Soft-Decision-Tree is the pytorch implementation of Distilling a Neural Network Into a Soft Decision Tree, paper recently published on Arxiv about adopting decision tree algorithm into neural network. "If we could take the knowledge acquired by the neural net and express the same knowledge in a model that relies on hierarchical decisions instead, explaining a particular decision would be much easier."
- python
- pytorch
I achieved 92.95% of test dataset accuracy on MNISTafter 40 epoches, without exploring enough of hyper-parameters (The paper achieved 94.45%). Higher accuracy might be achievable with searching hyper-parameters, or training longer epoches (if you can, please let me know :) )
$ python main.py