Recommender Systems – A Walkthrough
summary
A practical roadmap: recommender pipelines use only user–item interactions and the best-performing method is matrix factorization learned by stochastic gradient descent; demonstration trains on roughly 1 million datapoints (about 6k users × 4k items) with a compact implementation (a few dozen lines) and runs single‑core in seconds, improving baseline RMSE from ~1.12 down to 0.92. Also mentions production options (MyMediaLite .NET, parallel/Hadoop variants) and basic hyperparameter tuning.
AI-written summary · about
chapters (8)
- 0:00Recommender systems intro
- 3:09What RS solve and origins
- 9:56Input data evolution
- 18:25Live RS demo
- 21:57RS algorithm classes
- 37:40Theory to code walkthrough
- 52:06Performance comparison
- 53:51Q&A on optimization
related talks
41mRecommender Systems: Crafting Effective Personalised Experiences
01h 31mHands on - Build a Recommender system
37mA primer on recommendation systems.
32mLet's talk about Recommendation Systems!
01h 38mBuilding Recommender Systems Using Python
20mBuilding Recommender system
01h 21mUnlocking Personalization: A Deep Dive into Modern Recommendation Algorithms
42mDeep Learning Recommender Architecture: Master Thesis to ...