Skip to content
forked from lvyilin/RecogIB

Official codebase for the IJCAI 2023 paper, Recognizable Information Bottleneck

License

Notifications You must be signed in to change notification settings

CLIAgroup/RecogIB

 
 

Repository files navigation

Recognizable Information Bottleneck

This is the official repo of Recognizable Information Bottleneck (RIB).

Requirement

The code is implemented using PyTorch framework. Use pip or conda to install all dependencies:

pip install -r requirements.txt

or

conda install --file requirements.txt

Usage

Run the following command to get all the experimental results in the paper:

./run.sh `which python` <your log dir> <your data dir>

It will automatically invoke all the GPUs on the machine to run the experiments. It took about 16 hours to run through on our machine (8 × NVIDIA RTX A4000).

Citation

@inproceedings{ijcai2023p0448,
  title     = {Recognizable Information Bottleneck},
  author    = {Lyu, Yilin and Liu, Xin and Song, Mingyang and Wang, Xinyue and Peng, Yaxin and Zeng, Tieyong and Jing, Liping},
  booktitle = {Proceedings of the Thirty-Second International Joint Conference on
               Artificial Intelligence, {IJCAI-23}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Edith Elkind},
  pages     = {4028--4036},
  year      = {2023},
  month     = {8},
  note      = {Main Track},
  doi       = {10.24963/ijcai.2023/448},
  url       = {https://doi.org/10.24963/ijcai.2023/448},
}

License

MIT

About

Official codebase for the IJCAI 2023 paper, Recognizable Information Bottleneck

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 91.4%
  • Python 8.4%
  • Shell 0.2%