Publications Journal · 2021

EAML: Ensemble Self-Attention-Based Mutual Learning Network for Document Image Classification

S. Bakkali, Z. Ming, M. Coustaty, M. Rusiñol

International Journal on Document Analysis and Recognition (IJDAR) 24(3), pp. 251–268

In one paragraph

A self-attention fusion module inside an ensemble network learns image and text features together, and mutual learning transfers positive knowledge between the two modalities through a truncated Kullback-Leibler regulariser.

Cite

@article{bakkali2021eaml,
  title   = {{EAML: Ensemble Self-Attention-Based Mutual Learning Network for Document Image Classification}},
  author  = {Bakkali, Souhail and Ming, Zuheng and Coustaty, Mickaël and Rusiñol, Marçal},
  journal = {International Journal on Document Analysis and Recognition (IJDAR)},
  volume  = {24},
  number  = {3},
  pages   = {251--268},
  year    = {2021},
  doi     = {10.1007/s10032-021-00378-0},
  url     = {https://arxiv.org/abs/2305.06923}
}