Permutation Invariance, DeepSets and Universal Function Approximation

This is a blog post by Fabian Fuchs, Martin Engelcke and me about the ability of the DeepSets architecture to represent arbitrary permutation invariant functions. The post is very closely related to our 2019 ICML paper On the Limitations of Representing Functions on Sets, but should be significantly easier to process.

We wrote the post in collaboration with Ferenc Huszár and hosted it on his website: https://www.inference.vc/deepsets-modeling-permutation-invariance/