Deep Clustering

This repository stores the models used for experimental computations in the work "Deep Robust Spectral Clustering" (link to abstract, request the full paper through mail).

Contents

• Models

  • Robust Spectral Map
  • SpectralNet (fork coming soon)
  • Deep K-Means (will be implemented)

• Graph Kernels

  • Nearest Neighbors Kernel
  • Stoachastic Nearest Neighbors Kernel

• Experimental Computations

Requirements

numpy==1.26.4
scikit_learn==1.4.1.post1
scipy==1.12.0
torch==2.2.0
tqdm==4.66.2
torchvision==0.17.0

To-do

• Add notebook with experiments
• Add Wiki/Documentation
• Add SpectralNet model (as fork)
• Add Deep K-Means
• Add option for pre-computed kernel (scipy sparse) + fit support
• Add option for dynamic kernel reshaping