This directory contains the code for MixRL: Data Mixing Augmentation for Regression using Reinforcement Learning.
To run MixRL, you need PyTorch, Jupyter Notebook, and CUDA.

The directory contains a total of 4 files and 1 child directories for a dataset (NO2): 
1 README, 2 python files, 1 jupyter notebook, 
and 1 child directories containing 6 numpy files for the NO2 dataset.
a dataset consists of a training set, validation set, and test set.

The jupyter notebook file demonstrates how to run MixRL.

There are two python files: models.py and mixrl.py.
The models.py file contains a Mixup value network architecture, the forward propagation of the network,
and functions for reinforcement learning.
The mixrl.py file contains one class: MixRL.
The MixRL class contains utility functions for computing measures, computing the validation loss,
and functions for the MixRL algorithm including the training of the Mixup value network, finding the best threshold
that minimizes the validation loss, and running Mixup with knn options.
More detailed explanations are in the python files.

Thanks!
