Theoretical Guarantees for the Architope Modification

A recently introduced architecture transformation, called the architope modification, was shown to bypass the composite pattern learning bottlenecks of feed-forward networks. This architecture modification acts by partitioning the input space and optimally redistributing the neurons of a single feed-forward network across several sub-networks, each specialized on a single part of the generated partition... (read more)

PDF Abstract

Results from the Paper

  Submit results from this paper to get state-of-the-art GitHub badges and help the community compare results to other papers.

Methods used in the Paper