Neural Characteristic Activation Analysis and Geometric Parameterization for ReLU Networks (NeurIPS 2024)

This is the Official TensorFlow 2 Implementation of Geometric Parameterization, proposed in the paper Neural Characteristic Activation Analysis and Geometric Parameterization for ReLU Networks (published at NeurIPS 2024). Please read our paper for detailed descriptions of the proposed geometric parameterization.

All scripts and notebooks have been tested with python==3.8.12 and tensorflow==2.8.0.

If you find our paper and/or code useful for your research, please consider citing our paper:

@article{chen2024neural,
  title={Neural Characteristic Activation Analysis and Geometric Parameterization for ReLU Networks},
  author={Wenlin Chen and Hong Ge},
  journal={Advances in Neural Information Processing Systems},
  volume={37},
  year={2024}
}