Index

Demistifying E(3)-equivariant neural networks!

Check out the website: https://killiansheriff.github.io/DemistifyingE3NN/index.html

Welcome to our blog post on demistifying E(3)-equivariant neural networks! This blog aim to introduces the following concepts, in order to understand the mathematical tools leading to learnable equivariant operations:

1. Group_representation
2. Irreducible Representations
3. Transformation under Euclidean Symmetries
4. Spherical Harmonics
5. Principle of Equivariance and Polynomials
6. Learnable Tensor Products
7. MLIAP and Data Efficiency

Disclaimer

This post serves as an explanation of the paper by Geiger et al [1]. It was written as a final project for the Fall 2022 Final Project section of the 6.S898: Deep Learning course at the Massachusetts Institute of Technology. Needless to say that the intuition presented herein are heavily influenced by the explanations of Geiger et al.

References

1. Geiger, M. & Smidt, T. e3nn: Euclidean Neural Networks. Arxiv (2022).

JupyterBook

jupyter-book create blog/
jupyter-book build blog/

Citation

If used, please cite:

@software{killian_sheriff_2022_7430281,
  author       = {Killian Sheriff and
                  Yifan Cao},
  title        = {killiansheriff/blog\_e3nn: blog\_e3nn},
  month        = dec,
  year         = 2022,
  publisher    = {Zenodo},
  version      = {blog\_e3nn},
  doi          = {10.5281/zenodo.7430281},
  url          = {https://doi.org/10.5281/zenodo.7430281}
}