PyTorch implementation of neural additive models in Neural Additive Models (PyTorch) is adopted for this implementation of SurvNAM.
For neural additive models, check out:
- Neural Additive Models: Interpretable Machine Learning with Neural Nets.
- TensorFlow OG Implementation
For random survival forests (RSF):
scikit-learn>=1.0.2
numpy>=1.21.5
pandas>=1.3.5
tqdm>=4.54.0
setuptools>=61.2.0
In Python:
from nam import NeuralAdditiveModel
model = NeuralAdditiveModel(input_size=x_train.shape[-1],
shallow_units=100,
hidden_units=(64, 32, 32),
shallow_layer=ExULayer,
hidden_layer=ReLULayer,
hidden_dropout=0.1,
feature_dropout=0.1)
logits, feature_nn_outputs = model.forward(x)If you use this code in your research, please cite the following paper:
Utkin, L. V., Satyukov, E. D., & Konstantinov, A. V. (2022). SurvNAM: The machine learning survival model explanation. Neural Networks, 147, 81-102.
@article{utkin2022survnam,
title={SurvNAM: The machine learning survival model explanation},
author={Utkin, Lev V and Satyukov, Egor D and Konstantinov, Andrei V},
journal={Neural Networks},
volume={147},
pages={81--102},
year={2022},
publisher={Elsevier}
}
Agarwal, R., Frosst, N., Zhang, X., Caruana, R., & Hinton, G. E. (2020). Neural additive models: Interpretable machine learning with neural nets. arXiv preprint arXiv:2004.13912
@article{agarwal2020neural,
title={Neural additive models: Interpretable machine learning with neural nets},
author={Agarwal, Rishabh and Frosst, Nicholas and Zhang, Xuezhou and
Caruana, Rich and Hinton, Geoffrey E},
journal={arXiv preprint arXiv:2004.13912},
year={2020}
}