SupervisedMF 0.0.4

A package for various supervised matrix factorization methods

Supervised Matrix Factorization

This Python package contains source codes for algorithms for Supervised Matrix Factorization (SMF) in the papers [1] and [2]:

Installation

To install the package, run the following command in your environment:

python3 -m pip install SupervisedMF

Check your installation by trying to import the main classes in this package:

>>> from SMF import SMF_BCD
>>> from SMF import SMF_LPGD

Pytorch Version

If you are looking to use the Pytorch version of the Supervised Matrix Factorization algorithms, please first install torch and its related dependencies in your environment using the appropriate command from the official installation page.

For example, if you want to install torch for Linux with CUDA 12.1 using pip, run the following command:

pip3 install torch torchvision torchaudio

References

[1] Lee, Joowon, Hanbaek Lyu, and Weixin Yao. "Exponentially convergent algorithms for supervised matrix factorization." Advances in Neural Information Processing Systems 36 (2024).

[2] Lee, Joowon, Hanbaek Lyu, and Weixin Yao. "Supervised Matrix Factorization: Local Landscape Analysis and Applications." Forty-first International Conference on Machine Learning (2024).