A personal collection of code for training all kinds of recommender systems that I may want to use again. Some code is from courses, some hand written.
Contains:
- Content Filtering
- Factorization Machines
- Feature Engineering (various methods for creating feature embedding vectors)
- Reinforcement Learning (Multi-Armed Bandits)
Coming soon:
-
Factorization Machines in pytorch
-
Deep Factorization Machines (a Deep Hybrid Recommender system similar to the FM) in pytorch
-
Feature Engineering - VAE in pytorch
-
Deep Hybrid Recommender system using Vampire in pytorch
-
User-based and Item-based Collaborative Filtering
- Nearest neighbors approach
- Matrix decomposition by gradient descent
- Nonnegative matrix factroization
- Probabalistic Matrix Factorization
- Deep Collaborative Filtering with autoencoders
-
Association rules
-
Deep outfits. ("pairs well with")
-
"Describe what you are looking" for bot - Information retrieval (cosine sim with document embeddings. take in any type of document embeddings) (This is the same as content, but it uses sentiment)
-
LDA recommender system
