do you also analyze the raw video/audio somehow to augment the metadata?

The video embeddings in the paper are learned purely based on observing what users co-watch in sessions. In this sense, they can be thought of as latent factors in more traditional collaborative filtering approaches. When we inspect them, nearby vectors have a surprising amount of semantic similarity. Features about the videos such as titles and tags, as well as features derived from audio and video, are introduced i…

Are you guys using similar model like Word2Vec for obtaining the video embeddings?

Re: Deep Neural Networks for YouTube Recommendations

#39

Post by pastullo »

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.

This is probably offtopic but I unfortunately have to agree with few others here, saying that my recommendations have never been that great.

The system suggests me lots of click-baits and low content quality videos (with massive views though). It's very rare that i get a great video that i eventually really enjoy in my recommandations.

My guess is that the system can't really tell if the video itself is made of good quality, brings good and fresh content. Is that the case? how do you guys rate and measure the intrinsic video quality?

Re: Deep Neural Networks for YouTube Recommendations

#40

Post by pastullo »

firasd wrote: post #34
rainy-day wrote: post #28

I don't feel like anyone has gotten recommendations right, even though one seemingly obvious approach has not been tried by anyone: allow ratings of favorite works across all media: movies, tv shows, books, music, radio programs, youtube videos. Make a very easy, efficient UI to add ratings. This way you will avoid superficial matches: if I just watched an excellent steampunk cartoon, let's offer a zillion of throwaw…

You're kind of suggesting people go in reverse. Ratings were the initial way these things worked but then they moved to more implicit signals. Netflix used to be all about star ratings back in the day; now they want to measure what you're actually watching. I think the issue of a system determining whether you like the steampunk genre vs the quality of only that particular steampunk video is separate from the issue o…

But also view-time or view-count don't tell the full story of how much you liked that video.

I am not happy with YT recommendations because they suggest crap videos to me and not the finest one available for that topic, just as he said.

The system should rather suggest me a different topic but with the best quality/content available, rather than a super similar video with crappier quality/content.

Post reply on HN