English

Bayesian Flow Networks in Continual Learning

Machine Learning 2023-10-19 v1 Computer Vision and Pattern Recognition Machine Learning

Abstract

Bayesian Flow Networks (BFNs) has been recently proposed as one of the most promising direction to universal generative modelling, having ability to learn any of the data type. Their power comes from the expressiveness of neural networks and Bayesian inference which make them suitable in the context of continual learning. We delve into the mechanics behind BFNs and conduct the experiments to empirically verify the generative capabilities on non-stationary data.

Keywords

Cite

@article{arxiv.2310.12001,
  title  = {Bayesian Flow Networks in Continual Learning},
  author = {Mateusz Pyla and Kamil Deja and Bartłomiej Twardowski and Tomasz Trzciński},
  journal= {arXiv preprint arXiv:2310.12001},
  year   = {2023}
}

Comments

Submitted to NeurIPS 2023 Workshop on Diffusion Models

R2 v1 2026-06-28T12:54:26.783Z