tree: b6e3c0f6bd578cb196e1d5e1a27b16ee8ea388e2 [path history] [tgz]
  1. .gitignore
  2. README.md
  3. crossentropy.py
  4. demo-MF.R
  5. demo1-MF.ipynb
  6. demo1-MF2-fancy.ipynb
  7. demo2-binary.ipynb
  8. demo3-dssm.ipynb
  9. matrix_fact.py
  10. movielens_data.py
  11. negativesample.py
  12. randomproj.py
  13. recotools.py
  14. symbol_alexnet.py
example/recommenders/README.md

Recommender Systems with Sparse Data

This directory has a set of examples of how to build various kinds of recommender systems using MXNet. It also includes a set of tools for using sparse data.

Examples

The examples are driven by notebook files.

Re-usable code

These examples use and demonstrate a number of layers and other tools that can be used outside of these examples. They are all available from the recotools package.

Negative Sampling

  • NegativeSamplingDataIter

Loss Layers

  • CosineLoss
  • CrossEntropyLoss

Sparse Data Projection layers

  • SparseRandomProjection
  • SparseBagOfWordProjection

Acknowledgements

Thanks to xlvector for the first Matrix Factorization example that provided the basis for these examples.

MovieLens data from GroupLens.