Answer
In a recent paper published by Google, YouTube engineers analyzed in greater detail the inner workings of YouTube’s recommendation algorithm. The paper was presented at the 10th ACM Conference on Recommender Systems last week in Boston.
YouTube recommendations are driven by Google Brain, which was recently open-sourced as TensorFlow. By using TensorFlow one can experiment with different deep neural network architectures using distributed training. The system consists of two neural networks. The first one, candidate generation, takes as input user’s watch history and using collaborative filtering selects videos in the range of hundreds. An important distinction between development and final deployment to production is that during development Google uses offline metrics for the performance of algorithms but the final decision comes from living A/B testing between the best performing algorithms.
Candidate generation uses the implicit feedback of video watched by users to train the model. Explicit feedback such as a thumbs up or a thumbs down to a video are in general rare compared to implicit and this is an even bigger issue with long-tail videos that are not popular.
To accelerate training of the model for newly uploaded videos, the age of each training example is fed in as a feature. Another key aspect for discovering and surfacing new content is to use all YouTube videos watched, even on partner sites, for the training of the algorithm.
This way collaborative filtering can pick up viral videos right away. Finally, by adding more features and depth like searches and age of video other than the actual watches, YouTube was able to improve offline holdout precision results.
The second neural network is used for Ranking the few hundreds of videos in order.
This is much simpler as a problem to candidate generation as the number of videos is smaller and more information is available for each video and its relationship with the user. This system uses logistic regression to score each video and then A/B testing is continuously used for further improvement. The metric used here is expected watch time, as expected click can promote clickbait.
To train it on watch time rather than click through rate, the system uses a weighted variation of logistic regression with watch time as the weight for positive interactions and a unit weight for negative ones. This works out partly because the fraction of positive impressions is small compared to the total.
YouTube’s recommendation system is one of the most sophisticated and heavily used recommendation systems in the industry. The paper just scratches the surface but nonetheless gives several useful insights regarding engineering deep learning systems.
Source:https://www.infoq.com/news/2016/09/How-YouTube-Recommendation-Works
In a recent paper published by Google, YouTube engineers analyzed in greater detail the inner workings of YouTube’s recommendation algorithm. The paper was presented at the 10th ACM Conference on Recommender Systems last week in Boston.
YouTube recommendations are driven by Google Brain, which was recently open-sourced as TensorFlow. By using TensorFlow one can experiment with different deep neural network architectures using distributed training. The system consists of two neural networks. The first one, candidate generation, takes as input user’s watch history and using collaborative filtering selects videos in the range of hundreds. An important distinction between development and final deployment to production is that during development Google uses offline metrics for the performance of algorithms but the final decision comes from living A/B testing between the best performing algorithms.
Candidate generation uses the implicit feedback of video watched by users to train the model. Explicit feedback such as a thumbs up or a thumbs down to a video are in general rare compared to implicit and this is an even bigger issue with long-tail videos that are not popular.
To accelerate training of the model for newly uploaded videos, the age of each training example is fed in as a feature. Another key aspect for discovering and surfacing new content is to use all YouTube videos watched, even on partner sites, for the training of the algorithm.
This way collaborative filtering can pick up viral videos right away. Finally, by adding more features and depth like searches and age of video other than the actual watches, YouTube was able to improve offline holdout precision results.
The second neural network is used for Ranking the few hundreds of videos in order.
This is much simpler as a problem to candidate generation as the number of videos is smaller and more information is available for each video and its relationship with the user. This system uses logistic regression to score each video and then A/B testing is continuously used for further improvement. The metric used here is expected watch time, as expected click can promote clickbait.
To train it on watch time rather than click through rate, the system uses a weighted variation of logistic regression with watch time as the weight for positive interactions and a unit weight for negative ones. This works out partly because the fraction of positive impressions is small compared to the total.
YouTube’s recommendation system is one of the most sophisticated and heavily used recommendation systems in the industry. The paper just scratches the surface but nonetheless gives several useful insights regarding engineering deep learning systems.
Source:https://www.infoq.com/news/2016/09/How-YouTube-Recommendation-Works
