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

Fast Abductive Learning by Similarity-based Consistency Optimization

This is the repository for holding the sample code of Fast Abductive Learning by Similarity-based Consistency Optimization in NeurIPS 2021.

This code is only tested in Linux environment.

Environment Dependency

• Ubuntu 18.04
• Python 3.7
• PyTorch 1.7
• CuPy 8.3
• tqdm
• scikit-learn
• opencv-python

To create the above environment with Anaconda, you can run the following command (cudatoolkit=10.1 for old GPUs, cudatoolkit=11.3 for new GPUs / new drivers):

(cudatoolkit=10.1)

conda create -n ablsim python=3.7 -y
conda activate ablsim
conda install -c conda-forge cupy cudatoolkit=10.1 -y
conda install pytorch==1.7.0 torchvision==0.8.0 torchaudio==0.7.0 cudatoolkit=10.1 -c pytorch -y
pip install tqdm opencv-python scikit-learn matplotlib

(cudatoolkit=11.3)

conda create -n ablsim python=3.7 -y
conda activate ablsim
conda install -c conda-forge cupy cudatoolkit=11.3 -y
conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch -y
pip install tqdm opencv-python scikit-learn matplotlib

Running Code

To reproduce the experiment results, simply run the following code:

Download the Handwritten_Math_Symbols.zip from google drive and unzip it:

unzip Handwritten_Math_Symbols.zip -d data

• MNIST (CIFAR-10) Addition

  python main_1_2.py --dataset 2ADD --images handwritten 
  python main_1_2.py --dataset 2ADD --images CIFAR 
  

• Handwritten Formula Recognition

  python main_1_2.py --dataset HWF --images handwritten
  python main_1_2.py --dataset HWF --images CIFAR 
  

• CIFAR-10 Decimal Equation Decipherment

  Download the images.zip and ssl_mode.zip from google drive and unzip it:

  unzip images.zip -d data
  unzip ssl_model.zip
  python main_3.py --images CIFAR
  

To view or change the hyperparameters, please refer to the arg_init() function in the code.

Reference

@incollection{ablsim2021huang,
	author = {Huang, Yu-Xuan and Dai, Wang-Zhou and Cai, Le-Wen and Muggleton, Stephen H and Jiang, Yuan},
	booktitle = {Advances in Neural Information Processing Systems 34},
	pages = {26574--26584},
	title = {Fast Abductive Learning by Similarity-based Consistency Optimization},
	year = {2021}
}