Von Mises-Fisher Sampling of GloVe Vectors

Abstract

A recent publication introduced von Mises-Fisher exploration (vMF-exp), a scalable sampling method for exploring large action sets in reinforcement learning problems where hyperspherical embedding vectors represent these actions. We present the first experimental validation of vMF-exp’s key theoretical and scalability properties on a publicly available real-world dataset, confirming the potential of this method.

Cite

Text

Bendada et al. "Von Mises-Fisher Sampling of GloVe Vectors." ICLR 2025 Workshops: FPI, 2025.

Markdown

[Bendada et al. "Von Mises-Fisher Sampling of GloVe Vectors." ICLR 2025 Workshops: FPI, 2025.](https://mlanthology.org/iclrw/2025/bendada2025iclrw-von/)

BibTeX

@inproceedings{bendada2025iclrw-von,
  title     = {{Von Mises-Fisher Sampling of GloVe Vectors}},
  author    = {Bendada, Walid and Salha-Galvan, Guillaume and Hennequin, Romain and Bontempelli, Théo and Bouabça, Thomas and Cazenave, Tristan},
  booktitle = {ICLR 2025 Workshops: FPI},
  year      = {2025},
  url       = {https://mlanthology.org/iclrw/2025/bendada2025iclrw-von/}
}