On minimizing the training set fill distance in machine learning regression

This repository contains the code and experiments to replicate the results presented in the paper titled "On minimizing the training set fill distance in machine learning regression"

Paolo Climaco and Jochen Garcke
Institut für Numerische Simulation, Universität Bonn, Germany
Fraunhofer SCAI, Sankt Augustin, Germany

Contact [email protected] for questions about code and data.

Python Packages

-Python (>= 3.7)
-Pytorch 1.11.0
-Install packages in requirements.txt

Repository Structure

.
├── datasets/                   # Folder containing code to access data. Used also for data storage. 
│   ├── Datasets_Class.py           # Code for dowloading and reading datasets.
│    
├── notebooks/                  # Folder containing jupiter notebooks.
│   ├── experiments_QM7,ipynb       # Jupyter Notebook replicating experiments on QM7.
│   ├── access_data.ipynb           # Jupyter Notebook explaining how to access preprocessed datasets.
│
├── Passive_Sampling/           # Folder containing code to select datapoint with FPS.
│   ├──farthest_point.py            # Code for implementing the  Fartehst Point Sampling (FPS).
│
├── utils/                      # Folder containing basic code to run and plot experiments.
│   ├──FNN.py                       # Code containg the FNN architecture, training and testing procedures.
│   ├──plots.py                     # Code plotting the result of the experiments.

└── README.md                   # Project README file.
└── requirements.txt            # python packages required to run code.