Hypernetwork Approach to Bayesian MAML (Student Abstract)

Piotr Borycki, Piotr Kubacki, Marcin Przewiezlikowski, Tomasz Kusmierczyk, Jacek Tabor, Przemyslaw Spurek

AAAI 2025 pp. 29325-29327

doi:10.1609/AAAI.V39I28.35239 /aaai/2025/borycki2025aaai-hypernetwork/

Abstract

The main goal of Few-Shot learning algorithms is to enable learning from small amounts of data. One of the most popular and elegant Few-Shot learning approaches is Model-Agnostic Meta-Learning (MAML). In this paper, we propose a novel framework for Bayesian MAML called BH-MAML, which employs Hypernetworks for weight updates. It learns the universal weights point-wise, but a probabilistic structure is added when adapted for specific tasks. In such a framework, we can use simple Gaussian distributions or more complicated posteriors induced by Continuous Normalizing Flows.

PDF AAAI Semantic Scholar

Cite

Text

Borycki et al. "Hypernetwork Approach to Bayesian MAML (Student Abstract)." AAAI Conference on Artificial Intelligence, 2025. doi:10.1609/AAAI.V39I28.35239

Markdown

[Borycki et al. "Hypernetwork Approach to Bayesian MAML (Student Abstract)." AAAI Conference on Artificial Intelligence, 2025.](https://mlanthology.org/aaai/2025/borycki2025aaai-hypernetwork/) doi:10.1609/AAAI.V39I28.35239

BibTeX

@inproceedings{borycki2025aaai-hypernetwork,
  title     = {{Hypernetwork Approach to Bayesian MAML (Student Abstract)}},
  author    = {Borycki, Piotr and Kubacki, Piotr and Przewiezlikowski, Marcin and Kusmierczyk, Tomasz and Tabor, Jacek and Spurek, Przemyslaw},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2025},
  pages     = {29325-29327},
  doi       = {10.1609/AAAI.V39I28.35239},
  url       = {https://mlanthology.org/aaai/2025/borycki2025aaai-hypernetwork/}
}