I was reading this HBR article “Business of artificial Intelligence” in which it says
Jeff Wilke, who leads Amazon’s consumer business, says that supervised learning systems have largely replaced the memory-based filtering algorithms that were used to make personalized recommendations to customers.
Does “memory-based filtering algorithms” refer to collaborative filtering?
I stand to be corrected but I believe “memory-based filtering algorithms” refers to something like KNN where you need to keep all the training data around in order to make a prediction (find nearest / most similar users in ratings space then recommend movies that those nearby users rated highly).
In contrast, something like matrix factorization or the model from lesson 4 is ’model-based’ in the sense that you train a model then use the model parameters (the factors) and don’t need the training data any more.
In a sense, memory based algorithms need to remember all the training data while model-based algorithms build a model from the data and then can forget the raw data entirely.
There’s a distinction on “memory based” vs. “model based” CF here: https://en.wikipedia.org/wiki/Collaborative_filtering#Memory-based