https://github.com/natanaelalmeida/matrix-factorization
Understand how Matrix Factorization and SGD are used in personalized recommendation systems.
https://github.com/natanaelalmeida/matrix-factorization
collaborative-filtering machine-learning mean-square-error mse music-recommendation music-recommendation-system python recommendation-algorithms recommender-system rust sgd sgd-optimizer stochastic-gradient-descent
Last synced: 30 days ago
JSON representation
Understand how Matrix Factorization and SGD are used in personalized recommendation systems.
- Host: GitHub
- URL: https://github.com/natanaelalmeida/matrix-factorization
- Owner: natanaelalmeida
- License: mit
- Created: 2025-02-11T20:27:30.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-02-11T22:16:27.000Z (over 1 year ago)
- Last Synced: 2025-02-11T23:26:56.070Z (over 1 year ago)
- Topics: collaborative-filtering, machine-learning, mean-square-error, mse, music-recommendation, music-recommendation-system, python, recommendation-algorithms, recommender-system, rust, sgd, sgd-optimizer, stochastic-gradient-descent
- Language: Jupyter Notebook
- Homepage: https://physicscomputerlove.com/en/machine-learning/matrix-factorization/
- Size: 28.3 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
**Matrix factorization is a mathematical technique widely used in recommender systems, especially collaborative filtering, to predict users' preferences based on their interactions with items. This approach is crucial for streaming platforms, e-commerce and social networks, as it allows for personalized recommendations.**
The main idea is to "split" an interaction matrix $R$, with dimensions $[m, n]$, into two smaller matrices that represent latent characteristics of users and items:
$$\huge
R \approx U \cdot V^T
$$
1. $m$ is the number of users.
2. $n$ is the number of items.
3. The matrix $U (of \ dimensions \ [m,k])\ $ represents users in a latent feature space.
4. The matrix $V (of \ dimensions \ [n,k])\ $ represents items in a latent feature space.
The value $k$ defines the number of latent factors, which capture implicit patterns in interactions between users and items.
## Example
Imagine a futuristic scenario, at a party illuminated by soft neon lights and nice ambient sound in the background. Robbie, a curious and music-loving robot, finds himself at the center of a rather peculiar chatting circle. Around him, four iconic figures discuss their favorite music styles: Ripley, Darth Vader, Spock and Hermione. Robbie adjusts his hearing sensors...
Read the full article: [Matrix Factorization](https://physicscomputerlove.com/en/machine-learning/matrix-factorization/)