chenyaofo / cema

[ICLR 2024] Towards Robust and Efficient Cloud-Edge Elastic Model Adaptation via Selective Entropy Distillation

Towards Robust and Efficient Cloud-Edge Elastic Model Adaptation via Selective Entropy Distillation

This is the official project repository for Towards Robust and Efficient Cloud-Edge Elastic Model Adaptation via Selective Entropy Distillation [ICLR 2024], Authors: Yaofo Chen, Shuaicheng Niu, Yaowei Wang, Shoukai Xu, Hengjie Song, Mingkui Tan

Dependencies Installation:

Please refer to requirements.txt

Data Preparation:

Please download ImageNet-C and extract the tar files. Then use tools/make_tfrec.py to transfer raw image files into tfrecord format.

Example: ImageNet-C Experiments

python -m entry.run --conf conf/cema.conf -o outputs/

Citation

@inproceedings{chen2024towards,
  title={Towards Robust and Efficient Cloud-Edge Elastic Model Adaptation via Selective Entropy Distillation},
  author={Yaofo Chen and Shuaicheng Niu and Shoukai Xu and Hengjie Song and Yaowei Wang and Mingkui Tan},
  booktitle={International Conference on Learning Representations},
  year={2024},
}

cema's Issues

Code For other dataset?

Excellent Work! Can you provide the code for the other dataset used in the paper?

More modification request

After deleting some files in codebase, the code itself has not been modified.