smartyfh / wish

Few-Bits Semantic Hashing

WISH

Few-Bits Semantic Hashing

Code in Python for the paper "Unsupervised Few-Bits Semantic Hashing with Implicit Topics Modeling", Findings of EMNLP 2020. [paper]

Model

Train

1. Data Preprocessing

   python3 datapreprocessing.py
   

2. Training

   If the datasets have multiple labels for a single sample, run

   python3 train_WISH.py
   

   Otherwise, run

   python3 train_WISH.py --single_label_flag
   

Citation

@inproceedings{ye2020unsupervised,
  title={Unsupervised Few-Bits Semantic Hashing with Implicit Topics Modeling},
  author={Ye, Fanghua and Manotumruksa, Jarana and Yilmaz, Emine},
  booktitle={Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: Findings},
  pages={2566--2575},
  year={2020}
}

wish's Issues

Why use the real value \mu to calculate KL loss rather than z?

Thank you for your publication. I have a question about this line:

WISH/train_WISH.py

Line 65 in a0a7c63

kl_loss = WISH.calculate_KL_loss(mu)

In section 3.3.3 of the paper, I find the binary value z is used to calculate the Kl loss. While in this code, the real value \mu is used instead. Is there something wrong with my understanding?