ulmmcm

Unsupervised Learning for Multi-Model Consensus Maximization. Based on the single model scenario developed by Thomas Probst et al.

setup

1. download 'data' from https://drive.google.com/drive/folders/1i_20qdSdSLnVVkvEaG6lJWl92c6HGNYQ?usp=sharing
2. place 'data' inside the repository in such a way the directory tree appears as follows:

-- ulmmcm  
      |  
      | -- bin  
      | -- data  
      | -- model
      | -- results
      | -- syndalib             
      | -- utils                
      | -- losses.py             
      | -- metrics.py  
      | -- requirements.txt  

3. setup venv using requirements.txt file:
   pip install -r requirements.txt

train

1. modify values under "user defined variables" in the script bin/train.py as you wish
2. run train.py

test

1. run test.py script
2. choose what scenario to test among those listed in the menu
   (results are already stored in results/test)

contacts

contact me at william.bonvini@outlook.com for any question or doubt!