The paper characterizes convolutional layer decompositions as hypergraphs, enumerates them, and finds by genetic search that nonlinear decompositions can beat existing light-weight layers on small benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Einconv: Exploring Unexplored Tensor Network Decompositions for Convolutional Neural Networks
The paper characterizes convolutional layer decompositions as hypergraphs, enumerates them, and finds by genetic search that nonlinear decompositions can beat existing light-weight layers on small benchmarks.