In reading the YouTube Paper, I came across one technical section that seemed very subtle and was not readily apparent why it was true: the watch time estimation using weighted logistic regression. It’s easy to gloss over this detail, but as it turns out, many people before me were curious about this section as well. The previous links have already explored one view into the why behind this formula, but I would like to formalize the process and explain it end to end.

In this post, we explore two approaches for training a linear regression online and compare their behavior on non-stationary data. We begin by deriving online linear regression and the Kalman filter update from first principles. Then, we empirically show differences between the two methods by fitting models using each algorithm on non-stationary data and looking at the residuals.

In our previous posts, we walked through the sampled softmax and in-batch negatives as loss functions for training a retrieval model. In this post, we will explore the best of both worlds in the so-called mixed negative sampling. This method will prove the to be the most data efficient and yield the best performance of the previous two methods. The post contains a new implementation of all three methods which will allow all three to be trained from (approximately) the same data loader and compared.

Consider the retrieval problem in a recommendation system. One way to model the problem is to create a big classification model predicting the next item a user will click or watch. This basic approach could be extended to large catalogs via a sampled softmax loss, as discussed in our last post. Naturally, the resulting dataset will only contain positive examples (i.e., what was clicked). In this post, we explore a more optimized version of this approach for more complex models via in-batch negatives. We derive the LogQ Correction to improve the estimate of the gradient and present code in PyTorch to implement the method, building on the previous blog post.

Sometimes I’m asked how to start learning about recommender systems, and there’s a few go-to papers I always mention; however, without a proper map, it could be alittle difficult for the uninitiated. So, to try to make a gentle introduction, I will walk through an implementation of the candidate generation model based on Deep Neural Networks for YouTube, which I will sometimes refer to as the “Deep Nets” paper. This paper (and its talk) is jam-packed with all sorts of interesting practical recommendation system knowledge and is the perfect starting place for people hoping to understand large-scale recommendation systems. I will implement the key components of the candidate generation model and train the model on the MovieLens dataset in PyTorch, a typical benchmark dataset in the RecSys space. There will also be a few modern flourishes and comments on new developments as applicable. I’ll conclude with outlining a natural extension to this approach to predict multiple outcomes.