Collaborative filtering — how does it use unsupervised methods?
mediumAnswer
- Learn user + item embeddings from interaction matrix (implicit or explicit ratings).
- No content features required — 'unsupervised' in the sense that only interactions are used.
- Methods: (1) SVD / matrix factorization (Netflix Prize), (2) ALS (Spark), (3) Neural CF, (4) Two-tower / dual-encoder models, (5) Graph-based (LightGCN).
- Cold-start problem (new user / item) needs content features → hybrid.
Check yourself — multiple choice
- Same as classification
- Learn user + item embeddings from interaction matrix (implicit / ratings); MF / ALS / Neural CF / two-tower / LightGCN; cold-start needs content features → hybrid
- Random
- Not real
CF: user + item embeddings from interactions; MF/ALS/dual-tower/graph.
#applications
Practise Unsupervised Learning
214 interview questions in this topic.