The Predictive Mixture Learner

Get started:

git clone https://github.com/kobybibas/mixture_learner_experiments.git --recursive
cd mixture_learner_experiments
pip install -r requirements.txt

Download the experimnets

This repository relies on https://github.com/kobybibas/deep_pnml_experiments.

Download raw results of deep pNML: https://drive.google.com/open?id=1sMCZo2aoei7UxahQONAOf8gLp5Hjb1WQ

After the extraction the main directory is as follows, notice the new output directory:

.
├── LICENSE
├── README.md
├── deep_pnml_raw_results.zip
├── externals
├── notebooks
├── output
└── src

Analyze the results using jupyter notebooks:

notebooks
├── mixture_learner_linear_regression.ipynb
├── mixture_lenet_cifar10.ipynb
├── mixture_out_of_distribution.ipynb
├── mixture_random_labels.ipynb
├── mixture_resnet18_cifar10.ipynb
└── result_summary.ipynb