New algorithms for large scale learning with kernels

RKHS project for the course "Generalization Properties" by Aymeric Dieuleveut

Authors: Corentin Ambroise, Luis Montero and Margaux Zaffran

Data

• CTSlices can be found here.

References

• "Random Features for Large-Scale Kernel Machines", A. Rahimi, B. Recht
• "Generalization properties of learning with random features", A. Rudi, L. Rosasco
• "Less is More: Nystrom Computational Regularization", R. Camoriano Rosasco

Set up

This repository requires python, but also needs some libraries:

• pandas
• numpy
• scikit-learn
• matplotlib
• jupyter
• googledrivedownloader
• os

If you need to install some of them, you can simply execute pip install -r requirements.txt on your terminal.

All implemented algorithms can be found in rhks module.

To reproduce the results you can run Large_Scale_Learning.ipynb