🌟 New! ABLkit released: A toolkit for Abductive Learning with high flexibility, user-friendly interface, and optimized performance. Welcome to try it out!🚀

Deciphering Raw Data in Neuro-Symbolic Learning with Provable Guarantees

This is the repository for holding the sample code of Deciphering Raw Data in Neuro-Symbolic Learning with Provable Guarantees in AAAI 2024.

Getting Started

Our code relies on PyTorch, which will be automatically installed when you follow the instructions below.

conda create -n abl-tl python=3.8
conda activate abl-tl
pip install -r requirements.txt

Running Experiments

• ABL-TL on ConjEq.

  python main.py --train_loss TL --kb ConjEq
  

• TL-Risk on Conjunction.

  python main.py --train_loss TL --kb Conjunction
  

Citing this work

@inproceedings{tao2024deciphering,
  title={Deciphering Raw Data in Neuro-Symbolic Learning with Provable Guarantees},
  author={Tao, Lue and Huang, Yu-Xuan and Dai, Wang-Zhou and Jiang, Yuan},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  year={2024}
}