laixinn / CIVis

Towards Better Modeling With Missing Data: A Contrastive Learning-Based Visual Analytics Perspective

Video demo

CIVis integrates a Contrastive Learing (CL) framework to enable modeling dataset with missing values, avoiding imputation, and is a visual analytics system to allow users with limited background CL knowledge to iteratively improve model training. Finally a accurate and trustworthy model makes prediction for downstream tasks, fed by incomplete data.

How to run the system

1. Install python packages (Exclude PyTorch if you already install.)

   cd backend
   pip install -r requirement.txt

2. Install frontend packages

   cd frontend
   npm install

3. Run backend

   cd backend
   /opt/conda/bin/flask run --host=0.0.0.0 --port=5000

4. Run frontend

   cd frontend
   npm run start

How to cite

If this paper and tool helps your research projects, please considering citing our paper:

@article{xie2023towards,
  title={Towards Better Modeling With Missing Data: A Contrastive Learning-Based Visual Analytics Perspective},
  author={Xie, Laixin and Ouyang, Yang and Chen, Longfei and Wu, Ziming and Li, Quan},
  journal={IEEE Transactions on Visualization and Computer Graphics},
  year={2023},
  publisher={IEEE}
}