/MDL

🔥[IEEE TPAMI 2023] Official repository TPAMI 2023 paper "Exploiting Field Dependencies for Learning on Categorical Data"

Primary LanguagePythonOtherNOASSERTION

MDL

This repository is the official implementation of the code for paper Exploiting Field Dependencies for Learning on Categorical Data.

Requirements

The code has been tested with:

  • Python 3.6.12
  • PyTorch 1.10.1
  • lmdb 1.1.1
  • tqdm 4.57.0

Training and Evaluation (e.g. on Avazu dataset)

python run_mdl.py   --dataset-path <path_to_data> --inner-step 4 --inner-step-size 1 --lr 0.05 --wdcy 1e-8 --lmbd 100 --ebd-dim 40

Citation

If you find this repository helpful, please consider to cite the following paper:

@ARTICLE{10192074,
  author={Li, Zhibin and Koniusz, Piotr and Zhang, Lu and Pagendam, Daniel Edward and Moghadam, Peyman},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
  title={Exploiting Field Dependencies for Learning on Categorical Data}, 
  year={2023},
  volume={},
  number={},
  pages={1-14},
  doi={10.1109/TPAMI.2023.3298028}}