Summary
In this chapter, we have discussed how to develop scalable recommendation systems with TensorFlow. We have seen some of the theoretical backgrounds of recommendation systems and using a collaborative filtering approach in developing recommendation systems. Later in the chapter, we saw how to use SVD, and K-means, to develop a movie recommendation system.
Finally, we saw how to use FMs and a variation called NFM to develop more accurate recommendation systems that can handle large-scale sparse matrixes. We have seen that the best way to handle the cold-start problem is to use a collaborative filtering approach with FMs.
The next chapter is about designing an ML system driven by criticisms and rewards. We will see how to apply RL algorithms to make a predictive model for real-life datasets.
