/EDL

An implementation using PyTorch for the paper "Evidential Deep Learning to Quantify Classification Uncertainty".

Primary LanguagePython

Classification Uncertainty

🧐 About

An easy-to-run implementation using PyTorch for the paper " Evidential Deep Learning to Quantify Classification Uncertainty ".

🎈 Usage

Requirements

  • python 3.8.8
  • numpy 1.19.2
  • pytorch 1.7.1
  • torchvision 0.8.2
  • scikit-learn 0.24.1
  • scikit-image 0.18.2
  • scipy 1.6.2

Running

python main.py

📝 Experiments

Acc on valid dataset.

Model MNIST
softmax 0.9909
EDL using log 0.9714
EDL using digamma 0.9766
EDL using mse 0.9745

💝 Acknowledgement