English

Efficient Learning of Convolution Weights as Gaussian Mixture Model Posteriors

Computer Vision and Pattern Recognition 2024-07-03 v2

Abstract

In this paper, we showed that the feature map of a convolution layer is equivalent to the unnormalized log posterior of a special kind of Gaussian mixture for image modeling. Then we expanded the model to drive diverse features and proposed a corresponding EM algorithm to learn the model. Learning convolution weights using this approach is efficient, guaranteed to converge, and does not need supervised information. Code is available at: https://github.com/LifanLiang/CALM.

Keywords

Cite

@article{arxiv.2401.17400,
  title  = {Efficient Learning of Convolution Weights as Gaussian Mixture Model Posteriors},
  author = {Lifan Liang},
  journal= {arXiv preprint arXiv:2401.17400},
  year   = {2024}
}
R2 v1 2026-06-28T14:32:25.637Z