Collected molecules will appear here. Add from search or explore.
Provides a framework for performing structured pruning on Group Equivariant Convolutional Neural Networks (G-CNNs) to ensure that the resulting compact models maintain their geometric transformation invariance (specifically C4 rotations).
citations
0
co_authors
1
The project addresses a niche but valid technical problem: standard pruning techniques often break the symmetry constraints of equivariant networks. While it provides a novel combination of group theory and model compression, the project has zero traction (0 stars) and functions primarily as a code accompaniment to a research paper.
TECH STACK
INTEGRATION
reference_implementation
READINESS