🍀 PyTorch Implementation of Few-shot Partial Multi-view Learning 🍀

♨ An overview of the proposed few-shot partial multi-view learning task:

♨ A schematic illustration for the unified Gaussian dense-anchoring approach:

♨ Preparation

(1) Please download the data and put it into the root path of the project ./UGDA/;
(2) Please run the commend: pip install -r requirments.txt;

♨ Script Running

please use the following commend to run the scripts:
bash runcmd/{data name}/run.sh

The results will be around to the followings:

dataset	0	0.1	0.2	0.3	0.4	0.5
Cub-googlenet-doc2vec	95.59	90.76	87.72	83.28	80.16	76.69
Handwritten	89.08	85.44	80.51	76.13	71.27	66.77
Caltech102	59.13	54.15	51.35	47.81	44.29	42.05
Scene15	72.37	70.15	67.11	65.83	63.26	62.17
Animal	89.86	84.34	77.96	72.59	66.71	62.25
ORL	95.79	92.64	86.59	81.41	72.44	64.78

📌 To improve readability, we have comprehensively polished the code before releasing it, including comprehensive cleaning and re-organization, which may result in slight differences from the original one. Please be free to leave your questions in the issue panel. The paper is available at here: https://ieeexplore.ieee.org/abstract/document/10123043/. If our work is helpful for your research, please consider to cite our paper or give our project a start.

@ARTICLE{10123043,
  author={Zhou, Yuan and Guo, Yanrong and Hao, Shijie and Hong, Richang and Luo, Jiebo},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
  title={Few-Shot Partial Multi-View Learning}, 
  year={2023},
  volume={45},
  number={10},
  pages={11824-11841},
  doi={10.1109/TPAMI.2023.3275162}}
}

Acknowledgment

Thanks for the great works ALICE, C-FSCIL, and NC-FSCIL.