Skip to content

glarkstoat/SVM-using-SGD

Repository files navigation

MMD-Assignment #2: SVM-using-SGD

Team: Emin Guliev, Justus Rass & Christian Wiskott.

Requirements

  • Python 3
  • Numpy
  • Matplotlib
  • Jupyter
  • Seaborn
  • Tqdm

The DataLoader module assumes that the folder "datasets" is in the root folder. If that's not the case, pass your file-path to the DataLoader keyword-argument >path<.

Description

Application and comparison of a LinearSVM model using regular vs. Random Fourier Features on two toydata sets as well as the MNIST dataset. Includes a parallel implementation of the SVM + comparison of runtimes.

The included jupyther-notebooks provide all results used in the report and can be executed using a jupyther environment.

About

Mining Massive Data Project - University of Vienna

Resources

Stars

Watchers

Forks

Contributors