Matrix Factorization & SVD (ALS)
Latent Factor Decomposition with Alternating Least Squares & Implicit Feedback
#Recommenders#SVD#ALS#Latent Factors#Netflix Prize
Choose Presentation Mode:
STAGE 1 / 7— Anti-Pattern
Section 1: The LLM Anti-Pattern vs Right-Sized Model
The Naive Generative LLM Approach:
Prompting an LLM with a user’s historical watch history of 500 movies and asking it to rank 100,000 catalog candidates.
Why It Fails in Production:
LLMs cannot perform dot products across 100k candidate vectors in real time, cost thousands in token context, and suffer from recency bias.
Targeted Algorithm (Matrix Factorization & SVD (ALS))
Latency:0.2ms (Dot product lookup)
Cost / 1M Ops:$0.00
Determinism:Optimal Latent Factor Match
Generative LLM Alternative
Latency:4,000ms
Cost / 1M Ops:$18,000
Determinism:Popularity-Biased Hallucinations