Publications of Torsten Hoefler

Nikoli Dryden, Torsten Hoefler:

Spatial Mixture-of-Experts

(In Advances in Neural Information Processing Systems 35, presented in New Orleans, Louisiana, Dec. 2022)

Abstract

Many data have an underlying dependence on spatial location; it may be weather on the Earth, a simulation on a mesh, or a registered image. Yet this feature is rarely taken advantage of, and violates common assumptions made by many neural network layers, such as translation equivariance. Further, many works that do incorporate locality fail to capture fine-grained structure. To address this, we introduce the Spatial Mixture-of-Experts (SMoE) layer, a sparsely-gated layer that learns spatial structure in the input domain and routes experts at a fine-grained level to utilize it. We also develop new techniques to train SMoEs, including a self-supervised routing loss and damping expert errors. Finally, we show strong results for SMoEs on numerous tasks, and set new state-of-the-art results for medium-range weather prediction and post-processing ensemble weather forecasts.

Documents

download article:

download slides:

Recorded talk (best effort)

BibTeX

@inproceedings{smoe,
  author={Nikoli Dryden and Torsten Hoefler},
  title={{Spatial Mixture-of-Experts}},
  year={2022},
  month={Dec.},
  booktitle={Advances in Neural Information Processing Systems 35},
  location={New Orleans, Louisiana},
  source={http://www.unixer.de/~htor/publications/},
}


serving: 18.118.208.127:13849	© Torsten Hoefler